Clinical and molecular response of acute myeloid leukemia harboring non-canonical FLT3 N676K driver mutations to contemporary FLT3 inhibitors

Treatment of acute myeloid leukemia (AML) has been enhanced by the development and regulatory approval of several novel agents, including midostaurin and gilteritinib (FLT3 inhibitors), venetoclax (BCL2 inhibitor), ivosidenib (IDH1 inhibitor), and enasidenib (IDH2 inhibitor).¹ A challenge during the era of molecular therapies, however, is determining the efficacy of these agents for patients with AML harboring atypical driver mutations. These atypical drivers were underrepresented in seminal clinical trials that led to the approval of targeted AML therapies, thereby limiting availability of data for clinical decision making.² The non-canonical FLT3 N676K variant was initially described as an acquired resistance mechanism in patients with FLT3 internal tandem duplication (ITD) mutations treated with midostaurin.³ In vivo studies demonstrated FLT3 N676K-mutated AML is sensitive to midostaurin and quizartinib, but suggested that co-operating ITD mutations confer resistance to both agents.⁴ Clinical reports of FLT3 N676K-mutated AML are limited to those of two individuals, both of whom developed FLT3 N676K mutations at relapse.^4,5 Treatment outcomes for de novo disease with FLT3 N676K mutations are lacking, and limited data exist regarding the utility of FLT3 inhibitors for FLT3 N676K-mutated AML patients. The aim of this study was to use clinical and genomic data to investigate the efficacy of FLT3 inhibitors, both as monotherapy and in combination with other agents, for FLT3 N676K-mutated AML.

We performed a retrospective analysis of patients with AML receiving care at the University of Chicago. The study was approved by the Institutional Review Board and conducted according to the Declaration of Helsinki. Our practice utilizes a validated 1,213 gene next-generation sequencing (NGS) panel that has been previously described.^6-8 NGS is employed at presentation and at subsequent time points to assess response or disease status. In cases of morphologic remission, NGS is not performed due to the anticipated lack of detectable tumor DNA. We identified nine patients with AML and FLT3 N676K mutations. N676K was the only FLT3 mutation detected in seven patients, whereas two patients had co-incident ITD or tyrosine kinase domain (TKD) mutations at some point during their clinical course. Two patients in the study were referred from outside of our institution and had an unknown FLT3 mutation status at first presentation. The remaining seven patients were assessed by upfront NGS, and all patients had molecular reassessment longitudinally, including at relapse, by our in-house assay. The median age at AML diagnosis was 41 years (range, 17-79), with a mean presenting white blood cell (WBC) count of 53,300/µL. Four of seven (57%) patients with available cytogenetic data had normal cytogenetics. Laboratory and clinical data can be found in Table 1.

In order to infer antileukemic activity of FLT3 inhibitors for FLT3 N676K-mutated AML and to characterize relapse dynamics, we analyzed FLT3 N676K variant allele frequency (VAF) kinetics in patients for whom longitudinal NGS data were available. Regardless of disease status at the time of FLT3 inhibitor use (newly diagnosed vs. relapse), patients receiving FLT3 inhibitors had declines in FLT3 N676K VAF. For individuals in whom FLT3 N676K was the only FLT3 mutation (patients 1, 2, 9), a mean treatment time of 95 days led to undetectable FLT3 N676K. Suppression of FLT3 N676K VAF generally paralleled clinical response and likelihood of survival at the time of our analysis (Figure 1A). One exception was patient 9, who achieved morphologic and molecular remission with gilteritinib salvage therapy but who unfortunately died of post-transplant veno-occlusive disease after exposure to gemtuzumab ozogamicin and subsequent allogeneic transplant.

One older patient pursued comfort care immediately after diagnosis. Six patients were treated with “7+3” induction therapy, with three of six patients having FLT3 inhibitors added to induction chemotherapy (midostaurin, n=2; sorafenib, n=1). Patient 1 had a complete morphologic and molecular remission. They proceeded to allogeneic transplantation after induction 7+3 therapy with sorafenib (discontinued for gastrointestinal toxicity) and subsequent midostaurin during pretransplant consolidation. Day 30 post-transplant bone marrow studies demonstrated a complete morphologic and molecular remission on midostaurin maintenance therapy.

Two additional patients (patients 5 and 6) harbored de novo disease with co-occurring FLT3 mutations, one with a FLT3 ITD mutation (patient 5) and the other a FLT3 TKD mutation (patient 6). Each had midostaurin added to 7+3 induction. By days +71 and +32 after midostaurin treatment, respectively, both demonstrated remission with no detectable FLT3 VAF (Figure 1B).

Patient 5 had de novo disease with co-occurring FLT3 mutations (FLT3 ITD F612_G613ins25 and N676K). She received 7+3+midostaurin induction and had a morphologic and molecular complete remission 71 days after initiating midostaurin. Induction was complicated by fungal pneumonia and repeated episodes of acute kidney injury. She was not a candidate for cytotoxic consolidation therapy and started gilteritinib. She continued to experience multiple episodes of acute kidney injury unrelated to gilteritinib. Gilteritinib was held during these episodes, and she eventually presented with 25% circulating blasts after approximately 100 days of intermittent gilteritinib administration. NGS at relapse showed ascendancy of the same FLT3 ITD clone (VAF 43%) that was present at diagnosis, but an absence of the FLT3 N676K mutation. The patient chose comfort measures.

Table 1.Clinical and laboratory characteristics of patients with FLT3 N676-mutated acute myeloid leukemia.

Patient 6 had de novo disease with co-occurring FLT3 TKD (D835V) and N676K mutations. She underwent induction with 7+3+midostaurin, which led to morphologic and molecular remission 32 days after initiating midostaurin. She proceeded to hematopoietic cell transplantation in first remission. Despite MRD negativity at transplant, she relapsed after 6 months. She did not receive post-transplant FLT3 inhibitor therapy. NGS performed at relapse demonstrated expansion of the FLT3 N676K population (VAF 33%) and absence of the original FLT3 TKD clone. Salvage measures with donor lymphocyte infusion and high dose cytarabine/mitoxantrone were unsuccessful. She died of complications from central nervous system leukemic infiltration.

We also analyzed the spectrum of other pathogenic mutations co-existing with FLT3 N676K in our cohort (Online Supplementary Figure S1). Co-mutational clusters were most notable for FLT3 N676K and either FLT3 ITD or FLT3 TKD mutations (Figure 1C). In order to understand the structural properties of therapeutic inhibition of FLT3 in the presence of the N676K mutation, we utilized PyMOL (Schrödinger), an open source molecular graphics tool that is commonly used for visualization of macromolecules, to study the FLT3 TKD harboring N676K in the presence and absence gilteritinib.⁹ Upon activation, three residues, Asp-Phe-Gly (DFG), shift inward (DFG-in) from the inactive state (DFG-out). Mutations at D835 within the TKD favor the active DFG-in state and promote resistance to type II FLT3 inhibitors.¹⁰ Gilteritinib and other type I FLT3 inhibitors bind directly to the ATP-binding site, maintaining their activity regardless of DFG conformation.¹¹ As shown in Figure 2, the N676K mutation does not prohibit the transition from DFG-out to DFG-in or the interaction of gilteritinib with the ATP-binding site.

A recent analysis of the mutational landscape of patients with FLT3-mutated AML treated on CALGB 10603 (RATIFY) showed 26 of 275 (5.5%) patients harbored non-canonical FLT3 mutations.² Ten of these 26 patients (38%) had FLT3 N676K-mutated AML.² Growing clinical application of NGS will increase the identification of atypical driver mutations. Robust clinical series focused on FLT3 N676K-mutated AML patients are lacking, and the benefit of FLT3 inhibitor therapies in this population was previously unknown.

Figure 1.Clinical and molecular characterization of acute myleoid leukemia harboring atypical FLT3 N676K mutations. (A) Longitudinal detection of FLT3 N676K by next-generation sequencing from the time of FLT3 inhibitor initiation. (B) Clinical course and relapse dynamics of 2 patients with FLT3 N676K-mutated acute myeloid leukemia (AML) and coincident internal tandem duplication or tyrosine kinase domain mutations. (C) Co-mutational networks of FLT3 N676K-mutated AML seen in 2 or more patients within the cohort. Chord thickness reflects the number of co-occurrences between 2 genes.

Figure 2.Macromolecular modeling of the FLT3 tyrosine kinase domain. (A) PyMol visualization models of the uninhibited FLT3 tyrosine kinase domain with N676K (residue shown in orange) and activation loop (yellow). Residues F612 and D835 (corresponding to internal tandem duplication and tyrosine kinase domain mutations, respectively) are shown in magenta. (B) Upon activation, the 18-fluoro-deoxyglucose (FDG)-containing α helix (salmon) translocates inward. Inhibition by gilteritinib at the ATP-binding site is depicted in the inactive, FDG-out (C) and activated, FDG-in (D) state with the FDG-containing α helix shown in red.

Here, we used clinical and genomic data to assess the utility of FLT3 inhibitors in the largest series of FLT3 N676K AML patients described to date. Although previously described to be enriched in populations of core binding factor AML,⁴ none of the seven patients in our cohort with available metaphase cytogenetic data available had core-binding factor AML. Seven of nine (78%) individuals received intensive induction chemotherapy. FLT3 inhibitors were utilized in three patients during frontline induction in combination with cytotoxic chemotherapy, and in five patients during subsequent lines of therapy. We observed reduction, and in some instances, complete molecular suppression of detectable FLT3 N676K VAF on NGS, underscoring the activity of FLT3 inhibitors in this population, regardless of the line of therapy. All patients with FLT3 N676K mutations who were treated with FLT3 inhibitors had a best response of MRD negativity via flow cytometry or NGS at some point during their care.

Consistent with previous evidence,^3,4 concurrent canonical ITD or TKD FLT3 mutations were associated with loss of treatment response. In silico modeling of FLT3 in the presence of gilteritinib suggests that the mechanism of N676K-mediated resistance is not due to disruption of FLT3 inhibitor binding at the ATP-binding site but is likely influenced by other allosteric forces on the protein structure. Individuals with FLT3 N676K-mutated AML in our cohort whose treatment included FLT3 inhibitors had longer median survival (940 days) than those who did not (408 days, excluding patient 4 who immediately pursued comfort measures). This difference was not significant, likely because of the small size of this study (P=0.2). The three patients who remained in an ongoing remission at the time of manuscript submission, however, were all treated with FLT3 inhibitors. With emerging evidence supporting the role of post-transplant FLT3 inhibitor maintenance therapy for suppression of FLT3 ITD-mutated AML,¹² further studies evaluating the durability of FLT3 inhibitor maintenance for patients with non-canonical driver mutations in both transplant and non-transplant settings is warranted.

In conclusion, this is the largest study to date demonstrating that the atypical FLT3 N676K driver mutation is sensitive to contemporary FLT3 inhibitors, such as midostaurin and gilteritinib. This mutation has been infrequently detected in seminal studies of FLT3 inhibitors. However, our data demonstrate FLT3 inhibitors should be included both in upfront induction setting and relapsed/refractory settings for patients harboring the atypical FLT3 N676K mutation.

Footnotes

• Received September 22, 2022
• Accepted January 9, 2023

Correspondence

M.W. Drazer - mdrazer@medicine.bsd.uchicago.edu

Disclosures

ASA has acted as a consultant for AbbVie and Magenta Therapeutics. WS has acted as a consultant or advisor to Adaptive Biotechnologies, Jazz Pharmaceuticals, Agios, Kite, a Gilead company, Kura Oncology, GlaxoSmithKline, MorphoSys, Pfizer, Servier, has received honoraria from AbbVie, has received royalties for a chapter in UpToDate, and has received travel accommodation from Pfizer. OO has acted as a consultant for Abbvie, Impact Biomedicines, Celgene, Novartis, BMS, Taiho Pharmaceutical, CTI, Threadwell therapeutics, Bristol-Myers Squibb/Celgene, and has received research support to her institution from Celgene, Daichii Sankyo, Uncyte, Astex Pharmaceuticals, NS Pharma, AbbVie, Janssen Oncology, OncoTherapy Science, Agios, AstraZeneca, CTI BioPharma Corp, Kartos Therapeutics and Aprea AB. RAL has acted as a consultant or advisor to Ariad/Takeda, Celgene/BMS, CVS/Caremark, Epizyme, Immunogen, Novartis, and Servier, and has received clinical research support to his institution from Astellas, Cellectis, Daiichi Sankyo, Forty Seven/Gilead, Novartis, and Rafael Pharmaceuticals, and royalties from UpToDate. HL has acted as a consultant or advisor to Agios, Pfizer, Nkarta, CTI Biopharm, Servier, NGM Biopharma, has acted as a speaker/lecturer for SITC, CAHON, Academy for Continued Healthcare Learning, and has received research support from Miltenyi Biotec. LAG has received royalties from UptoDate, Inc. for a co-authored article on germline predisposition to hematopoietic malignancies. MJT reports grant support from AbbVie, Merck, Syndax, and TG Therapeutics and has received personal fees from AbbVie, Adaptive Biotechnologies, AstraZeneca, Celgene, Pharmacyclics, and Genentech. ASD has acted as a consultant or advisor to Jazz Pharmaceuticals and has served on a speakers’ bureau for Jazz Pharmaceuticals. AAP has received honoraria from AbbVie and research funding from Celgene/BMS, Pfizer and Kronos Bio. CKD has received consulting/advisory fees from Daiichi Sankyo and Sun Pharma. ASD has received fees for consulting and is serving as a member of a speakers’ bureau for Jazz Pharmaceuticals. ED has received honoraria from AbbVie. GDW has received honoraria and has served on an advisory board for Diagnostica Stago. The remaining authors have no conflicts of interest to declare.

Contributions

MWD conceived the study. MWD and GWR developed the concept and design of the study. MWD, GWR, FW and AS collected and analyzed the data. MWD, GWR, ASA, SK, WS, OO, RAL, HL, LAG, MJT, AAP, CKD, ASD, MTN and ED cared for the patients described. GDW, SG, CF, JPS, PW provided pathological support. GWR and MWD drafted the manuscript. All authors contributed to editing the manuscript.

Data-sharing statement

Data from the current work are available on request.