Cytomolecular Mechanisms of Relapse and Post‐Relapse Outcomes After Frontline FLT3 Inhibitor‐Based Therapy in FLT3‐Mutated AML

For patients with newly diagnosed FLT3-mutated AML, FLT3 inhibitors (FLT3i) are standard of care in combination with intensive chemotherapy (IC) for younger, fit patients and are increasingly being used in combination with a hypomethylating agent (HMA) and venetoclax for older patients who are unsuitable for IC [1-5]. However, relapse after frontline treatment with a FLT3i-containing regimen is common, with a 3-year cumulative incidence of relapse about 30%–35% with IC + FLT3i regimens [1, 6]. Studies focusing on mechanisms of relapse to FLT3i-based therapy have shown that clonal emergence of pro-survival RAS-MAPK pathway mutations is a common off-target mechanism of resistance to type 1 FLT3i, while the development of a new FLT3-TKD mutation is a common resistance mechanism to type 2 FLT3i [7, 8]. However, it is not known whether these molecular mechanisms of relapse differ based on the type of frontline regimen used or by prior allogeneic stem cell transplantation (ASCT) at first remission. We therefore conducted a retrospective study including adult patients with newly diagnosed FLT3-mutated AML [ITD and/or TKD (D835/D836)] who received frontline FLT3i-containing regimen at our institution between September 2013 and July 2024. Mutation and cytogenetic data from diagnostic bone marrow samples and at the time of relapse were compared. Additional study details are available in the Data S1. A total of 272 patients met study inclusion criteria (Figure S1). The median age was 66 years (range 18–91). Two hundred fifteen patients had a FLT3-ITD mutation, 31 had a FLT3-TKD mutation, and 26 had both FLT3-ITD and FLT3-TKD mutations. The induction regimen used was IC + FLT3i in 107 patients (39%), 6 of whom also received venetoclax, and lower-intensity therapy (LIT) + FLT3i in 165 patients (61%), 93 of whom received HMA + venetoclax + FLT3i (referred to as the "triplet" regimen). For the entire cohort, gilteritinib was the most commonly used FLT3i (n = 105, 39%), followed by sorafenib (n = 96, 35%), quizartinib (n = 54, 20%), midostaurin (n = 16, 6%), and crenolanib (n = 1, 0.4%). Morphological response was achieved in 227 patients (84%), including complete remission (CR) in 164 patients (60%), complete remission with incomplete hematologic recovery (CRi) in 39 patients (14%), and morphologic leukemia free state (MLFS) in 24 patients (9%). A total of 96 patients (35% of the entire cohort; 42% of responders) underwent ASCT in first morphological remission. Post-ASCT FLT3i maintenance was given to 52 (55%) of the 95 patients with available post-ASCT data. After a median follow-up of 46 months (95% confidence interval [CI]: 34.2–58.3 months), 80 patients (35% of responders) relapsed and are the primary focus of this analysis. Relapse rates (among responders) in patients receiving frontline IC + FLT3i (with/without venetoclax), HMA + venetoclax + FLT3i, and LIT + FLT3i (without venetoclax) were 23% (p < 0.01), 31% (p < 0.01), and 66% (ref), respectively. Patients who underwent ASCT in 1st remission had lower relapse rates compared to those who did not (23% versus 44%, p < 0.01). The median time to relapse from the date of initial response was 6.6 months (95% CI: 5.7–8.1 months). The baseline characteristics of the relapsed cohort (n = 80) are shown in Table S1. Of note, 22 patients (28%) in the relapsed cohort had received ASCT in first remission, with a median time from ASCT to relapse of 5.7 months (95% CI: 3.1–9.9 months). Twelve of these patients (55% of post-ASCT relapses) had received post-ASCT FLT3i maintenance. Loss of FLT3-ITD mutation at the time of relapse was noted in 27/67 (40%) evaluable patients, and loss of FLT3-TKD mutation at relapse was noted in 8/13 (62%) evaluable patients. Taken together, loss of a FLT3 mutation (FLT3-ITD and/or FLT3-TKD) at relapse was noted in 34/72 patients (47%) (Figure 1A). A total of 28/72 patients (39%) tested negative for both FLT3-ITD and FLT3-TKD at relapse (hence referred to as "FLT3 wild type relapses" for later analyses). Patients who either received IC + FLT3i or the HMA + venetoclax + FLT3i triplet were more likely to have loss of a FLT3 mutation at relapse compared with those who received LIT + FLT3i without venetoclax (57% versus 32%, respectively; p = 0.05). There were no differences in rates of loss of a FLT3 mutation at relapse when stratified by 1st generation versus 2nd generation FLT3i (47% in both groups; p > 0.99), ASCT in first remission versus no ASCT (50% versus 46%; p = 0.8), and use of post ASCT FLT3i maintenance versus no FLT3i maintenance in transplanted patients (58% versus 40%; p = 0.67). However, there was a trend toward higher rates of FLT3 wild type relapses (i.e., negative for both FLT3-ITD and FLT3-TKD mutations) in transplanted patients who received post-ASCT FLT3i maintenance compared to those who did not (58% versus 20%, p = 0.10). Among patients in the relapsed cohort who were FLT3-TKD wild type at baseline with available testing at relapse, 8/59 (14%) had a new emergent FLT3-TKD mutation at relapse, with a median variant allelic frequency (VAF) of 29% (range, 5%–59%). Among these, 7 (88%) had received frontline type 2 FLT3i that is not active against FLT3-TKD mutations (sorafenib in 5 patients, quizartinib in 2 patients). A total of 56 patients underwent mutational profiling at relapse, and new/emergent non-FLT3 mutations were detected in 28 patients (50%). The most common emergent mutations included: DNA methylation mutations (DNMT3A, TET2, IDH1, IDH2) in 12 patients (21%), median VAF 4% (range, 1%–35%), RAS pathway mutations (i.e., KRAS, NRAS, PTPN11, CBL, NF1) in 9 patients (16%), median VAF 9% (range, 2%–37%) and a WT1 mutation in 7 patients (13%), median VAF 6% (range, 2%–40%). Patients who received an HMA + venetoclax + FLT3i triplet regimen had numerically higher rates of emergent RAS pathway mutations at relapse compared with those who received other regimens (25% versus 9%, respectively; p = 0.15; Figure 1B), though this was not statistically significant. Among post-ASCT relapses, emergent non-FLT3 mutations were noted in 5/21 patients (24%) with available molecular testing at relapse. These included WT1 (n = 2), PTPN11 (n = 1), TET2 (n = 1), RUNX1 (n = 1), SF3B1 (n = 1), and BCORL1 (n = 1). Patients who relapsed after undergoing ASCT in first remission had significantly lower rates of emergent non-FLT3 mutations at relapse compared with those who did not undergo ASCT (24% versus 66%, respectively; p < 0.01). Data comparing molecular patterns of relapse specifically between patients receiving IC + FLT3i and HMA + venetoclax + FLT3i regimens is shown in Table S2. Additionally, data regarding cytogenetic clonal evolution at relapse is also shown in the Supporting Information. Outcomes after relapse were poor. At data cut off, five patients (6%) from the relapsed cohort (n = 80) were still alive with a median OS after relapse of 5.5 months (95% CI: 3.3–6.5 months; Figure 2A). When stratified by FLT3 status at relapse, FLT3 wild type relapses had a trend toward higher median OS compared with FLT3-mutated relapses (7.0 versus 4.1 months, respectively; p = 0.07; Figure 2B). There was no difference in OS among patients who had emergent non-FLT3 mutations at relapse compared to those who did not (5.5 versus 6.5 months, respectively; p = 0.33). A total of 51 patients (64%) from the relapsed cohort received at least one salvage therapy; this included IC in 11 patients (22%), LIT in 39 patients (77%), and immediate ASCT in 1 patient (2%); 23/29 patients (79%) with a persistent FLT3 mutation at relapse received a FLT3i-containing salvage regimen. Following the first salvage therapy, composite complete remission (CR/CRi) was attained in 12/51 patients (24%), including CR in 5 patients (10%) and CRi in 7 patients (14%); MLFS was the best response in an additional 10 patients (20%). Patients with FLT3 wild type relapses had higher CR/CRi rates following salvage therapy compared to those with FLT3-mutated relapses (41% versus 10%, respectively; p = 0.02). Patients who had relapsed after frontline IC + FLT3i and HMA + venetoclax + FLT3i had similar CR/CRi rates with the first salvage therapy (29% in both groups). Among patients who received salvage therapy after relapse, the median OS following the start of salvage therapy was 6.2 months (95% CI: 3.9–7.0 months; Figure 2C). Patients with FLT3 wild type relapses had a trend toward superior OS after salvage therapy as compared with FLT3-mutated relapses (8.6 months versus 5.7 months, respectively; p = 0.13; Figure 2D). Overall, our study adds to the existing literature on the mechanisms of relapse following frontline FLT3i therapy in FLT3-mutated AML and is the first to demonstrate that the use of an HMA + venetoclax + FLT3i triplet regimen in induction and ASCT in first remission may influence these mechanisms. We found that loss of FLT3 and emergence of new DNA methylation/RAS pathway/WT1 mutations are common at relapse following frontline FLT3i therapy. FLT3 mutations are lost in approximately 50% of patients at relapse, particularly when more effective frontline therapies are used, and patients with persistent FLT3 mutations at relapse have the worst outcomes. We show that ASCT may alter the patterns of molecular clonal evolution following frontline FLT3i therapy. In our cohort, post-ASCT relapses had lower rates of new emergent non-FLT3 mutations at relapse; however, there was no difference in the rates of FLT3 loss at relapse based on ASCT in first remission. These findings suggest that post-ASCT relapses in FLT3-mutated AML are mechanistically complex and cannot be explained by molecular clonal evolution alone. Our data underscores the need for novel therapies targeting the RAS/MAPK pathway in AML, with emergent RAS pathway mutations being common at relapse, particularly after a frontline HMA + venetoclax + FLT3i triplet regimen. RAS pathway mutations are a well-established resistance mechanism to both venetoclax [9] and FLT3i [10], and therefore it is plausible that co-administration of these agents confers a survival advantage and selection of RAS-mutated clones. We also show that FLT3 loss is commonly observed after these frontline FLT3i-based therapies. Therefore, efforts to improve outcomes in newly diagnosed FLT3-mutated AML should focus not only on eradication of FLT3-mutated subclones and established resistance mutations (e.g., RAS pathway) but also on prevention of FLT3 wild type relapses. S.A. and N.J.S. designed the manuscript. S.A., S.P., and J.M. collected the data. S.A., O.K., and N.J.S. analyzed the data. S.L. provided the laboratory data. S.A. made the figures. N.D., F.R., C.D.D., T.M.K., G.B., E.J., M.Y., G.C.I., N.J.S., and H.K. provided patients. S.A. wrote the first manuscript draft. N.J.S. and S.A. revised the drafts. The final draft was reviewed and approved by all authors. BioRender.com was used to create Figures 1 and S1: permission to publish is available on request. This study was approved by the MD Anderson Cancer Center institutional review board. S.A.: No relevant COI. S.L.: Consultancy/ad board: Guidepoint, QualWorld, Gerson Lehrman Group, AlphaSight, Arima, Qiagen, Opinion Health, Abbvie, Daiichi Sankyo, BluePrint Medicine, Caris Diagnostics, Recordati, and Servier; Kura Oncology, Syndax; Grant and contract support: Astellas, Amgen; Stock from: Abbvie, PureTech. N.D.: Grants/contracts: AbbVie Inc., Amgen, Astellas Pharma, Bristol Myers Squibb Company, Daiichi Sankyo Company, FATE Therapeutics, Genentech, Gilead Sciences Inc., Glycomimetics, Hanmi, ImmunoGen Inc., Kite Pharma Inc., Novimmune, Pfizer Inc., Servier Pharmaceuticals LLC, Trillium, and Trovagene; and Personal/consulting fees: AbbVie, Agios Pharmaceuticals Inc., Amgen, Arog Pharmaceuticals, Astellas Pharma, Bristol Myers Squibb Company, Celgene, Daiichi Sankyo Company, Genentech, Gilead Sciences Inc., ImmunoGen Inc., Jazz Pharmaceuticals, Kite Pharma Inc., Novartis, Pfizer Inc., Servier Pharmaceuticals LLC, Shattuck Labs, Stemline/Menarini, Syndax, and Trillium. F.R.: Grants/contracts: Bristol Myers Squibb Company and Taiho Pharmaceutical; Personal/consulting fees: Amgen, Astellas Pharma, Bristol Myers Squibb Company, Celgene Corporation, Novartis Foundation, and Taiho Pharmaceutical; and Support for other professional activities: Celgene Corporation. C.D.D.: Research funding: Abbvie, Astex, ImmuneOnc, BeiGene, BMS, Foghorn, Remix Rigel, Servier Schrodinger; Consulting: Abbvie, Astellas, AstraZeneca, Beigene, BMS, Genmab, Astellas, GSK, Molecular Partners, Ryvu, Schrodinger, Servier, Stemline; Honoraria: Abbvie, Astellas, BMS, Jazz, Servier; Support for attending meetings/travel: Servier; Data Safety Board: Genmab. T.M.K.: Grants/contracts: Ascentage Pharma, Bristol Myers Squibb Company, Genentech, Incyte Corporation, and Jazz Pharmaceuticals; Personal/consulting fees: AbbVie, Agios Pharmaceuticals Inc., Daiichi Sankyo Company, Genentech, Novartis, Servier Pharmaceuticals, and Syndax. G.B.: No relevant COI. E.J.: No relevant COI. M.Y.: No relevant COI. G.C.I.: Personal/consulting fees: Novartis, Sanofi, and Syndax; Data and safety monitoring: Novartis. J.M.: No relevant COI. O.K.: No relevant COI. S.P.: No relevant COI. H.K.: Grants to the institution: AbbVie, Amgen, Ascentage, BMS, Daiichi-Sankyo, Immunogen, Novartis; Consulting: AbbVie, Amgen, Ascentage, Ipsen Biopharmaceuticals, KAHR Medical, Novartis, Pfizer, Shenzhen Target Rx, Stemline, Takeda. N.J.S.: Consulting: Pfizer Inc., GSK, Autolus, and Adaptive Biotechnologies; Research funding: Takeda Oncology, Astellas Pharma Inc., Xencor, GSK, NextCure, Ascentage, Novartis, Hemogenyx, and Vironexis; Honoraria: Adaptive Biotechnologies, Amgen, Takeda, Pfizer Inc., Astellas Pharma Inc., and Sanofi. Blinded clinical data can be procured from corresponding author on reasonable request. Data S1. Figure S1. Study consort diagram. Table S1. Baseline characteristics of patients with FLT3-mutated AML who relapsed after frontline FLT3 inhibitor-based therapy (N = 80). Table S2. Comparison of molecular patterns at relapse after frontline IC + FLT3i versus HMA + VEN + FLT3i regimens. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.