Abstract Acute myeloid leukemia (AML) progresses through the sequential accumulation of somatic mutations, with signaling transducer mutation typically preceding the acquisition of epigenetic alterations. However, recent single-cell DNA sequencing (scDNA-seq) studies have identified cases that do not follow this canonical mutational hierarchy. The prevalence of these non-classical mutation orders, and the biological mechanisms by which mutation timing drives phenotypic heterogeneity remain unclear. To address this, we analyzed publicly available scDNA-seq data from 150 newly diagnosed normal karyotype AML patients to reconstruct their evolutionary trajectories (Morita et al., 2020; Miles et al., 2020). We observed that 78% of the signaling mutation events—particularly FLT3-ITD, NRAS, and PTPN11—were terminal subclonal events in branch evolution. Surprisingly, 22% of the signaling mutations serve as initiating events, following a linear evolutionary trajectory and patients harboring these initiating signaling mutations exhibited significantly higher white blood cell and blast counts. These data suggest that the timing of signaling mutation acquisition is strongly linked to the resulting clonal evolution pattern, with FLT3-ITD emerging as the most prominent driver among all signaling events. To investigate the pathogenic mechanisms of initiating FLT3-ITD and the characteristics of the resultant AML, we therefore performed multimodal profiling (scDNA+protein-seq, scRNA-seq and patient-derived xenograft models) on 16 FLT3-ITD positive AML cases. Eight cases had FLT3-ITD as the initiating mutation (Initiating FLT3-ITD AML, IFA), and eight as subclonal events (Subclonal FLT3-ITD AML, SFA). Comparing tumor cell proportions identified by scRNA-seq with the mutant clone fractions from scDNA-seq in the same patient, we confirmed that leukemic transformation preceded acquisition of FLT3-ITD in SFA. Integrated genotype-immunophenotype analysis of paired samples—adjusted for co-mutations—revealed that initiating FLT3-ITD drove an expansion of hematopoietic stem and progenitor cells, whereas subclonal FLT3-ITD did not. Bayesian modeling further showed that initiating FLT3-ITD selectively upregulated leukemia stem cell (LSC) markers CD69 and CD123, resulting in a significantly larger CD123+CD69+ population (74% vs 35%, P=0.04). Consistently, in PDX models, IFA samples exhibited higher engraftment rates (5/8 vs 3/8). ScRNA-seq further revealed expansion of multipotent progenitor (MPP) and lymphoid-primed multipotent progenitor (LMPP) populations (29% vs 13%, P=0.04) and significant enrichment of LSC transcriptional signatures in IFA compared to SFA. Unsupervised clustering identified progenitor-enriched clusters in IFA characterized by downregulation of G2M checkpoint and E2F target pathways, indicative of a quiescent, stem-like state. Unexpectedly, canonical FLT3-ITD downstream pathways showed no significant differences between IFA, SFA, and FLT3-WT AML patients. Instead, in IFA, we observed significant upregulation of Hedgehog pathway genes (GLI2, ALDH1A1), known regulators of cell cycle and LSC quiescence. We observed frequent emergence of chr13q (FLT3 locus) uniparental disomy (UPD) in Initiating FLT3-ITD AML patients (15/23). Notably, we found the acquisition or expansion of UPD clones in IFA PDX models. Inferred from scRNA-seq, these UPD clones displayed increased proliferative capacity without losing LSC traits, potentially via RB1 inactivation. Drug sensitivity testing in PDX models revealed that IFA samples responded to chemotherapy and gilteritinib with distinct patterns, with gilteritinib selectively reducing leukemic blasts and inducing differentiation. However, single-cell genotyping of residual cells showed a marked rise in UPD clone post-gilteritinib. Longitudinal follow-up of these models confirmed that relapse was driven by the expansion of UPD clones. These data suggested chr13q is a required evolutionary event in the development and drug resistant progression of IFA. Our study demonstrates that initiating FLT3-ITD mutations reprograms leukemic cells to a quiescent, stem-like state, enforcing a linear evolutionary pattern, compared to subclonal FLT3-ITD. Clonal dominance is ensured by non-canonical FLT3 signaling and late-emerging chr13q UPD. These findings underscore the critical role of clonal timing in shaping AML biology and highlight new therapeutic vulnerabilities in initiating FLT3-ITD-driven AML.
He et al. (Mon,) studied this question.