FLT3 inhibitor maintenance after allogeneic hematopoietic stem cell transplantation (allo-SCT) is supported by randomized data, mainly for sorafenib, from trial settings with uniform selection, timing, and duration. Routine practice varies on all three.
A German multicenter registry analysis of FLT3-ITD-mutated acute myeloid leukemia (AML) transplanted in first complete remission found better long-term survival with maintenance, most clearly with sorafenib. Relapse incidence did not differ significantly, and differences in patient selection and transplant era may account for part of the survival gap.
Maintenance After Transplant
FLT3 mutations occur in about 25%-30% of newly diagnosed AML. Internal tandem duplications (FLT3-ITD) raise relapse risk, so recurrence remains a major concern even after successful transplantation. Maintenance aims to suppress residual leukemic clones.
Evidence is strongest for sorafenib. The randomized SORMAIN trial showed improved relapse-free survival over placebo, and a separate randomized Chinese study reported less relapse and better survival. Evidence for other inhibitors is less consistent. The RADIUS trial of midostaurin showed no statistically significant benefit, and the phase III MORPHO trial of gilteritinib missed its primary endpoint in the overall population, though patients with detectable MRD appeared more likely to benefit.
Study Design
This retrospective study of the German Registry for Hematopoietic Stem Cell Transplantation and Cell Therapy included adults with FLT3-ITD-mutated AML who underwent allo-SCT in first complete remission between 2011 and 2023 at 13 centers. Of 927 patients identified, 404 were excluded because post-transplant maintenance status was unknown, leaving 523.
- 116 patients (22%) received FLT3 inhibitor maintenance
- 407 patients (78%) did not
Among maintenance recipients, 49% received sorafenib, 37% midostaurin, and 5% gilteritinib, and the agent was unknown in 9%. Physicians decided on maintenance based on clinical recovery, GVHD, donor characteristics, drug availability, and reimbursement, and no protocol was shared across centers. Sorafenib started at a median of 74 days after transplant and continued for a median of 486 days. Midostaurin and gilteritinib were generally given for shorter periods.
At transplant, 39% of patients were MRD-negative, 32% MRD-positive, and 29% had unknown status. MRD was assessed locally by flow cytometry or molecular methods, including NPM1 mutation and leukemia-associated fusion transcripts.
To limit immortal-time bias, all analyses ran from a day +100 landmark and excluded patients who died or relapsed earlier. Results therefore apply to patients alive and relapse-free at day +100. Endpoints were overall survival, RFS, cumulative incidence of relapse, non-relapse mortality, and GVHD-free, relapse-free survival (GRFS).
Results: Better OS and RFS With Maintenance
Five-year outcomes were:
Overall survival: 77% with maintenance versus 62% without.
Relapse-free survival: 75% versus 55%.
GVHD-free, relapse-free survival: 51% versus 39%.
Both OS and RFS associations remained significant after multivariable adjustment: the hazard ratio for patients without maintenance was 2.25 for OS and 1.72 for RFS. Maintenance recipients did not have significantly more acute or chronic GVHD. The retrospective design can’t establish that maintenance caused these differences.
Survival Gain Without a Significant Relapse Reduction
Five-year relapse incidence was 22% with maintenance and 28% without, which was not significant. Five-year NRM was 3% versus 17%. Maintenance recipients were more often transplanted in recent years, had better performance status, and had fully matched donors. Clinicians also started maintenance in patients who had recovered well, kept organ function, and had no prohibitive GVHD, and each factor could lower NRM independently.
The authors discuss a possible immunomodulatory effect of sorafenib through interleukin-15-mediated enhancement of donor T-cell and natural killer cell activity. The registry didn’t capture infections or causes of death. The NRM difference is more plausibly explained, at least in part, by selection and better transplant care over time. Randomized sorafenib trials showed benefit mainly through relapse prevention, which makes the strong NRM signal here a point of difference.
MRD Status
In MRD-positive patients, maintenance was associated with five-year OS of 76% versus 59% and RFS of 72% versus 55%. The OS association was significant, while differences in relapse and NRM were not. This fits the rationale for maintenance when residual leukemia persists.
In MRD-negative patients, five-year OS was 86% with maintenance versus 62% without, and RFS was 78% versus 53%. These patients were not necessarily free of leukemia: assay sensitivity varied between centers, and no uniform, highly sensitive FLT3-ITD MRD platform was used. The comparisons are observational subgroups, so they support further study of maintenance in MRD-negative disease but can’t show whether all such patients need it.
Sorafenib
In an exploratory analysis of about 50 sorafenib recipients, five-year outcomes were:
OS: 85% with sorafenib versus 62% without maintenance
RFS: 84% versus 55%
Relapse incidence: 13% versus 28%
NRM: 3% versus 17%
Both survival differences were significant. Relapse incidence and NRM were numerically lower but not significantly different in this smaller group.
In MRD-negative patients, sorafenib was associated with five-year OS of 90% versus 62% and RFS of 86% versus 53%. These differences were significant, though the subgroup had only 24 sorafenib-treated patients and the HR confidence intervals were wide. In MRD-positive patients, outcomes were numerically favorable without significant differences. The lack of significance in this small subgroup isn’t evidence that sorafenib is ineffective with residual leukemia, given the prospective data supporting maintenance in that setting.
The results agree with randomized evidence and extend it to routine practice, but the cohort was small and nonrandomized. They can’t establish the size of sorafenib’s causal benefit or its superiority over other inhibitors. No significant difference emerged between sorafenib and midostaurin for OS, RFS, relapse, NRM, or GRFS. Sorafenib was given for longer, and prescribing patterns changed over the study period, so tolerability, access, and selection may have influenced results.
Comparison With Randomized Trials
SORMAIN reduced the risk of relapse or death versus placebo, with two-year RFS of 85% versus 53%, and the Chinese randomized study showed durable benefit with extended follow-up. These trials give stronger evidence of antileukemic activity because allocation was randomized. The registry adds long-term outcomes in patients treated outside trial protocols.
The apparent benefit in MRD-negative patients contrasts with MORPHO, where gilteritinib’s benefit appeared concentrated in patients with detectable MRD. Differences in drug mechanism, MRD assays, populations, and design prevent direct comparison, and the registry doesn’t settle whether maintenance should be selected by MRD status, prior FLT3 inhibitor exposure, or other biology. Prior exposure matters as quizartinib and other targeted agents move into pre-transplant treatment.
Limitations
Maintenance was prescribed by physician judgment, so treated patients likely differed systematically from untreated ones. Adjustment for transplant year can’t remove residual differences in supportive care, donor selection, and drug availability across a decade of changing practice. The landmark design addressed immortal-time bias but dropped early relapses and deaths.
Excluding 404 patients with unknown maintenance status limits generalizability. MRD testing was heterogeneous, FLT3-ITD allelic ratios and post-relapse treatment data were incomplete, and few patients received each individual inhibitor, which limits drug-specific and MRD-stratified comparisons.
Reference: Massoud R, Flossdorf S, Hanke F, et al. Impact of FMS-like tyrosine kinase 3 inhibitor maintenance on post-transplant outcomes in acute myeloid leukemia with FMS-like tyrosine kinase 3 mutations: a real-world German registry analysis highlighting sorafenib. Haematologica. 2026;111(10). DOI: 10.3324/haematol.2025.300176
Machine Learning Improves AML Relapse Risk Prediction Beyond ELN 2022: HARMONY Study