Risk: the class's pharmacodynamic logic measures menin and reports it as the fusion, in a drug class dosed in infants
also called PD endpoint risk · menin displacement as a surrogate · MEIS1 as a biomarker
The tests used to show these drugs are working measure menin coming off DNA, or one gene switching off. Neither shows what happened to the leukaemia's fusion protein. In a trial in babies, where you can take very few samples, measuring the wrong thing is expensive.
The same thing, for a scientist
Three surrogate readouts carry the mechanistic case for this class, and each has a specific failure mode that the findings in this beat make concrete. **(1) Menin chromatin displacement.** Near-complete at 100 nM revumenib [E1]. It therefore **saturates at the bottom of the clinical dose range and cannot grade depth of response**. Worse, it is reported from depth-normalised data, in which a global change is partly rescaled away, so the true extent is uncertain in the flattering direction. And it says nothing about the fusion: in the same drug state the fusion is not globally displaced [E2]. **(2) MEIS1 expression.** The standard pharmacodynamic biomarker, used as such from the earliest preclinical work [E3]. MEIS1 is one of the three loci where the fusion genuinely does leave [E2] - so it is **the least representative locus in the genome** for inferring what happened elsewhere. A full MEIS1 response is compatible with the fusion remaining on 99% of its sites. **(3) Differentiation markers and blast clearance.** Real and clinically meaningful, and not in question. They are outcome measures, not target-engagement measures, and cannot be used to argue that the oncoprotein was removed. **Why the stakes are paediatric.** Revumenib is approved for relapsed or refractory KMT2A-rearranged acute leukaemia including children, and the class is being taken into infants, where marrow volume strictly limits correlative sampling. **A correlative plan that spends the single available aspirate on menin ChIP answers a question that was already answered in 2019 and forgoes the one measurement nobody has: what the fusion did.** The alternative costs no extra sample - profile the fusion in the same CUT&RUN reaction, add a spike-in, and analyse for gains as well as losses.
What to measure instead, at no extra sample cost
| current | proposed | extra material needed |
|---|---|---|
| menin CUT&RUN | menin and MLL1n (or tag) in the same reaction | none - a second antibody aliquot |
| depth normalisation | spike-in calibration | none - spike chromatin into the same tube |
| loss analysis at target genes | genome-wide differential, both directions | none - analysis only |
| MEIS1 qPCR | MEIS1 plus a retained-locus control (MYB, HOXA9) | none |
| steady-state RNA-seq | add nascent transcription if cell number allows | more cells - the one item that costs |
Four of the five cost nothing but a protocol amendment, and together they would convert a saturating engagement assay into a graded one.
| kind | delivery |
|---|---|
| severity | moderate - affects interpretation and trial design rather than patient safety directly |
| mitigation | Profile the fusion alongside menin in the same correlative reaction; add a chromatin spike-in for calibration; analyse genome-wide in both directions; pair MEIS1 with a retained-locus control such as MYB or HOXA9 |
| shown for fusion | no - none of the three surrogates measures the fate of a KMT2A fusion protein |
affects
[ "drug-revumenib", "drug-ziftomenib", "drug-bleximenib" ]
surrogates in use
[ "menin chromatin displacement", "MEIS1 expression", "differentiation markers and blast clearance" ]
failure modes
[ "saturates at 100 nM so cannot grade depth of response", "measures the exceptional locus and generalises it", "outcome measure presented as target engagement", "conflates displacement with degradation" ]
- Correlative sampling in infants is scarce and irreplaceable. Spending it on a saturating assay is a real cost to the field even though it harms no individual patient.
- A surrogate that reads full engagement in a non-responder invites the conclusion that the target is wrong, when the correct conclusion may be that the target was engaged and the fusion stayed.
- Nothing here questions the drug's efficacy, which rests on response rates in patients and not on any chromatin measurement. The risk is to the mechanistic inferences that guide the next combination, the next dose and the next trial.
- What do the AUGMENT-101 and the paediatric trial correlative plans actually specify? This entry reasons from the preclinical literature; the protocols themselves would settle which surrogates are collected and whether fusion profiling is already included.
- Is residual fusion occupancy measurable from a clinical marrow aspirate at all? CUT&Tag has been run on patient KMT2A-rearranged samples at scale, so the answer is probably yes, and the limiting factor is a paired on-treatment timepoint.
- Does residual fusion occupancy at diagnosis predict depth of response? If the sensitive locus set varies between patients, its size is a candidate predictive biomarker that could be measured once, before treatment.
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Revumenib at 100 nM produces near-complete and global displacement of menin from chromatin in KMT2A::AFF1 cells, with no further room to fall at 1 or 5 microM in wild-type cells - so menin displacement saturates at the low end of the exposure range.
MEN1 mutations mediate clinical resistance to menin inhibition. · Perner F, Stein EM, Wenge DV, et al. · Nature 2023What this evidence cannot show
MV4;11 cells; treatment duration for the ChIP-seq is not stated; depth-normalised ChIP-seq without spike-in, so 'near-complete' is a statement about relative signal. Whether plasma concentrations in paediatric dosing reach and sustain 100 nM intracellularly was not assessed for this entry.
Evidence that argues against this (1)
- Saturation of a target-engagement marker at the lowest effective dose is normal and desirable pharmacology; the problem is only that a saturating marker cannot then be used to explain differences in response between patients, which is how it is often read. Krivtsov AV et al. 2019
- E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
In the same drug state the fusion is not globally displaced, and MEIS1 is one of only three named loci at which more than 2-fold fusion occupancy loss occurs, while HOXA and MYB retain occupancy.
A Menin-MLL Inhibitor Induces Specific Chromatin Changes and Eradicates Disease in Models of MLL-Rearranged Leukemia. · Krivtsov AV, Evans K, Gadrey JY, et al. · Cancer Cell 2019What this evidence cannot show
Shared-epitope antibody except in ML2; TSS-anchored 2-fold thresholding; day 3; adult-derived lines. The inference that MEIS1 is unrepresentative depends on the retained loci being genuinely retained, which the HOXA cluster's repetitive structure makes worth re-checking.
Evidence that argues against this (1)
- If MEIS1 loss is the proximate cause of the therapeutic effect, then measuring MEIS1 is measuring the mechanism, and its unrepresentativeness of genome-wide occupancy is irrelevant to its value as a biomarker. That is a defensible position and it is the one the field implicitly holds. Klossowski S et al. 2020
- E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
MEIS1 was established as the pharmacodynamic biomarker for this class in the preclinical work that took a menin inhibitor to remission in KMT2A-rearranged and NPM1-mutated models, alongside differentiation and proliferation readouts.
Menin inhibitor MI-3454 induces remission in MLL1-rearranged and NPM1-mutated models of leukemia. · Klossowski S, Miao H, Kempinska K, et al. · J Clin Invest 2020What this evidence cannot show
MI-3454, a preclinical compound; the biomarker's use in the clinical trials is inferred from the preclinical framing and from the class literature rather than read off a protocol document here. Differentiation and remission endpoints in that work are strong and are not in question.
Evidence that argues against this (1)
- MEIS1 has the practical virtues a biomarker needs - large dynamic range, measurable by qPCR from few cells, tightly coupled to response - and no proposed alternative matches that. Criticising it without supplying a deployable replacement is not actionable for a trialist. Perner F et al. 2023
- E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
A menin inhibitor also degrades menin, shortening its half-life through CHIP/Hsp70-mediated ubiquitination with no change in MEN1 mRNA, so even the menin displacement readout conflates target engagement with target destruction.
Disruption of the menin-MLL interaction triggers menin protein degradation via ubiquitin-proteasome pathway. · Wu Y, Doepner M, Hojnacki T, et al. · Am J Cancer Res 2019What this evidence cannot show
MI-503 in MV4;11 and THP-1, a tool compound; no equivalent measurement for the clinical agents, and the article has no DOI on its PubMed record (PMC6726985).
Evidence that argues against this (1)
- If the clinical compounds do not degrade menin, this failure mode does not apply to them, and the only published degradation data are for tool compounds. A single western time course with revumenib would resolve it. Krivtsov AV et al. 2019
the model's recruitment variable versus a menin-displacement readout
MEIS1's status as the class biomarker is why the loss set matters clinically
the approved agent whose pharmacodynamic logic this risk concerns
the trial in which the correlative logic was established
the class pharmacodynamic biomarker, and one of the three unrepresentative loci
correlative sampling is scarcest in the population the class is being taken into
the surrogate failure modes follow from the measurement's single dimension