Menin inhibition with revumenib for NPM1-mutated relapsed or refractory acute myeloid leukemia: the AUGMENT-101 study Open Access

Acute myeloid leukemia (AML) is a clonal hematopoietic malignancy, characterized by the rapid growth of myeloid stem cells that fail to differentiate into functional cells.¹ Nucleophosmin 1 (NPM1) mutations, the most common genetic aberrations in adult AML, are found in ∼28% of new cases.² NPM1 is an intracellular chaperone protein predominantly found in the nucleolus with roles vital for cellular processes, such as genetic stability.^3,4 Mutations in NPM1 lead to cytoplasmic translocation of the protein; however, a fraction of mutant NPM1 remaining in the nucleolus interacts with the lysine methyltransferase 2A (KMT2A)–menin complex and influences oncogenic gene transcription.^4-6 In NPM1-mutated (NPM1m) AML, as in KMT2A-rearranged (KMT2Ar) acute leukemias, the KMT2A–menin interaction leads to aberrant homeobox (HOX)–mediated and Meis homeobox 1 (MEIS1) oncogenic expression that blocks stem cell differentiation.^4-8 Selectively blocking the interactions between KMT2A and menin reverses the aberrant expression of these critical leukemogenic targets (ie, MEIS1 and HOX).^7,8 

Currently, no therapies are approved for patients with NPM1m AML. Although NPM1m AML displays favorable response rates with intensive chemotherapy or venetoclax plus hypomethylating agents in frontline settings,^2,9,10 there is no standard-of-care or targeted therapies in case of relapsed or refractory (R/R) disease.¹¹ On recurrence, the disease becomes difficult to treat, especially if the disease-free interval is short.¹² After first-line therapy, patients with NPM1m AML often have improved outcomes compared with those without NPM1m^9,10; however, ∼50% of adult patients with NPM1m AML experience progressive disease or death.^13,14 Responses to salvage therapy can be achieved after first relapse,^15,16 but time to second relapse, response to subsequent salvage therapies, and overall survival (OS) shorten with each subsequent line of therapy, which is similar to patients with NPM1 wild-type AML.¹⁷ Outcomes after venetoclax failure are poor, with overall response rates (ORRs) ranging from 6% to 23%.^18-21 Comutations are also very common in NPM1m AML, including at relapse; FLT3-ITD (fms-related receptor tyrosine kinase 3 internal tandem duplication), DNMT3A (DNA methyltransferase 3 alpha), and WT1 (WT1 transcription factor) comutations are associated with poorer outcomes and an increased incidence of relapse after achieving measurable residual disease negativity.²² The poor outcomes associated with R/R NPM1m AML, combined with limited treatment options and high relapse rates, highlight the need for new therapies to improve patient outcomes.

Revumenib is a first-in-class, potent, oral menin inhibitor that selectively blocks the KMT2A–menin interaction,^8,23 resulting in downregulation of MEIS1 and HOX expression and consequently enabling terminal differentiation of leukemic to normal hematopoietic cells.^8,23-25 Revumenib was recently approved for the treatment of R/R KMT2A-translocated acute leukemia in patients ≥1 year of age based on data from AUGMENT-101 (ClinicalTrials.gov identifier: NCT04065399), an ongoing phase 1/2 study in patients with R/R KMT2Ar acute leukemia or R/R NPM1m AML.²⁶ Here, we report the primary efficacy analysis of revumenib in patients with R/R NPM1m AML from the phase 2 portion of the AUGMENT-101 study.

AUGMENT-101 is a phase 1/2, open-label, dose-escalation and -expansion study of revumenib in adult and pediatric (≥30 days old) patients with documented R/R NPM1m or KMT2Ar acute leukemias. NPM1 mutation status was determined by local testing for eligibility; all treated patients with NPM1m AML were in the safety analysis population. Patients with centrally confirmed NPM1 mutation (using Focus Myeloid panel, Flagship Biosciences, Inc; or NPM1 Mutation assay, Invivoscribe, Inc) and ≥5% blasts in the bone marrow at baseline within 28 days before the start of study treatment were considered efficacy evaluable. As prespecified in the statistical analysis plan and protocol, the first 64 adult patients in the study who met these criteria were included in the adult efficacy-evaluable population. There was no restriction on the number or types of previous therapies. Comutation testing was performed locally but was not required. Patients with central nervous system (CNS) disease at the most recent relapse were eligible if no active CNS disease remained present at the start of study therapy. Ongoing intrathecal therapy or prophylaxis was allowed concurrent with revumenib. See the Study Population section in the supplemental Clinical Study Protocol for full inclusion and exclusion criteria.

Revumenib was administered orally in capsule, tablet, or liquid formulation every 12 hours in 28-day continuous cycles. The recommended phase 2 dose of revumenib was 270 mg (160 mg/m² if body weight <40 kg) every 12 hours or, given that revumenib is a cytochrome P450 3A4 (CYP3A4) substrate, 160 mg (95 mg/m² if body weight <40 kg) every 12 hours if patients were also receiving a strong CYP3A4 inhibitor.²⁵ Revumenib treatment was continued until lack of response after up to 4 cycles, disease progression, unacceptable adverse events (AEs), withdrawal of consent, investigator decision, or loss to follow-up (see supplemental Clinical Study Protocol for further details).

Patients who achieved composite complete remission (CRc; complete remission [CR] + CR with partial hematologic recovery [CRh] + CR with incomplete hematologic recovery [CRi] + CR with incomplete platelet recovery [CRp]), morphological leukemia-free state, or partial remission were allowed to undergo allogeneic hematopoietic stem cell transplant (HSCT) without leaving the study. Revumenib was stopped before the HSCT conditioning regimen but, per a protocol amendment (approved on 24 September 2021, before enrollment of all patients in this analysis), could be resumed as maintenance therapy after allogeneic HSCT if the patient was between 30 and 180 days post-HSCT, had successful engraftment, did not have acute or chronic graft-versus-host disease that required systemic immunosuppression, and remained in CRc.

The primary end points were the rate of CR + CRh and the evaluation of safety and tolerability of revumenib. Secondary end points included rate of CRc, overall response rate (ORR) (CRc + morphological leukemia-free state + partial remission), time to response (defined as the number of months from the date of first dose to the date of initial response [CR + CRh or CRc]), duration of response, event-free survival (defined as the number of months from the date of first dose to the date of first documented relapse/progression or death, whichever occurred first; supplemental Table 1, available on the Blood website), and OS. Responses were assessed by the investigators according to the European LeukemiaNet 2017 response criteria.²⁷ CRp was defined as all CR criteria except for platelet count <100 × 10⁹/L. Details on secondary end points and response definitions are available in the supplemental Clinical Study Protocol.

AEs were collected from the time of the first revumenib dose to 30 days after the last dose, including maintenance therapy, and were graded using the National Cancer Institute Common Terminology Criteria for Adverse Events version 5.0. AEs of special interest included differentiation syndrome (an expected on-target effect of inducing differentiation of leukemia cells into normal hematopoietic cells), prolongation of the corrected QT interval by Fridericia (QTcF) interval of grade ≥2, and peripheral neuropathy. Study investigators received guidelines for managing AEs of special interest (supplemental Methods). Hydroxyurea for cytoreduction, intrathecal chemotherapy for CNS prophylaxis, and steroids for differentiation syndrome were allowed during the study.

Transcriptional changes after 1 cycle of revumenib were evaluated for target genes of interest. Samples from bone marrow aspirates (RNA) were isolated using the Maxwell RSC simplyRNA kit (Promega Corporation) and quantified using the NanoDrop 2000 (Thermo Fisher Scientific Inc). Total RNA quality and molecular weight distribution were evaluated using the Agilent 2100 Bioanalyzer (Agilent Technologies, Inc). Multiplex gene analysis was performed using a custom-designed QuantiGene assay (Thermo Fisher Scientific Inc). Seven genes of interest (MEIS1, HOXA9, PBX3, FLT3, CD11b, CD14, and CD13) and 5 housekeeping genes (PGK1, B2M, RPL13A, POLR2A, and HPRT1) were included in the custom assay. PGK1 and HPRT1 were selected as housekeeping genes for analysis because of variability found in the other housekeeper genes. The raw data were analyzed using QuantiGene Plex Data Analysis software (Thermo Fisher Scientific Inc) to generalize normalized expression data. All steps were performed by Flagship Biosciences.

Baseline genetic alterations were evaluated to determine whether a particular comutation was associated with response. Bone marrow–derived DNA from adult patients in the protocol-defined, efficacy-evaluable population, collected before the start of revumenib treatment, was genomically profiled using the Focus Myeloid panel conducted by Flagship Biosciences. The gene coordinates used were part of the standard TruSight Myeloid Sequencing Panel (Illumina, Inc., San Diego, CA; supplemental Table 2), which is commercially available.²⁸ The analytical sensitivity of this assay is 5% at >500× read depth, with 100% accuracy demonstrated during validation.²⁸ Somatic short variants were identified and analyzed according to the manufacturer’s instructions.

The sample size was driven by the primary analysis in the adult efficacy-evaluable population. The number of adult efficacy-evaluable patients in the R/R NPM1m AML population evaluated in each stage, including the minimum number of responders needed to continue to the next stage, were determined based on the minimax version of Simon’s 2-stage design,²⁹ with 90% power and a 1-sided significance level of 2.5%. For this primary analysis, 64 adult efficacy-evaluable patients were included in the R/R NPM1m AML population. The primary hypothesis test used a null hypothesis of a 10% CR + CRh rate. A CR + CRh rate >10% was considered the lower threshold for antileukemic activity.

Safety was summarized for all patients with R/R NPM1m AML who received ≥1 dose of revumenib (safety population). No formal statistical hypothesis testing for safety analyses was conducted. Time-to-event end points were estimated using the Kaplan-Meier method, and descriptive statistics were used for other clinical or laboratory variables, with subgroup analyses performed for efficacy.

Statistical analysis of comutation data was performed by Fios Genomics Ltd to identify associations between clinical outcomes and gene mutations. In all analyses, an applied threshold of unadjusted P value <.05 defined an association as statistically significant.

Mutation data provided the input for statistical hypothesis testing, in which features that were significantly different between sample groups were identified. Statistical comparisons were performed using Fisher exact test. Significance values (P values) were adjusted for multiple testing by controlling the false discovery rate. For each binary comparison (eg, responders vs nonresponders), a positive log2 odds ratio indicated a positive association between responders and the presence of a mutation relative to nonresponders, whereas a negative log2 odds ratio indicated a positive association between nonresponders and the presence of a mutation relative to responders.

The study was conducted in accordance with principles outlined in the Declaration of Helsinki and the good clinical practice guidelines of the International Council for Harmonisation. The protocol and amendments were approved by the relevant authorities and institutional review board or ethics committee at participating centers, and all patients or their legal guardians provided written informed consent. Important changes to the methods to expand eligibility were to adjust the design to include children and to allow patients to resume revumenib treatment posttransplant. Throughout the study, an independent data monitoring committee monitored safety and efficacy according to predefined parameters detailed in the protocol and provided recommendations for continuing or terminating the study. The details of the study design are provided in the study protocol.

From 1 October 2021 to 18 September 2024, 84 patients with R/R NPM1m AML in 9 countries received ≥1 dose of revumenib and comprised the safety population (Figure 1). At baseline, the median age was 63 years (range, 11-84), and 83 patients (98.8%) were adults (Table 1). FLT3-ITD and FLT3-tyrosine kinase domain comutations were identified in 31.0% and 7.1% of patients, respectively; other comutations included isocitrate dehydrogenase 1 (IDH1; 13.1%), IDH2 (11.9%), TP53 (tumor protein 53; 4.8%), and RAS (3.6%). Patients were heavily pretreated, with 34.5% having received ≥3 previous lines of therapy (median, 2 [range, 1-7]); 73.8% received prior venetoclax, 38.1% received prior FLT3 inhibitor therapy, 6.0% received prior IDH1 inhibitor therapy, and 6.0% received prior IDH2 inhibitor therapy. Almost one-fourth of patients (23.8%) had undergone a prior HSCT, 8.3% had extramedullary disease at enrollment, and 2.4% had active CNS disease at their most recent relapse. The primary efficacy analysis included 64 efficacy-evaluable adults with centrally confirmed NPM1m AML and ≥5% blasts in bone marrow at baseline (Figure 1).

Treatment-emergent AEs were experienced by 83 patients (98.8%; grade ≥3, 77 [91.7%]; supplemental Table 3). A total of 66 of 84 patients (78.6%) experienced ≥1 treatment-related AE (TRAE) with revumenib (Table 2) and 50 patients (59.5%) experienced a grade ≥3 TRAE. Dose modifications occurred in 64 patients (76.2%), with 56 patients (66.7%) requiring dose interruptions (supplemental Table 4). TRAEs leading to dose reduction occurred in 10 patients (11.9%; Table 2). Four patients (4.8%) discontinued revumenib due to a TRAE (cardiac arrest, differentiation syndrome, osteomyelitis [n = 1 patient each], and 1 patient experienced QTcF prolongation and syncope). One patient (1.2%) died because of a treatment-related event (cardiac arrest); the investigator reported 2 possible causes (intracranial hemorrhage or arrythmia), and an autopsy was not performed (Table 2).

Treatment-emergent differentiation syndrome (any grade) occurred in 16 patients (19.0%), of whom 9 (10.7%) had a grade 3 event, 2 (2.4%) had a grade 4 event, and none had a grade 5 event. All 16 patients were treated with corticosteroids, with the addition of hydroxyurea for associated leukocytosis in 5 patients. Differentiation syndrome led to interruption of revumenib in 7 patients and discontinuation in 1 patient with a grade 3 event. The median time to initial onset was 10 days (range, 4-34), and median duration of the initial event of differentiation syndrome was 14.5 days (range, 3-57).

Treatment-emergent QTcF prolongation (any grade) occurred in 36 patients (42.9%), of whom 17 (20.2%) had a grade 3 event, 2 (2.4%) had a grade 4 event, and none had a grade 5 event. QTcF prolongation was managed per treatment guidelines as described in the supplemental Methods; the median times to initial onset and median duration of the initial event were 8 days (range, 1-84) and 4 days (range, 1-14), respectively. QTcF prolongation resulted in dose interruption in 18 patients, dose reduction in 8, and discontinuation in 1.

Grade ≥3 treatment-related cytopenias and electrolyte imbalances, which occurred in ≥5% of patients, included anemia (12 [14.3%]) and thrombocytopenia (8 [9.5%]). Revumenib dose reductions due to cytopenias were infrequent (thrombocytopenia, 1 [1.2%]; neutropenia, 1 [1.2%]; and white blood cell count decreased, 1 [1.2%]), and no patient discontinued due to grade ≥3 cytopenias or electrolyte imbalances.

The study met the primary efficacy end point for patients with R/R NPM1m AML, with 15 of 64 adult patients achieving CR or CRh (CR + CRh, 23.4%; 95% confidence interval [CI], 13.8-35.7; 1-sided P = .0014; Table 3). The rate of CRc was 29.7% (95% CI, 18.9-42.4). The ORR was 46.9% (95% CI, 34.3-59.8).

The median time to first response was 1.84 months (range, 0.9-4.6), and the median time to first CR or CRh was 2.76 months (range, 1.8-8.8; Figure 2). Changes in neutrophil and platelet counts in patients achieving CR or CRh are found in supplemental Figure 1.

Although the study was not powered to evaluate differences among subgroups, responses were observed across the various subgroups assessed. Notably, responses were found regardless of previous HSCT (CR + CRh rate [yes vs no], 21.4% [3/14; 95% CI, 4.7-50.8] vs 24.0% [12/50; 95% CI, 13.1-38.2]) and number of previous lines of therapy (CR + CRh rate [1 vs 2 vs ≥3 prior lines of therapy], 25.0% [4/16; 95% CI, 7.3-52.4] vs 20.0% [5/25; 95% CI, 6.8-40.7] vs 26.1% [6/23; 95% CI, 10.2-48.4]; Figure 3). CR + CRh rates were numerically similar in adults <65 years of age (7/31 [22.6%]; 95% CI, 9.6-41.1) and ≥65 years of age (8/33 [24.2%]; 95% CI, 11.1-42.3). The CR + CRh rate was 16.7% (8/48; 95% CI, 7.5-30.2) and 43.8% (7/16; 95% CI, 19.8-70.1) in patients with and without previous venetoclax exposure, respectively (Figure 3). In patients with previous FLT3 inhibitor use, the CR + CRh rate was 13.3% (4/30; 95% CI, 3.8-30.7), the CRc rate was 20.0% (95% CI, 7.7-38.6; CR, n = 4; CRp, n = 1; and CR with incomplete hematologic recovery, n = 1), and the ORR was 40.0% (95% CI, 22.7-59.4). Patients with IDH1 or IDH2 comutations with NPM1 mutation at baseline achieved CR + CRh at higher rates than the overall population (75.0% [6/8] and 50.0% [4/8], respectively).

Across all responders, the median duration of CR + CRh was 4.7 months (95% CI, 1.2-8.2; Figure 4A), and the median duration of CRc was 4.7 months (95% CI, 1.9-8.2). Across all 64 adult efficacy-evaluable patients, the median event-free survival was 3.0 months (95% CI, 2.0-3.8; Figure 4B) and the median OS was 4.0 months (95% CI, 2.5-7.2; Figure 4C); the median OS in the 15 CR + CRh responders was 23.3 months (95% CI, 7.2 to not reached [NR]).

One pediatric patient (female, 11 years old) was not included in the protocol-defined primary efficacy analysis consisting of the adult efficacy-evaluable population but otherwise met the efficacy-evaluable criteria. This patient was diagnosed 98.2 months (8.2 years) before enrolling in the study, had an IDH2 mutation at baseline, had received 4 previous treatments (including venetoclax), and had undergone a previous HSCT. This patient was treated for 17.6 weeks and achieved a CRh response. Treatment continued for 2 cycles after CRh was achieved, at which point the patient relapsed and withdrew consent from the study.

Among 64 adult efficacy-evaluable patients who achieved an overall response (n = 30), 5 (16.7%) underwent an allogeneic HSCT while in remission. Furthermore, 3 of the 5 HSCT recipients resumed revumenib after HSCT. The duration of maintenance therapy with posttransplant revumenib ranged from 2 to 60 weeks, with no patients remaining on revumenib post-HSCT at the time of the data cutoff (discontinued due to AE [fatigue], progressive disease, or other reason [relapsed disease]; n = 1 each; Figure 2). No instances of differentiation syndrome, grade ≥2 QTcF prolongation, or grade 5 treatment-emergent AEs were observed among the patients who resumed revumenib post-HSCT.

Transcriptional changes were evaluated in 18 adult efficacy-evaluable patients with available RNA at baseline and after 1 cycle of revumenib treatment. After 1 cycle of revumenib, downregulation of most leukemogenic target genes (MEIS1, PBX3, and FLT3) was observed. HOXA9 expression was upregulated in all 5 nonresponders and downregulated in 10 of 13 responders. Expression of genes associated with differentiation (CD11b, CD14, and CD13) was markedly increased regardless of response (supplemental Figure 2).

Comutations were evaluated in the 54 adult efficacy-evaluable patients whose NPM1 mutation status was centrally confirmed by Flagship Biosciences on the Focus Myeloid NGS panel. In the NGS comutation vs response analysis, when responders were compared with nonresponders, no mutations were significantly associated with response or lack of response. However, IDH1 mutation was significantly associated with CR + CRh (P = .0084) or CRc (P = .0013) response vs nonresponse, and STAG2 mutation was significantly associated with nonresponse when compared with patients who achieved a CRc response (P = .049; supplemental Figure 3).

No currently approved therapies specifically target the NPM1 mutation in AML, in either the frontline or relapsed setting. Patients with NPM1m AML who relapse or are refractory to initial therapies have a poor prognosis.^12,17,30 Historical data suggest that only 48% and 10% of patients achieve CR when receiving high- or low-intensity treatments, respectively, as first salvage therapy, with CR rates decreasing with each subsequent line (second salvage CR, 30% and 8%; subsequent salvage CR, 11% and 2%, respectively).¹⁷ These dismal outcomes highlight the urgent need for improved therapies, especially for patients unable to tolerate intensive chemotherapy and/or patients whose disease relapses after treatment with venetoclax. Novel therapies, such as those directed at menin, including revumenib, the first US Food and Drug Administration–approved menin inhibitor, ziftomenib^31,32 and bleximenib,³³ provide a promising approach to targeting leukemogenesis driven by the KMT2A–menin interaction.^26,32,34,35 

Patients with R/R NPM1m AML who enrolled in the phase 2 part of AUGMENT-101 had high-risk baseline characteristics. In the safety population, the median age was 63 years (range, 11-84), and the median number of previous lines of therapy was 2 (range, 1-7), with 34.5% of patients having received ≥3 previous lines of therapy and 19.0% having received ≥4. Importantly, 73.8% of treated patients had previously received a venetoclax-containing treatment regimen, 38.1% had received previously an FLT3 inhibitor, and 23.8% had undergone HSCT previously, with 9.5% having received >1 previous HSCT. At study entry, 57.1% of patients had disease refractory to the most recent line of therapy. This clinical trial patient population is representative of the real-world population of patients with R/R NPM1m AML for whom standard-of-care therapies failed and/or who relapsed after HSCT, both of which are characteristics that confer poor outcomes independent of the presence of an NPM1 mutation.

The phase 2 portion of AUGMENT-101 met the primary end points in patients with R/R NPM1m AML. The CR + CRh rate achieved with revumenib in the adult efficacy-evaluable population was 23.4% (95% CI, 13.8-35.7; 1-sided P = .0014). Almost half of patients achieved a response (ORR, 46.9%), which allowed a subset of patients to proceed to HSCT. The 1 pediatric patient treated achieved CRh. Efficacy of revumenib monotherapy was observed across various subgroups, including by age, prior lines of therapy, previous venetoclax, and previous FLT3 inhibitor exposure. Notably, the CR + CRh rates in patients receiving previous venetoclax or a FLT3 inhibitor were 16.7% and 13.3%, respectively. Although these CR + CRh rates are lower than the 23.4% observed in the overall population with R/R NPM1m AML in this study, they were numerically greater than historical CR rates with salvage therapies after failure of venetoclax or a FLT3 inhibitor (4.2% and 6%, respectively).^18,19 The median time to first CR + CRh was 2.8 months (range, 1.8-8.8), and the median duration of response was an additional 4.7 months (95% CI, 1.2-8.2) thereafter. These clinical data confirm the oncogenic role between menin and KMT2A in NPM1m AML and demonstrate that disruption of this interaction with revumenib, an orally administered targeted inhibitor, provides meaningful antileukemic activity.

The safety profile of revumenib in R/R NPM1m AML was consistent with that found in other acute leukemias and was predictable, with AEs primarily related to the underlying disease, mechanism of action based on preclinical characterization of revumenib, and characteristics of this population (older aged and heavily pretreated). QTc prolongation and differentiation syndrome were known possible AEs with revumenib, and both were manageable. Differentiation syndrome was managed using steroids and hydroxyurea when necessary and seems to be a class effect of menin inhibition.³⁶ 

Gene expression analysis revealed that HOXA9 expression increased in all nonresponders but decreased in most responders in the first cycle of revumenib treatment. This observation suggests that changes in HOXA9 expression may be an early biomarker of revumenib response in NPM1m AML.³⁷ The comutation analysis revealed that several genes were mutated exclusively or predominantly in responders or nonresponders; however, none reached statistical significance when responders were compared with nonresponders. Notably, IDH1 was significantly associated with both CR + CRh and CRc responses vs nonresponse; STAG2 expression was significantly associated with nonresponse vs CRc response. These results should be interpreted with caution as bulk gene expression analysis is not able to distinguish between changes in cell composition and changes in gene expression in equivalent cells. Further studies are warranted to assess these possible relationships.

This single-agent study of revumenib has limitations that should be noted. Comutations were assessed locally at the discretion of the investigator, which may have resulted in inconsistencies due to variability in reporting. In addition, assessments of genetic markers of revumenib resistance have not yet been performed. Last, the nature of single-arm clinical trials has inherent limitations compared with studies with control arms.

These results build upon the first evidence from AUGMENT-101 which demonstrated that a targeted treatment could benefit patients with R/R NPM1m AML.²⁵ Treatment with revumenib, a selective, first-in-class menin inhibitor, continued to provide a meaningful clinical benefit and manageable safety profile for patients with R/R NPM1m AML. Additional studies assessing revumenib in combination for R/R NPM1m AML (SAVE [NCT05360160]) and in the frontline newly diagnosed AML setting for fit (SNDX-5613-0708 [NCT06226571]) and unfit (Beat AML [NCT03013998]; EVOLVE-2 [NCT06652438]) patients are ongoing.^38-41 In conclusion, these results suggest that revumenib has the potential to provide substantial improvement over currently available treatments in patients with R/R NPM1m AML.

The authors thank the study participants, their families, and all investigational site members involved in this study. The authors also thank Yu Gu, Sharda Jha, and Timothy O’Toole, employees of Syndax Pharmaceuticals, Inc, for their significant contributions to the statistical planning and analysis, data collection, and study operations, respectively. Study funded by Syndax Pharmaceuticals, Inc. Writing and editorial support were provided under the direction of the authors by JoAnna Anderson and Apurva Davé of Nucleus Global, an Inizio company, and funded by Syndax Pharmaceuticals, Inc.

Contribution: G.C.I., M.J.T., J.F.D., R.M.S., E.M.S., J.S.B., A.R.S., L.Y., and R.G.B. contributed to the conception, design, or planning of the study; M.L.A., G.C.I., E.M.S., R.M.S., T.K., E.T., J.S.B., A.R.S., L.Y., Y.C., and R.G.B. contributed to data analysis and interpretation; M.L.A., A.C.S., A.Ž., C.S.G., C.P., A.E.P., G.C.I., A.B., D.S.D., M.W.M.K., I.M., H.D., Y.C., L.Y., R.G.B., and A.R.S. contributed to drafting the manuscript; and all authors contributed to the critical review of the manuscript, contributed to the provision of study materials or data acquisition, approved the final manuscript for submission, and agree to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

Conflict-of-interest disclosure: M.L.A. reports advisory board participation for Syndax Pharmaceuticals, Inc. M.J.T. reports institutional funding from Syndax Pharmaceuticals, Inc, AbbVie, Merck, and Nurix; and honoraria from AbbVie. J.F.D. reports funding from MacroGenics, Wugen, BioLineRx, and Incyte; royalties/licenses from Magenta Therapeutics and Wugen; and consulting fees from RiverVest, Vertex Pharmaceuticals, and BioLineRx. E.M.S. reports consulting fees from AbbVie, Agios, Astellas Pharma, AstraZeneca, Celgene, Daiichi Sankyo, Genentech, Gilead, Jazz Pharmaceuticals, and Servier. A.C.S. reports research funding from AbbVie, Amgen, Astellas Pharma, Bristol Myers Squibb (BMS), GlycoMimetics, Kite/Gilead, Loxo, Novartis, Pfizer, Servier, and Syndax Pharmaceuticals, Inc; and advisory board participation for AbbVie, Amgen, Astellas Pharma, BMS, Jazz Pharmaceuticals, Kite/Gilead, Novartis, Paladin, Pfizer, Servier, and Teva. A.Ž. reports consultancy fees from AbbVie, Astellas Pharma, Pfizer, Novartis, and Johnson & Johnson; honoraria from AbbVie, Astellas Pharma, Novartis, and Johnson & Johnson; and travel expenses from AbbVie, Novartis, Johnson & Johnson, and Takeda. S.d.B. reports research funding from Auron; consultancy for Servier, BMS, GlaxoSmithKline (GSK), Syndax Pharmaceuticals, Inc, Remix, Auron, and Rigel; honoraria from BMS, AbbVie, Servier, Jazz Pharmaceuticals, Astellas Pharma, and Loxo; and travel fees from AbbVie, Servier, Johnson & Johnson, Daiichi Sankyo, and Rigel. C.S.G. reports consultancy fees from Otsuka Australia Pharmaceutical, Astellas Pharma, and AbbVie. G.N.M. reports consultancy fees from AbbVie, Agios, MacroGenics, Pfizer, Servier, and Stemline; scientific advisory committee participation for AbbVie, Agios, Astellas Pharma, BMS/Celgene, Forty Seven, Genentech, ImmunoGen, Orbital Therapeutics, Rigel, Servier, Stemline, and Wugen; and research funding from Aptose, GlycoMimetics, Forty Seven, Gilead, Jazz Pharmaceuticals, Astex, Syndax Pharmaceuticals, Inc, Immune-Onc, and BMS/Celgene. C.P. reports honoraria from AbbVie, Astellas Pharma, Servier, Menarini/Stemline, BMS, Pfizer, Amgen, Janssen, Incyte, and Novartis; and advisory board participation for Pfizer, Astellas Pharma, Janssen, GSK, Blueprint, Jazz Pharmaceuticals, AbbVie, Novartis, and Laboratoires Delbert. A.E.P. reports research funding from Syndax Pharmaceuticals, Inc; grants from Daiichi Sankyo, Astellas Pharma, AbbVie, and Syndax Pharmaceuticals, Inc; advisory board participation for AbbVie, Astellas Pharma, Daiichi Sankyo, Genentech, ImmunoGen, BMS, Aptose, Rigel, Curis, and Schrödinger; consultancy fees from AbbVie, Astellas Pharma, Daiichi Sankyo, Foghorn Therapeutics, and Beat AML; and honoraria from Astellas Pharma and Daiichi Sankyo. G.C.I. reports research funding from Celgene, Merck, Kura Oncology, Syndax Pharmaceuticals, Inc, Astex, NuProbe, and Novartis; and consultancy or advisory board fees from NuProbe, AbbVie, Novartis, Sanofi, AstraZeneca, Syndax Pharmaceuticals, Inc, and Kura Oncology. I. Aldoss reports consultancy fees from Amgen, Syndax Pharmaceuticals, Inc, Pfizer, Kite Pharma, Takeda, Jazz Pharmaceuticals, and Sobi; honoraria from Pfizer and Amgen; and research support from AbbVie and MacroGenics. A.B. reports advisory board participation for AbbVie, Amgen, Novartis, Pfizer, Takeda, Astellas Pharma, and Senti Bio; speaker fees from Amgen, BMS, and Pfizer; and consultancy fees from Shoreline Biosciences. D.S.D. reports advisory board participation for Tempus and Amgen; unpaid advocacy roles for the American Academy of Pediatrics, American Society of Pediatric Hematology/Oncology, and Iowa Cancer Consortium; and consultancy fees from Amgen and Y-mAbs Therapeutics. M.W.M.K. reports honoraria and consultancy fees from AbbVie, BMS/Celgene, Gilead, Johnson & Johnson, Jazz Pharmaceuticals, Pfizer, and Servier; and research support from Kura Oncology and Syndax Pharmaceuticals, Inc. I.M. reports advisory board participation with Syndax Pharmaceuticals, Inc. E.T. reports research funding from Prelude Therapeutics, Schrödinger, Incyte Corporation, and AstraZeneca; consultancy fees from Astellas Pharma and AbbVie; and advisory board participation for AbbVie, Astellas Pharma, Daiichi Sankyo, Servier Laboratories, and Rigel. I. Amitai reports honoraria from Novartis and AbbVie; and advisory board participation for Stemline and Alexion. H.D. reports advisory board participation with honoraria for AbbVie, AstraZeneca, Gilead, Janssen, Jazz Pharmaceuticals, Pfizer, Servier, Stemline, and Syndax Pharmaceuticals, Inc; institutional research support from AbbVie, Astellas Pharma, BMS, Celgene, Jazz Pharmaceuticals, Kronos Bio, and Servier; and travel expenses from AbbVie and Servier. C.G. reports honoraria from Jazz Pharmaceuticals. T.K. reports institutional research support from AbbVie, Amgen, Gilead, GlycoMimetics, and Syndax Pharmaceuticals, Inc; honoraria from Rigel and Servier; and advisory board participation for Rigel, Astellas Pharma, Takeda, and NeoGenomics. C.M.M. reports research funding from Syndax Pharmaceuticals, Inc, and Syros Pharmaceuticals, Inc; and advisory board participation for Kura Oncology. P.M. reports advisory board participation for AbbVie, Astellas Pharma, Daiichi Sankyo, Janssen, Servier, and Syndax Pharmaceuticals, Inc; speakers bureau participation for AbbVie, Astellas Pharma, Daiichi Sankyo, Janssen, and Servier; research support from AbbVie, Astellas Pharma, Daiichi Sankyo, Janssen, and Servier; and consultancy fees from Astellas Pharma and Daiichi Sankyo. P.J.S. reports research support from Chimerix, Inc, Amgen, and Abcuro, Inc; advisory board participation for BMS, Daiichi Sankyo, Gilead Sciences Takeda, and RJH BioSciences; serving as chief medical officer and on the board of directors for JSK Therapeutics; serving in a leadership role for the National Comprehensive Cancer Network; and holding a US patent (8,404,665; 9,005,656). R.M.S. reports consultancy fees from AbbVie, Actinium Pharmaceuticals, Inc, Amgen, Aptevo Therapeutics, Arog Pharmaceuticals, AvenCell, BerGenBio, BMS, Boston Scientific, Cellularity, CTI BioPharma, DAVA Oncology, Epizyme, Inc, GSK, Hemavant Sciences, Janssen, Jazz Pharmaceuticals, Kura Oncology, LAVA Therapeutics, Ligand Pharmaceuticals, Novartis, Rigel, Syros Pharmaceuticals, Syntrix Pharmaceuticals, and Takeda; and advisory board participation for Aptevo Therapeutics, Syntrix Pharmaceuticals, and Takeda. O.W. reports research support from AbbVie, honoraria from Astellas Pharma and Novartis; and serving in an advisory role for AbbVie, Astellas Pharma, Novartis, Pfizer, Medison, and Teva. J.G.H. reports employment with ICON plc, which received research funding from Syndax Pharmaceuticals, Inc. Y.C., L.Y., R.G.B., and A.R.S. report employment with Syndax Pharmaceuticals, Inc and stock ownership in the company. J.S.B. reports consultancy fees from Astellas Pharma, AstraZeneca, AbbVie, and Syndax Pharmaceuticals, Inc; advisory board participation for Astellas Pharma, AstraZeneca, AbbVie, and Syndax Pharmaceuticals, Inc; honoraria from the 2023 Ohio Academy of Family Physicians; and holding US patents on a leukemia diagnostic device (in prosecution) and for the epigenetic classification of leukemia (Patent Cooperation Treaty filed). The remaining authors declare no competing financial interests.

Correspondence: Martha L. Arellano, Hematology and Medical Oncology Fellowship Program, Winship Cancer Institute of Emory University School of Medicine, 1365 Clifton Rd, Suite B4129, Atlanta, GA 30322; email: marella@emory.edu.