The menin inhibitor revumenib in KMT2A-rearranged or NPM1-mutant leukaemia

Targeting critical epigenetic regulators reverses aberrant transcription in cancer, thereby restoring normal tissue function1–3. The interaction of menin with lysine methyltransferase 2A (KMT2A), an epigenetic regulator, is a dependence in acute leukaemia caused by either rearrangement of KMT2A or mutation of the nucleophosmin 1 gene (NPM1)4–6. KMT2A rearrangements occur in up to 10% of acute leukaemias and have an adverse prognosis, whereas NPM1 mutations occur in up to 30%, forming the most common genetic alteration in acute myeloid leukaemia7,8. Here, we describe the results of the first-in-human phase 1 clinical trial investigating revumenib (SNDX-5613), a potent and selective oral inhibitor of the menin–KMT2A interaction, in patients with relapsed or refractory acute leukaemia (ClinicalTrials.gov, NCT04065399). We show that therapy with revumenib was associated with a low frequency of grade 3 or higher treatment-related adverse events and a 30% rate of complete remission or complete remission with partial haematologic recovery (CR/CRh) in the efficacy analysis population. Asymptomatic prolongation of the QT interval on electrocardiography was identified as the only dose-limiting toxicity. Remissions occurred in leukaemias refractory to multiple previous lines of therapy. We demonstrate clearance of residual disease using sensitive clinical assays and identify hallmarks of differentiation into normal haematopoietic cells, including differentiation syndrome. These data establish menin inhibition as a therapeutic strategy for susceptible acute leukaemia subtypes.

Targeting critical epigenetic regulators reverses aberrant transcription in cancer, thereby restoring normal tissue function [1][2][3] . The interaction of menin with lysine methyltransferase 2A (KMT2A), an epigenetic regulator, is a dependence in acute leukaemia caused by either rearrangement of KMT2A or mutation of the nucleophosmin 1 gene (NPM1) [4][5][6] . KMT2A rearrangements occur in up to 10% of acute leukaemias and have an adverse prognosis, whereas NPM1 mutations occur in up to 30%, forming the most common genetic alteration in acute myeloid leukaemia 7,8 . Here, we describe the results of the first-in-human phase 1 clinical trial investigating revumenib (SNDX-5613), a potent and selective oral inhibitor of the menin-KMT2A interaction, in patients with relapsed or refractory acute leukaemia (ClinicalTrials. gov, NCT04065399). We show that therapy with revumenib was associated with a low frequency of grade 3 or higher treatment-related adverse events and a 30% rate of complete remission or complete remission with partial haematologic recovery (CR/CRh) in the efficacy analysis population. Asymptomatic prolongation of the QT interval on electrocardiography was identified as the only dose-limiting toxicity. Remissions occurred in leukaemias refractory to multiple previous lines of therapy. We demonstrate clearance of residual disease using sensitive clinical assays and identify hallmarks of differentiation into normal haematopoietic cells, including differentiation syndrome. These data establish menin inhibition as a therapeutic strategy for susceptible acute leukaemia subtypes.
The prognosis of acute leukaemias harbouring rearrangements of the gene lysine methyltransferase 2A (KMT2A), previously known as mixed-lineage leukaemia (MLL), is poor, with a 5-year overall survival of less than 25% (ref. 7 ). KMT2A rearrangements (KMT2Ar) occur in 80% of infant acute lymphoblastic leukaemia (ALL) and in 5-15% of children and adults with acute leukaemia, whether myeloid, lymphoid or mixed phenotype 9 . Mutated nucleophosmin 1 gene (NPM1) is the most common genetic alteration in adult acute myeloid leukaemia (AML), occurring in up to 30% of patients 8 . Currently there are no targeted therapies specifically approved for acute leukaemia with KMT2Ar or mutated NPM1.
Genetic rearrangements of KMT2A lead to aberrant expression of homeobox (HOX) genes and their DNA-binding cofactor MEIS1 (ref. 10 ). This gene expression programme, normally expressed in stem cells, causes a haematopoietic differentiation block and leukaemic transformation 11 . For leukaemias driven by KMT2Ar, menin is a critical oncogenic cofactor 4 . The menin-binding motif is preserved throughout all KMT2A fusion proteins, and menin is an essential cofactor for binding to HOX gene promoters 7 . Similarly, in AML with mutated NPM1, the interaction between wild-type KMT2A and menin leads to HOX-and MEIS1-mediated leukaemogenic transcription 6,12 . Blockade of the menin-KMT2A interaction disrupts the assembly of oncogenic KMT2A wild-type or fusion complexes on chromatin 12 . Preclinical studies showed that menin inhibition downregulates HOX and MEIS1 transcription and reverses leukaemogenesis in KMT2Ar-r-or NPM1-mutated leukaemia models 5,12,13 . Revumenib, previously known as SNDX-5613, is a potent, oral, selective inhibitor of the menin-KMT2A interaction. Treatment with revumenib led to abrogation of aberrant HOX gene expression and dramatic antileukaemic activity in those models (Extended Data Fig. 1; GEO accession no. for RNA sequencing (RNA-seq) data, GSE216730). In this first-in-human clinical trial we assessed the safety, maximum tolerated dose (MTD), recommended phase 2 dose (RP2D) and pharmacokinetic and pharmacodynamic profiles in patients with relapsed or refractory acute leukaemia, and present the clinical activity of revumenib in patients with KMT2Ar or mutated NPM1.
This phase 1 dose-escalation study was conducted across nine sites in the United States. Because revumenib is a substrate of cytochrome P450 3A4 (CYP3A4), two parallel dose-escalation cohorts, one without https://doi.org/10.1038/s41586-023-05812-3 (Arm A) and one with (Arm B) strong CYP3A4 inhibitors, were conducted. Revumenib was administered orally every 12 h (q12h) in continuous 28-day cycles. Although all patients could enrol at study initiation regardless of cytogenetic and mutational profile, given the strong preclinical rationale demonstrating that the menin-KMT2A interaction is a targetable vulnerability in acute leukaemia with KMT2Ar or mutated NPM1, an early amendment to the protocol restricted enrolment to patients with relapsed or refractory acute leukaemia with KMT2Ar or mutated NPM1 (88% of the total study population). Between 5 November 2019 and 31 March 2022, a total of 68 patients (37 in Arm A and 31 in Arm B) were enroled and treated. The data cutoff date for this analysis was 31 March 2022. Patient disposition is presented in Extended Data Fig. 2.
Baseline characteristics of patients are provided in Extended Data Table 1. There were 56 patients (82%) with relapsed or refractory AML, 11 (16%) with ALL and one with mixed-phenotype acute leukaemia (2%). Forty-six patients (68%) had KMT2Ar, 14 (21%) had mutated NPM1 and eight (12%) had neither KMT2Ar nor NPM1 mutations. Sixty patients were adults (at least 18 years old) and eight were children or adolescents (under 18 years of age). The median age was 42.5 years (range, 0.8-79). The median age of adult patients was 50.5 years (range, 19-79) and the median age among paediatric patients was 2.5 years (range, 0.8-16). The study population was heavily pretreated with a median of four previous lines of therapy (range, 1-12), and 31 patients (46%) had relapsed after an allogenic stem cell transplant. All patients who received at least one dose of revumenib were included in the safety analysis whereas only patients with KMT2Ar or mutated NPM1 were included in the efficacy analysis.
The dose-escalation schema is shown in Extended Data Fig. 3. The only dose-limiting toxicity observed in both Arm A and Arm B was grade 3 prolongation of the QT interval (over 500 ms) on electrocardiography (ECG), a toxicity predicted in preclinical animal studies. These events occurred without any clinical correlate or symptoms. This dose-limiting toxicity was observed in Arm A at dose levels of 226 mg q12h (n = 1) and 339 mg q12h (n = 2), and in Arm B at 113 mg q12h (n = 1) and 226 mg q12h (n = 2). The incidence of grade 3 QT prolongation was 10% at dose levels either at or below MTD.
A common occurrence in antileukaemia therapies that induce myeloid differentiation is the development of a clinically apparent differentiation syndrome. Differentiation syndrome is caused by cytokine alterations associated with haematopoietic differentiation, with common manifestations including fever, arthralgias, leukocytosis, pleural or pericardial effusions and respiratory or renal failure in severe cases 14,15 . This had been mostly reported in patients with myeloid leukaemias treated with all-trans retinoic acid, arsenic trioxide and isocitrate dehydrogenase inhibitors [14][15][16] . In this study, differentiation syndrome was reported in 11 patients (16%), all grade 2. All cases of differentiation syndrome resolved following treatment with corticosteroids, with the addition of hydroxyurea for associated leukocytosis seen in five of the 11 patients (Extended Data Figs. 4 and 5). One patient missed one day of dosing, but otherwise treatment with revumenib was uninterrupted because of differentiation syndrome. The median time of onset of differentiation syndrome was 18 days (range, 5-41) with manifestations including bone pain or arthralgia, pericardial effusion, pleural effusion, pulmonary infiltrates, increase in creatinine, rash, oedema and pyrexia.
Pharmacokinetic studies showed that dose-proportional exposure was achieved in both Arm A (no strong CYP3A4 inhibitor) and Arm B (with a strong CYP3A4 inhibitor). Steady-state levels were achieved in approximately 48 h, with no evidence of drug accumulation (Extended Data Fig. 6).
Following a review of prespecified protocol criteria on safety, tolerability and pharmacokinetic data, the revumenib doses of 226 mg q12h and 276 mg q12h in Arm A and 113 mg q12h and 163 mg q12h in Arm B met the prespecified criteria for RP2D.
The pharmacodynamic effects of revumenib were assessed through transcriptional changes following treatment using RNA-seq of bone marrow cells (Fig. 1). Consistent with the established mechanism of action, menin inhibition using revumenib resulted in downregulation of the critical leukaemogenic target genes MEIS1, homeobox A9 (HOXA9), pre-B-cell leukaemia transcription factor 3 (PBX3) and cyclin-dependent kinase 6 (CDK6), and an increase in expression of genes associated with differentiation, such as integrin alpha M (CD11b)      and CD14. We also observed transcriptional suppression of fms-like tyrosine kinase 3 (FLT3) following treatment, a putative transcriptional target of MEIS1; FLT3 is frequently mutated in multiple subtypes of AML, including KMT2Ar AML (10%), and is particularly common in AML with mutated NPM1 (60%) 8,[17][18][19] . The rate of complete remission or complete remission with partial haematologic recovery (CR/CRh) was 30% (95% confidence interval (CI) 18.8-43.2) in 18 of 60 evaluable patients with undetectable measurable residual disease (MRD), as assessed by multiparameter flow cytometry, in 14 of 18 patients (78%) who achieved CR/CRh (Fig. 2a and Table 2). The overall response rate (CR/CRh/complete remission with incomplete platelet recovery (CRp)/complete remission with incomplete haematologic recovery (CRi)/morphologic leukaemia-free state) was 53% (32 of 60 evaluable patients). The median time to CR/ CRh was 1.9 months (range, 0.9-4.9). Although imaging assessment was not mandated on study, responses were notably seen in both bone marrow and extramedullary sites in two of six evaluable patients with extramedullary leukaemia at enrolment, with representative imaging shown in Extended Data Fig. 7. In an exploratory descriptive analysis, we evaluated responses by both leukaemia lineage and the adult and paediatric (under 18 years of age) populations. Morphologic remissions were identified in 27 of 49 patients (55%) with AML (95% CI 40.2-69.3), in four of ten patients (40%) with ALL (95% CI 12.2-73.8) and in the one patient with a mixed-phenotype acute leukaemia. Morphologic remission was noted in four of eight paediatric patients (50%; 95% CI 15.7-84.3) and in 28 of 52 adults (54%; 95% CI 39.5-67.8).
In patients with KMT2Ar who achieved morphologic remission following the first cycle of therapy with revumenib, 21 of 25 (84%) retained the detectable fusions causing KMT2Ar during concomitant cytogenetic analyses. However, most patients later had clearance of these KMT2A rearrangements with subsequent cycles of therapy. The rate of complete cytogenetic response in patients with KMT2Ar who achieved morphologic clearance of their myeloblasts was 16 of 25 (64%) (Extended Data Table 3); these typically occurred after morphologic remission and, in some cases, after MRD negativity as assessed by immunophenotyping via flow cytometry. The median time to achieving complete cytogenetic response was 1.9 months (range, 0.9-2.8). These findings are consistent with the mechanism of action of differentiation agents, for which response leads first to a phenotypic differentiation of cells such that they retain the genetic fusion but are without aberrant morphology or leukaemia surface marker expression (as detected by flow cytometry) 20,21 . Subsequently with additional cycles following phenotypic remission, fusions are no longer detected by fluorescence in situ hybridization, probably because of apoptosis of differentiated cells or replacement by normal haematopoiesis.
With a median follow-up of 11.9 months (95% CI 4.9-14.8) in patients who achieved CR/CRh, the median duration of response was 9.1 months (95% CI 2.7 to not reached (NR)) (Fig. 2b). Median overall survival in the efficacy population, regardless of remission status, was 7 months (95% CI 4.3-11.6) (Extended Data Fig. 8) and median follow-up of overall survival in the efficacy population was 14.3 months (95% CI 10.6-16.7). Twelve patients received an allogeneic stem cell transplant as consolidation following response to revumenib, with nine in remission at the time of the data cutoff, seven of whom have been in remission for over 6 months. The median duration of response in those who achieved CR/ CRh with censoring at the time of allogeneic stem cell transplant was 4.4 months (95% CI 1.9-NR) (Extended Data Fig. 9).
Consistent with the preclinical hypothesis regarding the efficacy of menin inhibition in patients with NPM1 mutations or KMT2Ar, there were no responses in the eight patients enroled on study who lacked either a KMT2Ar or NPM1 mutation. In evaluable patients for whom a targeted sequencing panel was performed at baseline, no specific KMT2A translocation or comutation was clearly associated with response in a limited analysis (Extended Data Fig. 10). Notably among evaluable patients with KMT2Ar or mutated NPM1, 14 had a comutation in FLT3, of whom 13 had been treated with a FLT3 inhibitor before enrolment. Among the 11 patients with comutated NPM1 and FLT3, three responded whereas two of the three patients with KMT2Ar and concomitant FLT3 mutations achieved a response. All responders had previously received a FLT3 inhibitor.

Discussion
The treatment of patients who have relapsed or refractory acute leukaemia with KMT2Ar is challenging. The rate of CR or complete remission with incomplete count recovery in relapsed KMT2Ar AML following two previous lines of therapy is 9% (ref. 18 ). Despite notable progress in the treatment of childhood acute leukaemia, infant KMT2Ar acute leukaemias have remained a therapeutic challenge with high rates of resistance to multi-agent chemotherapy 22,23 . Mutations in NPM1 are the most common alterations in AML, and targeting of vulnerabilities associated with this subset expand the reach of targeted therapy to the largest portion of patients with this disease. The menin inhibitor revumenib has the potential to address these unmet needs. In this first-in-human clinical trial, we provide clinical evidence of the effectiveness of menin inhibition with an oral targeted therapy, which is the first epigenetic therapy that evicts protein complexes from chromatin, leading to remissions in patients with acute leukaemia. We found an encouraging clinical benefit, with deep molecular remissions and minimal toxicities, in a heavily pretreated population of both children and adults with advanced acute leukaemia. However, recent data suggest that clinical resistance to targeted menin inhibition is mediated through the acquisition of mutations in menin that prevent inhibitor binding and can lead to clinical relapse 24 . The emergence of resistance  Article mutations highlights the critical role of the menin-KMT2A interaction in the pathogenesis of KMT2Ar and NPM1-mutant acute leukaemia. Because revumenib disrupts the effect of epigenetic regulators in leukaemia that are dependent on the menin-KMT2A interaction, this leads to abrogation of aberrant gene expression and removal of the haematopoietic differentiation block 12,13 . In this study, we show evidence of this mechanism of action with differentiation of leukaemia cells, persistence of cytogenetic abnormalities at the time of morphologic response and differentiation syndrome associated with menin inhibition in some patients, which resolved following appropriate therapy.
Other leukaemia genotypes have recently been identified as susceptible to menin inhibition, such as AML with rearrangement of the nucleoporin 98 gene (NUP98), a common and adverse genotype among children with relapsed and refractory disease 25 . Another example is the rare occurrence of AML with translocations involving the meningioma-1 gene (MN1) 26 . Therefore, menin inhibitors have the potential to reach a larger subset of acute leukaemia with similar dependency on the menin-KMT2A interaction, which could be identified using precision approaches testing characteristic gene expression.
In conclusion, in children and adults with highly refractory acute leukaemia with KMT2Ar or NPM1 mutation, menin inhibition with revumenib monotherapy was associated with promising antileukaemic activity leading to deep and sustained remission.

Online content
Any methods, additional references, Nature Portfolio reporting summaries, source data, extended data, supplementary information, acknowledgements, peer review information; details of author contributions and competing interests; and statements of data and code availability are available at https://doi.org/10.1038/s41586-023-05812-3.

Study design and oversights
We conducted a phase 1, multicentre, open-label, dose-escalation study (ClinicalTrials.gov, NCT04065399). Revumenib was administered orally in either capsule or liquid formulation, q12h, in 28-day continuous cycles with dose adjustment by body surface area for patients weighing under 40 kg (Extended Data Fig. 3). We employed a rolling-6 design, an algorithm-based extension of the 3+3 design that allows for concurrent accrual of two to six patients onto a dose level, thereby shortening the duration of phase 1 without increasing the risk of toxicity 27 . Therefore, the number of patients enrolled followed this dose-escalation design without an absolute prior estimation of sample size. The study included two parallel dose-escalation cohorts, for patients either not taking (Arm A) or taking (Arm B) strong CYP3A4 inhibitors. Dose expansion occurred at safe and efficacious doses. Dose-limiting toxic effects were defined as nonhaematologic toxic effects of grade 3 or higher during cycle 1, or as haematologic toxicities directly attributed to the study drug rather than to the underlying disease.
The study was conducted in accordance with the principles of the All patients provided written informed consent. This study was designed by the sponsor (Syndax Pharmaceuticals) in collaboration with the investigators. The data were collected by the investigators and their research staff and were analysed by the sponsor and authors. Drafts of the manuscript were written by the first and last authors and revised in collaboration with all authors and the sponsor, all of whom vouch for the completeness and accuracy of the data and analyses, and for the adherence of the study to the protocol. Assistance in manuscript preparation for submission was provided by a professional service and paid for by the sponsor.

Patients
An early amendment to the protocol allowed patients aged 30 days and older to be enroled on this study; additionally, it restricted the eligibility criteria from any relapsed or refractory acute leukaemia to only those with KMT2Ar or mutated NPM1 in Arms A and B of the study. Mutational status was assessed at each site.

Study assessments
The primary objectives of the phase 1 portion of this study were to assess safety, the MTD or, if different, RP2D and to characterize the pharmacokinetic parameters of revumenib in arms based on concomitant CYP3A inhibitors. Exploratory objectives included assessment of antileukaemic activity in the efficacy population, which consisted of patients with KMT2Ar or mutated NPM1. All study assessments and analyses included both paediatric and adult patients.
Adverse events were graded with use of the Common Terminology Criteria for Adverse Events, v.5.0. Clinical efficacy was assessed by the investigators with use of a modified version of the 2017 European LeukemiaNet response criteria, to additionally include CRh 28 . Guidelines for managing prolongation of the QT interval included electrolyte repletion and adjustment of revumenib dose. Investigators were encouraged to administer corticosteroids following suspicion of differentiation syndrome based on guidelines included in the protocol. Measurable residual disease assessment was performed in participating institutions using multicolour flow cytometry or PCR [29][30][31] . In an exploratory mutational analysis we used the ArcherDx VariantPlex Myeloid Panel, a 75-gene, next-generation targeted sequencing product examining genes frequently mutated in myeloid and lymphoid malignancies. Powered by Anchored Multiple PCR chemistry, the panel enables deep strand-specific amplification of unique molecular barcoded DNA fragments for sequencing. Archer Analysis software was used for the analysis and interpretation of sequencing data in the detection of single-nucleotide variants and indels. Variant calling was performed using Invitae's Comprehensive Targeted Mutation File (v.1.6), and the Vision variant caller to report well-classified variants.

Statistical analysis
All patients who received at least one dose of revumenib were included in the safety analysis, whereas only those with KMT2Ar or mutated NPM1 were included in the efficacy analysis. Time-to-event end points were estimated using the Kaplan-Meier method. Descriptive statistics were used for other clinical, laboratory and pharmacokinetic variables. Clinical data were captured in Medidata Classic Rave 2021.2.0, and data analyses were performed using SAS v.9.4 and GraphPad Prism v.8.0. Changes in gene expression were analysed using a paired t-test.
RP2D was determined by the safety review committee based on review of pharmacokinetics, safety and tolerability data among evaluable patients.

Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Data availability
Syndax Pharmaceuticals (Syndax) is committed to providing qualified scientific researchers access to anonymized data and clinical study reports from the company's clinical trials for the purpose of conducting legitimate scientific research. Syndax is also obligated to protect the rights and privacy of trial participants and, as such, has a procedure in place for evaluating and fulfilling requests for sharing company clinical trial data with qualified external scientific researchers. Following submission of a request (to datarequest@syndax.com), Syndax will provide an outline of the process and requirements for submitting a data request. Feasible requests will be reviewed by a committee of Syndax subject matter experts to assess the scientific validity of the request and the qualifications of the requestors. In line with data privacy legislation, submitters of approved requests must enter into a standard data-sharing agreement with Syndax before Syndax may grant data access. Data will be made available for request after product approval in the United States and European Union, or after product development is discontinued. There are circumstances that may prevent Syndax from sharing requested data, including country-or region-specific regulations. If Syndax declines the request, it will communicate the decision to the investigator. Access to genetic or exploratory biomarker data requires a detailed statistical analysis plan that is collaboratively developed by the requester and Syndax subject matter experts.