The Hierarchy of Alternative Donors for Allogeneic Hematopoietic Stem Cell Transplantation in Poor Risk AML in CR1: 10/10 Matched Unrelated Donors Still to be Preferred over Haplo-Identical Donors or Umbilical Cord Blood

Allogeneic hematopoietic stem cell transplantation (alloHSCT) is the most effective post-remission treatment (PRT) for prevention of relapse in poor risk acute myeloid leukemia (AML) in first complete remission (CR1). Most patients, however, lack a human leukocyte antigen (HLA) matched related donor (MRD). For these patients, 10/10 or 9/10 HLA-matched adult unrelated donors (MUDs) are considered preferred alternative donors, but results with haplo-identical donors (haplo) and alloHSCT using umbilical cord blood (UCBT) have gradually improved the last decade and these alternative donors/stem cells sources, with the advantage of rapid availability, are increasingly being applied. We set out to compare outcome in poor risk AML patients in CR1 proceeding to PRT by alloHSCT using either MRD, 10/10 or 9/10 MUDs, haplo or UCBT in order to establish the hierarchy of alternative donors/stem cell sources.

2850 adult patients, transplanted between 2000 and 2012 and reported to the EBMT Acute Leukemia Working Party, with poor risk AML in CR1 receiving an alloHSCT using MRD (n=1716) or alternative donors, including 10/10 (n=635) or 9/10 MUDs (n=260), UCBT (n=142), or haplo (n=97) were eligible for this analysis. Poor risk AML was defined as either no CR after one cycle of induction chemotherapy, white blood cell count >100 x 10⁹/L at diagnosis, or cytogenetic/molecular abnormalities associated with adverse risk, according to European LeukemiaNET classification. Endpoints of the study included 3-year overall survival (OS), relapse-free survival (RFS), and cumulative incidences of relapse and non-relapse mortality (NRM). Patient characteristics as well as time intervals from diagnosis to CR1 and from CR1 to transplant were evaluated and included in a multivariable analysis.

Patient characteristics revealed no differences between the treatment groups with respect to the risk status of underlying AML. Recipients of haplo alloHSCT and UCBT were younger than MRD alloHSCT recipients (median ages, respectively, 43 and 45 versus 48 years, p<0.0001). The time interval from diagnosis to CR1 was longer for recipients of haplo alloHSCT as compared to MRD, MUD, or UCBT recipients (respectively: 65 versus 54, 51, and 55 days, p=0.013). The time interval between CR1 and transplant was longer for recipients of 10/10 and 9/10 MUD alloHSCT (respectively 113 and 137 days) as compared to UCBT (111 days) and haplo alloHSCT (106 days) versus MRD alloHSCT (79 days) (p<0.0001). 51% of patients needed more than 1 cycle of chemotherapy to obtain CR1 in all subgroups according to type of donor. Alternative donors were more frequently used in the recent years as compared to MRDs. Both reduced intensity conditioning and myeloablative conditioning (MAC) were applied, with a higher percentage of MAC in recipients of haplo alloHSCT (67%) and MRD alloHSCT (62%) versus patients receiving MUD alloHSCT (50%) and UCBT (44%). Prevention of GVHD differed, with more patients receiving in-vivo T-cell depletion before MUD and haplo alloHSCT (respectively 76% and 77%) versus 23% in MRD alloHSCT and 33% in UCBT.

3 year OS following MRD alloHSCT was 56%, which did not differ from 10/10 MUD alloHSCT, but which differed significantly from 9/10 MUD (45%), UCBT (44%), and haplo alloHSCT (49%) (p=0.0007). Multivariable analysis confirmed the impact of type of donor and also showed that higher age and longer time from diagnosis to CR1 adversely impacted on OS. A longer time from CR1 to transplant was associated with better OS and less relapse. 3 year RFS was 51% and 50% following MRD and 10/10 MUD alloHSCT, respectively, which was significantly (p=0.0009) better than following UCBT (35%), haplo alloHSCT (39%), or 9/10 MUD alloHSCT (42%). NRM depended on donor type and estimated 36% and 25% after haplo alloHSCT and UCBT, respectively, versus 15% following MRD alloHSCT at 3 years. Relapse did not significantly differ by donor type in multivariable analysis.

Collectively, these results suggest that alloHSCT with MRDs and 10/10 MUDs may still be preferred in patient with poor risk AML in CR1. Due to higher NRM, recipients of 9/10 MUD, UCBT, or haplo each showed a significant reduction of approximately 10% in OS and RFS as compared to MRD and 10/10 MUD alloHSCT. Therefore, the urge to immediately proceed to alloHSCT in poor risk AML might favor UCBT or haplo if the identification of 9/10 MUD is expected to be prolonged for months and a MRD or 10/10 MUD is not available.