Comparative Effectiveness of Busulfan/Cyclophosphamide Versus Busulfan/Fludarabine Myeloablative Conditioning for Allogeneic Hematopoietic Cell Transplantation in Acute Myeloid Leukemia

Background

Myeloablative conditioning (MAC) is a standard approach for allogeneic hematopoietic cell transplantation (alloHCT), but is associated with risks of morbidity and mortality. As regimen intensity affects post-transplant outcomes, assessment of pre-transplant cytogenetics and somatic mutations may refine which acute myeloid leukemia (AML) patients benefit the most. We compared the effectiveness of two approaches: busulfan/cyclophosphamide (Bu/Cy) and the MAC, but reduced toxicity regimen busulfan/fludarabine (Bu/Flu). Moreover, it is unclear whether somatic mutations in AML may differentially affect post-transplant outcomes between these regimens. We hypothesized that despite relative differences with these regimens, they may result in comparable outcomes.

Methods

We conducted a single center, retrospective analysis of adult AML patients in CR1 or CR2 who underwent a first T-cell replete HLA-8/8 matched related or unrelated donor alloHCT. Patients received either parenteral Bu (12.8 mg/kg total over 4 days) with Cy (120 mg/kg total over 2 days) or parenteral Bu (400 mg/m² total over 4 days) with Flu (160 mg/m² total over 4 days). Bu dose adjustment was not used in either cohort. In addition to cytogenetic risk group stratification by European LeukemiaNet criteria (Döhner H, et al, Blood 2016), a subset of patients had a 36-gene somatic mutation panel assessed prior to alloHCT by next-generation sequencing. Pre-transplant characteristics were compared between regimens with Chi-square, Fisher's exact, or Wilcoxon rank sum tests. Differing characteristics were included in a multivariable Fine and Gray regression model with results as hazard ratios (HR) and 95% confidence intervals (CI).

Results

From 2008 - 2017, 76 AML patients receiving Bu/Cy and 50 receiving Bu/Flu were identified meeting inclusion criteria (Bu/Flu used starting in 2010). Median age at transplant was 51 (21-61) years for Bu/Cy vs. 64 (34-74) for Bu/Flu (P<0.001). The cohorts were otherwise comparable in regards to gender, race, performance status, HCT-CI, and disease status at alloHCT. Bu/Cy vs. Bu/Flu patients had 16% vs. 10% favorable, 66% vs. 50% intermediate, and 18% vs. 40% adverse-risk cytogenetics (P=0.033). The most common somatic mutations in the Bu/Cy cohort were FLT3 (20%), NPM1 (18%), DNMT3A (16%), TET2 (9%), CEBPA (5%), and IDH1 (5%). In the Bu/Flu cohort, these were FLT3 (20%), NPM1 (18%), NRAS (12%), TET2 (12%), and DNMT3A (10%). There were no significant differences in somatic mutations between the cohorts, except for a higher incidence of NRAS in the Bu/Flu cohort (12% vs. 4%, P=0.029). Bu/Flu patients were more likely to have an unrelated donor (70% vs. 47%, P=0.012) and receive a peripheral blood stem cell (PBSC) graft (94% vs. 17%, P<0.001). As such, Bu/Flu patients had more rapid neutrophil (median 13 vs. 14 days, P=0.009) and platelet (median 15 vs. 20 days, P<0.001) recovery and a shorter length of hospital stay (LOS) (median 22 vs. 27 days, P<0.001). In multivariable analysis, Bu/Flu patients trended towards more chronic graft-versus-host-disease (GvHD; any stage) (HR 0.42, CI 0.16-1.11, P=0.08), but there were no other differences in CMV infections, other infections, acute GvHD, relapse, relapse mortality (RM), non-relapse mortality (NRM), relapse-free (RFS), and overall survival (OS). 58% of Bu/Cy and 56% of Bu/Flu patients remain alive with median follow-up of 59 and 22 months, respectively (P=0.003). The most common causes of death for these respective cohorts were relapse (50% vs. 41%) and infection (16% vs. 27%).

Conclusion

Bu/Cy and Bu/Flu in alloHCT for AML results in comparable incidences of infection, GvHD, RM, NRM, RFS, and OS. This was despite Bu/Flu patients being older and more likely to have adverse cytogenetics and an unrelated donor. Bu/Flu is better tolerated with less toxicity. Faster hematopoietic recovery and shorter LOS with Bu/Flu reflects PBSC graft use and has implications for health care resource utilization. Future prospective studies with larger cohorts and cost-effectiveness analyses comparing these conditioning strategies are warranted. In this analysis, no mutations appeared to be sensitive to use of the more intensive regimen, Bu/Cy. Further investigation of pre-transplant or post-transplant persistence of somatic mutations may risk stratify those who may benefit from more intensive or innovative approaches to prevent relapse after transplant.