Revumenib in colorectal and solid tumors: a menin inhibitor with zero responses

What was tried

Syndax Pharmaceuticals ran a Phase 1/2 study of revumenib (SNDX-5613) in colorectal cancer and other solid tumors (NCT05731947). Revumenib is an oral small-molecule inhibitor of the menin and MLL protein interaction (ChEMBL CHEMBL4650827). The study enrolled 41 participants who had failed at least one prior line of therapy, using a non-randomized, sequential design with dose escalation in Phase 1a, a signal-seeking Phase 1b expansion, and a planned randomized Phase 2 against investigator-choice chemotherapy (NCT05731947). It ran from April 2023 to a primary completion date of June 2025. Syndax then terminated it. The registry reason: "Study was terminated by sponsor after Phase 1b as pre-defined signals of efficacy were not met in the Phase 1b dose expansion cohort" (NCT05731947).

The biological hypothesis

Menin is a scaffold protein that binds MLL and helps assemble transcriptional complexes at HOX and MEIS1 loci. In KMT2A-rearranged and NPM1-mutant acute leukemias, this interaction is a genetic dependency, and revumenib exploits it. The AUGMENT-101 program showed single-agent activity in those leukemias and supported approval in KMT2A-rearranged disease (Issa 2023, Issa 2025). The solid-tumor rationale was different and more speculative. Menin and MLL also shape transcriptional programs downstream of Wnt and other pathways in epithelial tumors, and colorectal cancer carries near-universal Wnt activation. The hope was that menin inhibition would blunt those programs. Human genetics did not strongly support MEN1 as a colorectal driver. The Open Targets overall association score for MEN1 and colorectal carcinoma is 0.0867, most of it from literature co-mention rather than causal genetic evidence.

What actually happened

Revumenib did not produce a single objective response in this population. Overall response rate was 0 percent across every dose cohort, including the Phase 1b expansion at 270 mg (0 of 22), giving a 95 percent Wilson interval of 0 to 14.9 percent for that cohort and 0 to 8.6 percent across all 41 treated participants (NCT05731947 posted results). Disease control rate, which counts stable disease, reached 33.3 percent in one small Phase 1a cohort but fell to 10.0 percent in the Phase 1b expansion. Pharmacokinetics were not the problem. Maximum plasma concentrations rose with dose to roughly 3,000 to 4,000 ng/mL, in the range achieved in the leukemia program, so exposure was adequate. No dose-limiting toxicities were reported in Phase 1a. The drug reached its target and reached its concentrations. It did not shrink tumors.

Failure mechanism, best guess

This reads as a translational mismatch expressed as an efficacy failure. Menin is a validated dependency where a specific fusion or mutation rewires transcription around it, as in KMT2A-rearranged leukemia. Colorectal cancer does not carry that dependency. Wnt activation in colorectal tumors runs mainly through APC loss and beta-catenin stabilization, and menin sits far enough from that node that inhibiting it does not collapse the oncogenic program. The clean pharmacology and the flat response rate point to a target that is present but not load-bearing in this tumor. The archetype here is efficacy failure, driven by a hypothesis that transplanted a leukemia dependency into an epithelial setting where the genetics did not carry it.

How to prevent this next time

Two quantitative tools would have set a lower prior and a faster stop. First, a Bayesian futility rule on objective response. Starting from a Beta(1,1) prior and updating on the observed 0 of 41, the posterior probability that the true response rate exceeds a 20 percent threshold of interest is about 0.0001, and even the interim 0 of 22 would have driven that probability below any reasonable continuation boundary. The posterior takes the form

PP(success)=∫θ​P(final success∣θ,interim data)⋅π(θ∣interim data)dθ

Second, translational validation sequencing before the clinic. A requirement to show a menin dependency in colorectal models by genetic knockdown, with an effect size that separates from control, would have tested whether the target was load-bearing. Open Targets scored MEN1 and colorectal carcinoma at 0.0867, a weak prior that a historical base-rate adjustment on non-genetically-anchored oncology targets would have discounted further.

The single highest leverage change would have been a preclinical menin-dependency gate in colorectal models, with a pre-specified effect size, that the program had to clear before opening a solid-tumor cohort.

What this means for similar programs

Approval in one lineage does not license a mechanism everywhere. Revumenib works where a fusion or mutation creates a menin dependency, and the solid-tumor extension tested the mechanism outside that context. Programs repurposing a validated targeted agent into a new tumor type should treat the original genetic dependency as the asset, not the drug. When the new indication lacks that dependency, adequate exposure and a clean safety profile do not rescue efficacy, as the flat 0 percent response rate here showed.

Open questions

Would a biomarker-defined solid-tumor subset, for example tumors with a specific menin-dependent transcriptional signature, have shown activity that a broad colorectal population masked. Does menin inhibition have combination potential in colorectal cancer, paired with a Wnt-pathway agent, that single-agent testing could not reveal. The posted results report response and disease control but not the molecular profiling that would identify any responder-enriched subgroup.

Sources

1. • ClinicalTrials.gov NCT05731947, protocol and posted results: 41 treated, ORR 0 percent, DCR by cohort, revumenib Cmax, why-stopped text. - Issa GC et al. The menin inhibitor revumenib in KMT2A-rearranged or NPM1-mutant leukaemia. Nature 2023.- Issa GC et al. Menin Inhibition With Revumenib for KMT2A-Rearranged Acute Leukemia. J Clin Oncol 2025.- Martinez-Gamboa DA et al. Revumenib: a new era in acute leukemia treatment. Trends Cancer 2025.- ChEMBL CHEMBL4650827 (revumenib), max phase 4, small molecule. - Open Targets Platform: MEN1 and colorectal carcinoma overall association 0.0867. - Wilson score intervals and the Beta(1,1) posterior computed from posted-result counts. Available from: https://clinicaltrials.gov/study/NCT05731947.