Measurable residual disease in elderly acute myeloid leukemia: results from the PETHEMA-FLUGAZA phase 3 clinical trial

Abstract

The value of measurable residual disease (MRD) in elderly patients with acute myeloid leukemia (AML) is inconsistent between those treated with intensive vs hypomethylating drugs, and unknown after semi-intensive therapy. We investigated the role of MRD in refining complete remission (CR) and treatment duration in the phase 3 FLUGAZA clinical trial, which randomized 283 elderly AML patients to induction and consolidation with fludarabine plus cytarabine (FLUGA) vs 5-azacitidine. After consolidation, patients continued treatment if MRD was ≥0.01% or stopped if MRD was <0.01%, as assessed by multidimensional flow cytometry (MFC). On multivariate analysis including genetic risk and treatment arm, MRD status in patients achieving CR (N = 72) was the only independent prognostic factor for relapse-free survival (RFS) (HR, 3.45; P = .002). Achieving undetectable MRD significantly improved RFS of patients with adverse genetics (HR, 0.32; P = .013). Longer overall survival was observed in patients with undetectable MRD after induction though not after consolidation. Although leukemic cells from most patients displayed phenotypic aberrancies vs their normal counterpart (N = 259 of 265), CD34 progenitors from cases with undetectable MRD by MFC carried extensive genetic abnormalities identified by whole-exome sequencing. Interestingly, the number of genetic alterations significantly increased from diagnosis to MRD stages in patients treated with FLUGA vs 5-azacitidine (2.2-fold vs 1.1-fold; P = .001). This study supports MRD assessment to refine CR after semi-intensive therapy or hypomethylating agents, but unveils that improved sensitivity is warranted to individualize treatment and prolong survival of elderly AML patients achieving undetectable MRD.

Graphical abstract

Figure 1.

Scheme of the PETHEMA phase 3 FLUGAZA clinical trial. A total of 283 patients were accrued and 78 achieved complete response with or without incomplete blood count recovery (CR/CRi).

Figure 2.

Outcome according to CR_MRDstatus in older AML patients treated with low-intensive therapy or hypomethylating agents. RFS (A) and OS (B) of patients in CR/CRi according to persistence (ie, ≥0.01%) vs undetectable (ie, <0.01%) MRD after induction. (C) Patients with undetectable MRD are listed according to their outcome: OS is depicted by gray bars, whereas progression and deaths are identified by orange and purple boxes, respectively. P values were determined by the 2-sided log-rank test.

Figure 3.

Phenotypic divergence and overlap between leukemic cells in elderly AML vs their normal maturation-stage counterpart in healthy adults. (A) Bone marrow samples from older AML patients (N = 265) and healthy adults (N = 30; age range, 20-24 years) were immunophenotyped with the first 5 combinations of the EuroFlow panel for the diagnostic classification of MDS/AML. Patient-specific aberrant phenotypes were identified and the total leukemic cells were exported into new FCS files without the remaining nucleated cells. Files from patients in whom ≥50% leukemic cells expressed CD34 were merged with files containing CD34⁺ progenitors from healthy adults; files from patients whom <50% leukemic cells expressed CD34 but ≥50% expressed CD117, were merged with files containing CD117⁺ myeloid or erythroid progenitors from healthy adults, depending on the lineage commitment of leukemic cells; files from patients whom <50% leukemic cells expressed CD34 and CD117 were merged with files containing the total neutrophil, monocytic, or erythroid lineage from healthy adults, depending on the lineage commitment of leukemic cells. After merging FCS files of leukemic cells from patients and the corresponding normal-cell counterpart from healthy adults, cells were plotted using principal component analysis (PCA) and represented according to their median value of expression for any given 8 markers in the combination (leukemic cells) or to the standard deviation (SD) (normal counterpart). In each combination, a score of −1, 0, or 1 was given if leukemic cells were plotted inside, over or outside the SD of the normal counterpart. Thus, patients could be scored from −5 (full phenotypic overlap with the normal counterpart) to 5 (full phenotypic divergence with respect to the normal counterpart). (B) Number of patients with each score based on the principal component analysis of merged data.

Figure 4.

Genomic landscape of CD34 progenitors and leukemic cells at MRD stages. Patients with undetectable (A) (N = 4) and persistent MRD (B) (N = 6) after FLUGA (light purple) or AZA (light green). Genes are distributed in rows, and columns represent paired baseline and phenotypically normal CD34 progenitors (HPCs) in patients with undetectable MRD, or paired diagnostic and blasts in patients with persistent MRD. Venn diagram shows the percentage of mutations and CNAs in blasts or HPCs detectable only at baseline, detectable only at MRD, and detectable in both time points. P value was determined by Pearson χ² for differences between treatment arms.