Comparative analysis of melphalan versus busulphan T‐cell deplete conditioning using alemtuzumab in unrelated donor stem cell transplantation for acute myeloid leukaemia

Abstract

For most patients with intermediate or high-risk acute myeloid leukaemia (AML), allogeneic stem cell transplantation (allo-HSCT) offers the best chance of cure (D€ ohner et al, 2017). Reduced intensity conditioning (RIC) protocols, with greater emphasis on a graft-versus-leukaemia effect, allow allo-HSCT in older patients and those with co-morbidities (Sengsayadeth et al, 2015). The morbidity and mortality associated with acute and chronic graft-versus-host disease (aGVHD, cGVHD, respectively) has been reduced by in vivo T-cell depletion (TCD) (Soiffer et al, 2011; Walker et al, 2016) but there is limited evidence for optimal allo-HSCT protocols in AML (Bacigalupo et al, 2009; Jethava et al, 2017). We conducted a multi-centre retrospective analysis of two reduced toxicity alemtuzumab-based T-cell deplete protocols in common use in the United Kingdom for unrelated donor allo-HSCT in patients with AML: fludarabine-melphalan (FMC) and fludarabine-busulfan (FBC). We identified 117 patients from University College London Hospital (UCLH) (employing FMC), Royal Free London Hospital (RFH)(employing FMC) and Kings College Hospital, London (KCH) (employing FBC) receiving matched unrelated TCD-HSCT for AML in complete remission (CR) [defined as per (Cheson et al, 2003)] between January 2005 and December 2014. Patient demographic data are shown in Table I. Chimerism, GvHD and graft failure were assessed using published and/or standard methods with further details of transplant conditioning regimens, immunosuppression and statistical methods provided in Data S1. The primary endpoint was overall survival (OS), measured from day 0 to death from any cause. Secondary endpoints included cumulative incidence of relapse (CIR); relapse-free survival (RFS), measured from day 0 to first relapse or death; non-relapse mortality (NRM) measured from day 0 to death without relapse. The overall median age of the whole cohort (n = 117) was 55 years (range 19–68). The 5-year OS for the whole cohort was 48% (Fig 1A) with no significant difference between FMC (n = 70; 5-year OS 46%) and FBC n = 47; 5-year OS 52%, P = 0 2) (Fig 1B). Patients receiving FBC were younger (P = 0 04), more likely to have a fully matched (10/10) donor (P = 0 03) and received a lower dose of alemtuzumab (P < 0 001). No other significant differences between the two groups were observed. Median follow-up for patients alive at the end of the study was 37 months (range 11 2–97 5) for FMC (n = 27) and 42 months (range 4 6–106) with FBC (n = 33) regimen (P = 0 40). Patients aged ≥60 years had a poorer 5-year OS of 38% vs. 55% (<60 years) (P = 0 03; Fig 1C). Patients with a mismatched unrelated donor (MMUD) had inferior 5-year OS of 29% compared with 60% with 10/10 human leucocyte antigen (HLA)-matched unrelated donor (MUD) (P = 0 009; Fig 1D), with both these variables retained significance on adjusted Cox regression (Table SI). Patients <60 years age who underwent 10/10 HLA-matched MUD HSCT had an excellent 5-year OS of 73%, with no difference noted between the two regimens (P = 0 14). There was no significant difference in OS/RFS for other variables measured including cytogenetic risk group and FLT3-internal tandem duplication (ITD) (Fig 1E,F, Table SII). Overall NRM at 1 and 5 years was 20% and 25% respectively and not different between the two regimens (Fig 1G). Five-year NRM was higher for older patients (42% vs. 15%; P = 0 001; Fig 1H), in those with extensive versus limited cGVHD (43% vs. 5%, P = 0 0006; Fig 1I) and those using MMUD vs. 10/10 MUD (36% vs. 18% respectively; P = 0 03; Table SII). The 5-year RFS was 45% across the whole cohort with similar results between the 2 regimens (P = 0 68; Fig 1J). CIR was 16% and 30% at 1 and 5 years respectively, with no significant difference observed between the two regimens (Fig 1K). Five-year CIR was increased in male recipients (38% male vs. 18% female; P = 0 02) and patients with normal karyotype FLT3-ITD mutated (FLT3) versus wild type (FLT3) (52% vs. 25% respectively; P = 0 02). CIR was reduced in patients with cGVHD (all grades) (P = 0 02), most notably with extensive cGHVD (P = 0 006; Fig 1L, Table SI) but this did not translate into differences in OS (P = 0 39). The FMC cohort was more likely to develop grade 1 acute and limited cGVHD events compared to the FBC cohort (P < 0 001); while the incidence of grade 3-4 aGVHD and extensive cGVHD was similar in both cohorts (Table SIII). Interestingly, CIR was significantly lower in patients with a more strictly defined full donor T-cell chimerism (>97% CD3 FDC) (P = 0 03) but no such differences were observed using standard FDC (>95%) definition versus other chimerism groups (Table SIV). At the completion of the study, 33 patients had relapsed (FMC = 17, FBC = 16). Median survival following disease relapse was longer in the FBC versus FMC group (240 days