Backround. Low-dose total body irradiation (TBI)-based nonmyeloablative conditioning regimens are widely used in acute myeloid leukemia (AML) patients in first or second complete remission (CR) at the time of transplantation. With these regimens the burden for tumor eradication relies mostly on immune-mediated graft-versus-leukemia (GVL) effects that can be impacted by donor type. Here we assessed the impact of donor type in a large cohort of AML patients transplanted in CR after low-dose TBI-based nonmyeloablative conditioning regimens.

Methods. Inclusion criteria included >= 18 years of age, de novo or secondary AML in CR1 or CR2, transplantation between 2004 and 2015, conditioning with fludarabine + 2 Gy TBI with or without pre- or post-transplant cyclophosphamide, no in vitro or in vivo T cell depletion of the graft (besides post-transplant cyclophosphamide), and either an HLA-matched sibling donor (MSD), a 10/10 HLA-matched unrelated donor (MUD), an HLA-haplo-identical donor (Haplo) or a single or double umbilical cord blood (CBT).

Results. Data from 1357 patients receiving grafts from MSD (n=594), MUD (n=442), Haplo (n=77) or given single or double CBT (n=244) were included. The proportion of patients transplanted in CR1 was 87%, 80%, 66% and 60% in MSD, MUD, Haplo and CBT groups, respectively (P<0.001) while median patient age at transplantation was 58, 63, 57 and 55 years, respectively (P<0.001). The proportion of patients with secondary AML was 16%, 22%, 21% and 20% in MSD, MUD, Haplo and CBT groups (P=0.04). Median follow-up was 49, 25, 17 and 48 months, respectively (P<0.001). Graft rejection occurred in 2% of each MSD and MUD recipients, 7% of haplo patients and 8% of CBT recipients (P<0.001). Grade III-IV acute GVHD was more frequent in CBT (16%) than in MUD (9%), haplo (8%) or MSD (7%) recipients (P<0.001). In contrast, chronic GVHD was less frequent in CBT (27%) and in haplo (30%) patients than in MSD (50%) and MUD (52%) recipients (P<0.001). Two-year OS and GRFS were 60% and 30% in MSD patients, 56% and 41% in CBT recipients, 54% and 24% in MUD recipients, and 42% and 35% in haplo patients, respectively (P=0.02 for OS and P=0.03 for GRFS ; Figure 1). Two-year incidences of relapse and nonrelapse mortality (NRM) were 33% and 12% in MSD patients, 29% and 20% in MUD recipients, 34% and 17% in CBT recipients, and 37% and 21% in haplo patients, respectively (P=0.7 for relapse and P<0.001 for NRM). In multivariate analyses, in comparison to MSD patients, each other donor type was associated with a significantly higher NRM. Interestingly, CBT was associated with a higher incidence of grade III-IV acute GVHD (HR 2.9, P=0.001) but a lower incidence of chronic GVHD (HR=0.4, P<0.001) than MSD resulting in a significantly worse GVHD-free relapse free survival (GRFS) the first 100 days (HR=1.5, P=0.02) but significantly better GRFS beyond day 100 (HR=0.6, P=0.001). Finally, there was a suggestion for lower OS in haplo patients (HR 1.5, P=0.08). Factors associated with worse OS included intermediate and adverse-risk cytogenetics and secondary AML. AML relapse, GVHD and infections were the primary causes of death of 24%, 6% and 4% of MSD recipients, 21%, 7% and 7% of MUD recipients, 22%, 5% and 14% of haplo recipients, and 25%, 8% and 5% of CBT recipients, respectively.

Conclusions. In this large cohort of AML patients transplanted in CR following low-dose TBI-based nonmyeloablative conditioning regimen there were no significant differences in OS and LFS between MSD, MUD and UCB while there was a suggestion for lower LFS and OS with haplo due to a higher incidence of death from infection. Since this regimen relies nearly exclusively on GVL effects for tumor eradication, our data suggest that the intensity of GVL effects is comparable with these four donor types. Further studies including more haplo patients are needed to confirm these results.


Savani: Jazz Pharmaceuticals: Speakers Bureau. Mohty: Sanofi: Honoraria, Speakers Bureau.

Author notes


Asterisk with author names denotes non-ASH members.