Very short insertions in the FLT3 gene are of therapeutic significance in acute myeloid leukemia

TO THE EDITOR:

Internal tandem duplications (ITDs) of the Fms-like tyrosine kinase 3 (FLT3) gene occur in ∼25% to 30% of acute myeloid leukemia (AML) cases.^1,2 The presence of these FLT3-ITDs is associated with adverse outcomes, including higher relapse rates after conventional AML treatment and reduced overall survival.³ In contrast, point mutations occurring within the tyrosine kinase domain (TKD) of FLT3 are less common, accounting for 5% to 10% of AML cases, and have a neutral prognostic impact.⁴ Recent investigations have provided insight into the potential therapeutic benefits of combining AML chemotherapy with targeted tyrosine kinase inhibitors (TKIs) that specifically target mutated FLT3 receptors. In particular, TKIs such as midostaurin, gilteritinib, and sorafenib have shown promising results at various stages of AML therapy, including first-line treatment, relapse, and maintenance after allogeneic stem cell transplantation.^5-9 These benefits are particularly evident in patients harboring FLT3-ITD or TKD mutations. Therefore, the identification of pathogenic FLT3 mutations that can be targeted by TKIs has significant therapeutic implications for the treatment of AML.

ITDs occur predominantly in the juxtamembrane domain (JMD) of FLT3 and less frequently in the TKD, with the latter being associated with a poorer prognosis.¹⁰ Previous investigations have examined the size of ITDs and yielded conflicting results regarding their prognostic significance. Although the structure of the duplicated motif is heterogeneous, these studies consistently reported median ITD sizes ranging from 15 to 54 bp.^11-15 We identified 1974 patients with in-frame FLT3 insertions diagnosed at 11 tertiary care centers in France participating in the Groupe des Biologistes Moléculaires des Hémopathies Malignes (GBMHM) network. (Figure 1A; supplemental Table 1).¹⁶ Among these patients, 22 (1.1%) were found to have very short in-frame insertions (VSI) within FLT3, resulting in a gain of only 1 or 2 amino acids, a phenomenon that has already been described but not specifically studied (supplemental Table 2).^14,15 Interestingly, FLT3-VSIs were mainly observed at amino acid positions 572 to 578 and 591 to 599 in 25% and 60% of the analyzed cases in our cohort, respectively (Figure 1B). Surprisingly, the genomic landscape revealed an unexpected co-occurrence of 8 (40%) RUNX1 mutations, leading to a significant proportion of patients classified as ENL2017 adverse risk (Figure 1C; supplemental Table 3). FLT3-VSI alterations were detected together with FLT3-TKD and/or FLT3-ITD in 12 (54%) cases (supplemental Table 3). The characteristics of these patients are similar to those reported in classic FLT3-ITD cases, including leukocytosis, association with normal karyotype, and mutant or wild-type allele ratio values (supplemental Table 3). Overall survival was poor, with 30% of patients surviving after a median follow-up of 14.9 months, resulting in a median overall survival of 18.6 months (Figure 1D). Interestingly, a nonsignificant trend for decreased survival was observed in patients with simple FLT3 VSIs compared with composite cases with combined VSI and ITD abnormalities (supplemental Figure 1A).

To elucidate their biological relevance, we constructed expression vectors containing 4 FLT3-VSIs representative of the available sequences from 20 cases in our cohort (Figure 2A). These variants were compared with a standard FLT3-ITD derived from the human cell line MOLM-14.¹⁷ To evaluate the ability of these different forms to induce growth factor–independent proliferation, we used the following 2 cell lines: the human AML TF-1 cell line and the murine lymphoid Ba/F3 cell line. Notably, TF-1 cells are dependent on granulocyte-macrophage colony-stimulating factor for survival, whereas Ba/F3 cells are dependent on interleukin-3.¹⁸ This allowed us to assess the efficacy of the different FLT3 variants in promoting autonomous proliferation (supplemental Figure 1A).

Under cytokine deprivation conditions, our observations showed that all FLT3-VSI variants induced constitutive signaling through FLT3. This was evidenced by increased phosphorylation of STAT5 (Tyr-694) and FLT3 (Tyr-591) compared with empty vector transduced cells (Figure 2B).^19,20 However, we observed heterogeneity among the different FLT3-VSI, particularly in Ba/F3 cells. Furthermore, all VSI conferred cytokine-independent growth in vitro, comparable to cells transduced with the FLT3-ITD vector (Figure 2C). These results underscore that FLT3-VSIs activate FLT3 signaling and maintain autonomous cell growth at levels similar to that of classical FLT3-ITDs. We then evaluated the efficacy of 4 different small-molecule FLT3-targeting TKIs: quizartinib, gilteritinib, midostaurin, and ponatinib. Notably, all of these TKIs, when used at nanomolar concentrations, exhibited almost complete inhibition of FLT3 signaling in both TF-1 and Ba/F3 cells, as evidenced by modulation of FLT3, STAT5, and ERK phosphorylation levels (Figure 2D; supplemental Figure 1B). In addition, the 4 TKIs induced a dose-dependent reduction in cell viability after 48 hours of incubation. However, the dose-response profile showed a more robust effect in Ba/F3 cells compared with TF-1 cells, indicating a lower degree of oncogenic addiction to FLT3 signaling in the latter (Figure 2E; supplemental Figure 1C). Notably, the cytotoxic effects of TKIs on cell-lines expressing FLT3-VSI were found to be comparable, and in some cases, superior to those observed in cells transduced with a conventional ITD (Figure 2F). These results highlight the potential therapeutic efficacy of TKIs against FLT3-mutant AML harboring very short insertions in the FLT3 gene.

Structurally, the FLT3 receptor tyrosine kinase consists of several domains. Within the extracellular domain, immunoglobulin-like domains are responsible for binding the FLT3 ligand (FL). Upon ligand-induced dimerization, autophosphorylation occurs at specific tyrosine residues in the kinase domain, initiating downstream signaling cascades that regulate cell proliferation, survival, and differentiation.²¹ Physiologically, JMD interactions maintain FLT3 in an autoinhibited state by spanning the N- and C-terminal lobes of the catalytic domain. FL-induced activation disrupts the inhibitory contacts through autophosphorylation of conserved tyrosine residues in the JMD, whereas ITD impairs this autoinhibition mechanism. Notably, our results demonstrate that even a small insertion of 1 or 2 amino acids within essential JMD residues, mainly at position 592 to 599, significantly induces constitutive oncogenic activation of FLT3 signaling. Interestingly, this level of activation was comparable to that induced by longer insertions, underscoring the remarkable sensitivity of the JMD to even small changes in length.

The identification of FLT3-VSIs expands our understanding of the mutational landscape of FLT3-driven cancers. Although these mutations represent a small proportion (1.1%) of all FLT3 insertions, they may have clinical significance in certain patient populations. Notably, FLT3-VSI alterations were frequently associated with FLT3-TKD (27%) or FLT3-ITD (18%) in this cohort. Although the presence of composite FLT3 mutations is well-documented in patients who relapse after TKI treatment, data regarding the co-occurrence of distinct FLT3 mutations are limited and suggest that FLT3-TKD mutations can be found in >10% of diagnostic samples harboring FLT3-ITD.^22,23 Similarly, the correlation between FLT3-VSI and RUNX1 mutations is higher than expected for FLT3-ITD, which may contribute to the unfavorable outcome observed in our study and warrants further investigation in larger cohorts of patients with AML.²⁴ 

Through experimental analysis using cytokine-dependent hematopoietic cell lines, we found that these very short insertions activated FLT3 signaling, induced cytokine-independent growth, and demonstrated significant sensitivity to FLT3-targeting TKIs. These findings are significant because the validated commercial kit LeukoStrat CDx FLT3 Mutation Assay, currently used as a standard tool to select patients with AML for treatment with midostaurin, is only validated to detect FLT3-ITD between 30 and 279 bp.²⁵ Maintaining this detection threshold could therefore lead to misdiagnosis in a subset of patients with FLT3-VSI AML, a scenario in which TKI treatment has the potential to significantly improve clinical outcomes.

In conclusion, these findings have clinical implications for the diagnosis, prognosis, and therapeutic decision-making of patients with FLT3-driven malignancies.

Acknowledgments: The authors thank the Geneva Reader Assay Development and Screening core facilities and in particular Yves Cambet for technical support and insightful collaboration.

This work was supported by grants from the Copley May, Dr. Henri Dubois-Ferrière, Dinu Lipatti, Medic and Lombard Odier (Fonds Jean Pastré) Foundations, as well as Ligue Genevoise Contre le Cancer.

Contribution: J.T., S.M., C.L., P.S., and O.K. developed the methodology; J.T., S.M., P.S., and O.K. performed formal analysis; N.D., A.B., A.S., P.H., L.R., S.C., M.T., F.F., P.F-G., Y.L.B., A-S.A., L.M., S.T., E.D., and F.D. carried out the investigation; J.T. was responsible for funding acquisition, prepared the original draft of the manuscript, and performed visualization; and J.T., P.S., and O.K. conceptualized and validated the study, were responsible for project administration, supervised the study, and reviewed and edited the manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

A complete list of the members of the French GBMHM group appears in “Appendix.”

Correspondence: Jerome Tamburini, Faculty of Medicine, University of Geneva, rue Michel Servet 1, Geneva 1206, Switzerland; email: jerome.tamburinibonnefoy@unige.ch.