Impact of Conditioning Intensity of Allogeneic Transplantation for Acute Myeloid Leukemia With Genomic Evidence of Residual Disease

Abstract

Purpose: Patients with acute myeloid leukemia (AML) in remission remain at risk for relapse even after allogeneic hematopoietic cell transplantation (alloHCT). AML measurable residual disease (MRD) status before alloHCT has been shown to be prognostic. Whether modulation of the intensity of the alloHCT conditioning regimen in patients with AML who test positive for MRD can prevent relapse and improve survival is unknown.

Methods: Ultra-deep, error-corrected sequencing for 13 commonly mutated genes in AML was performed on preconditioning blood from patients treated in a phase III clinical trial that randomly assigned adult patients with myeloid malignancy in morphologic complete remission to myeloablative conditioning (MAC) or reduced-intensity conditioning (RIC).

Results: No mutations were detected in 32% of MAC and 37% of RIC recipients; these groups had similar survival (3-year overall survival [OS], 56% v 63%; P = .96). In patients with a detectable mutation (next-generation sequencing [NGS] positive), relapse (3-year cumulative incidence, 19% v 67%; P < .001) and survival (3-year OS, 61% v 43%; P = .02) was significantly different between the MAC and RIC arms, respectively. In multivariable analysis for NGS-positive patients, adjusting for disease risk and donor group, RIC was significantly associated with increased relapse (hazard ratio [HR], 6.38; 95% CI, 3.37 to 12.10; P < .001), decreased relapse-free survival (HR, 2.94; 95% CI, 1.84 to 4.69; P < .001), and decreased OS (HR, 1.97; 95% CI, 1.17 to 3.30; P = .01) compared with MAC. Models of AML MRD also showed benefit for MAC over RIC for those who tested positive.

Conclusion: This study provides evidence that MAC rather than RIC in patients with AML with genomic evidence of MRD before alloHCT can result in improved survival.

FIG 1.

Detection of mutations in the blood of patients with acute myeloid leukemia (AML) during complete remission (CR). (A) The allele frequency of each mutation detected in each gene in the blood of patients with AML during morphologic CR before transplantation is shown. Allele frequencies for FLT3 internal tandem duplication (ITD) and NPM1 variants detected at < 0.1% are plotted at 0.1%. The majority of variants (88%) had an allele frequency < 10%, with exceptions primarily occurring in genes known to be associated with clonal hematopoiesis. (B) The total number of mutations detected per patient and the distribution across patients are shown, including patients with at least 1 mutation detectable in DNMT3A, TET2, or ASXL1 (DTA) genes and those with no detectable DTA mutation. (C) A network analysis of the co-occurrence of mutations on the gene level within patient samples is shown. Each gene is depicted as a node, and mutational co-occurrence is depicted as an edge, where node color is representative of the total number of genes found to be concurrently mutated and the distance between nodes is proportional to the rate of co-occurrence across all patient samples.

FIG 2.

Impact of conditional intensity and mutational status on clinical outcomes. (A) Differences in rates of transplant-related mortality (TRM) were identified between subgroups defined by conditioning intensity and mutational status (P = .02). TRM was significantly higher in patients who underwent treatment with myeloablative conditioning (MAC) v reduced-intensity conditioning (RIC; P = .001), but there was no difference on the basis of mutational status (P = .8). Rates of relapse were different between subgroups (P < .001), with RIC having a higher relapse rate than MAC (P < .001) and the highest rate occurring in next-generation sequencing (NGS) positive patients who received RIC (P < .001). (B) In patients with no mutations detected (NGS negative), overall survival (OS) did not differ on the basis of conditioning intensity (3-year OS, 63% RIC v 56% MAC; P = .96). However, in those with detectable mutations, survival was significantly worse in those who received RIC (3-year OS, 43% RIC v 61% MAC; P = .02).

FIG 3.

Relapse outcomes of patients with specific gene mutations detected. The percentage of patients with mutations in each gene separated by conditioning intensity and excluding those with transplant-related mortality (TRM) is shown. For multiple genes (IDH1, IDH2, SF3B1, non–internal tandem duplication [ITD] FLT3, NPM1), differences in whether a patient experienced relapse v no relapse were observed on the basis of treatment with reduced-intensity conditioning (RIC) v myeloablative conditioning (MAC).

FIG 4.

Prognostic implications of detected gene mutations. (A) Patients with a detectable FLT3 internal tandem duplication (ITD) mutation had a significantly decreased rate of overall survival (OS) compared with patients with no detectable mutation (next-generation sequencing [NGS] negative; P < .001), while (B) patients with mutations detected only in ≥ 1 of DNMT3A, TET2, or ASXL1 (DTA) showed no difference in OS (P = .23). (C) A model was generated by considering those patients with mutations only in DTA as negative. Using this definition, OS for patients with detectable non-DTA mutations treated with reduced-intensity conditioning (RIC) was significantly worse than for other subgroups (P = .01). MAC, myeloablative conditioning.