Hypothesis: adaptive scheduling of menin inhibitors will fail, because MEN1 resistance mutations probably cost the cell nothing
There is a treatment strategy that deliberately leaves some drug-sensitive leukemia alive to keep the resistant cells in check. It only works if the resistant cells are worse off when the drug is away. For this drug, nobody has checked, and the closest comparison suggests they are not.
The same thing, for a scientist
Adaptive therapy depends on the resistant genotype being competitively inferior in the absence of drug; that assumption carried the prostate pilot, in which adaptive dosing extended median time to progression to at least 27 months at 47% of standard cumulative dose [E1], and it carried the melanoma result, in which resistant tumours were drug-dependent and regressed on drug withdrawal [E2]. We hypothesise the assumption fails here: MEN1 interface mutants preserve the menin-MLL1 interaction and merely impair small-molecule binding [E3], which is exactly the kind of lesion that should be near-neutral off drug — and the closest leukemia precedent, BCR-ABL T315I and Y253H, showed no growth advantage and no measured cost in competitive repopulation assays without imatinib [E4]. We therefore predict selection coefficients for MEN1 mutants off drug in the range -0.02 to +0.02, and consequently that adaptive scheduling delivers no durability benefit in this disease. This is a deliberately negative hypothesis, written to stop an attractive idea before resources are committed to it.
Why a negative hypothesis is worth a graph entry
Adaptive therapy is intellectually appealing, has a published clinical pilot, and is the kind of idea that will be proposed for this disease sooner or later. The ecological analogy that generates it in this graph also generates the reason to doubt it, and writing only the appealing half would be exactly the dressed-up reasoning the brief for this beat warns against.
The structural argument is short. A MEN1 interface mutation changes a residue that contacts a drug, and the published mechanism is explicit that it preserves binding to the natural ligand MLL1 3. A mutation that breaks a contact with a synthetic molecule while leaving the physiological interaction intact has, by construction, almost no reason to impair fitness in the drug's absence. Contrast the melanoma case, where resistance ran through over-expression of the drug target, a change with obvious metabolic and signalling cost 2.
The one way it could be wrong, and it is worth taking seriously
Fitness effects of resistance mutations are mutation-specific, not class-specific. Among BCR-ABL kinase-domain mutants, M351T and H396P were less potent transformers than unmutated BCR-ABL while Y253F and E255K were more potent 5. If one of the four recurrent MEN1 mutants happens to carry a real cost, adaptive scheduling might work against that mutant specifically — and MEN1 mutation is reported in about 39% of post-revumenib relapses 6, so a mutant-specific strategy would still address a substantial fraction of failures.
This is why the hypothesis specifies a measurement across all four recurrent mutants rather than a class-level claim, and why a mixed result is a real possible outcome rather than a hedge.
The second reason adaptive therapy is a poor fit here, independent of fitness
Even with a fitness cost, adaptive therapy requires a real-time burden signal precise enough to time the on-off cycles; the prostate pilot used PSA. Acute leukemia has no continuous equivalent — marrow MRD is intermittent, invasive and slow, and peripheral blast counts are unreliable in remission. And the objective differs: adaptive therapy explicitly aims at containment rather than eradication, which is defensible in metastatic prostate cancer and much harder to defend in a child whose disease is potentially curable and whose relapse hazard is front-loaded.
So there are two independent reasons to expect failure, one biological and one operational, and the experiment addresses only the first. If the fitness measurement comes back favourable, the operational objection still stands and should be confronted before any schedule is proposed.
| statement | MEN1 drug-interface resistance mutations (M327I, G331R, G331D, T349M) carry no meaningful fitness cost in the absence of menin inhibitor, with selection coefficients between -0.02 and +0.02 in competition against unmutated cells; consequently adaptive or intermittent scheduling of menin inhibitors will not suppress resistant clones and will not extend durability, and any benefit of intermittent dosing in this disease must come from reduced toxicity rather than from evolutionary suppression. |
|---|---|
| origin | analogy |
| source analogy | analogy-ecology-competitive-release-and-adaptive-therapy |
| prior probability | 0.6 |
| prior reasoning | Higher than most entries here because it is a prediction of no effect grounded in mechanism: the mutation breaks a contact with a synthetic molecule and preserves the physiological one, so there is no structural reason for a cost. The closest leukemia precedent shows neutrality. The 40% of the prior held back reflects the demonstrated mutation-specificity of fitness effects among BCR-ABL mutants, and the fact that menin has clients other than MLL1 whose engagement by the mutants has not been assessed. |
| relationship to other entries | Shares an experiment with hyp-agency-revocation-hysteresis-pulsed-menin and is deliberately in tension with it: that hypothesis argues for intermittent dosing on kinetic and toxicity grounds, this one denies it the evolutionary justification. Both can be true, and if both are, intermittent dosing should be proposed as a toxicity strategy and never as a resistance strategy. |
| decision value | Prevents a plausible but probably unfounded evolutionary rationale from being attached to a schedule change in children, and forces any intermittent-dosing proposal to justify itself on toxicity alone. |
predictions
[ "Selection coefficients for all four recurrent MEN1 interface mutants in drug-free competition fall between -0.02 and +0.02.", "Intermittent dosing in a resistance-competent in vivo model does not reduce the emerging frequency of MEN1 mutants relative to continuous dosing at matched cumulative exposure.", "MEN1-mutant clones detected in patients do not decline in variant allele frequency during treatment interruptions.", "Any benefit observed from intermittent dosing tracks reduced toxicity and is not accompanied by a change in time to resistance.", "If one mutant does show a cost, it is a specific mutant rather than the class, mirroring the BCR-ABL pattern." ]
experiments
[ "exp-men1-mutant-fitness-cost-competition" ]
analyses
[]
falsifiers
[ "One or more MEN1 interface mutants shows a selection coefficient below -0.05 in drug-free competition, which would make adaptive scheduling rational against that mutant.", "Intermittent dosing reduces the emergence rate of MEN1 mutants in a resistance-competent in vivo model.", "MEN1-mutant variant allele frequency declines measurably during treatment interruptions in patients.", "Menin-inhibitor-resistant cells prove drug-dependent in the way BRAF-inhibitor-resistant melanoma is, regressing on withdrawal." ]
- If mutants are neutral off drug, then treatment interruptions for toxicity — which are common — neither help nor harm with respect to resistance, and the reassuring intuition that a break lets sensitive cells recover ground is simply false.
- Confirming neutrality also means resistant clones persist indefinitely once they arise, so a patient who has ever had a MEN1 mutant detected should be assumed to still carry it, which argues against menin-inhibitor re-challenge and for switching mechanism entirely.
- Do MEN1 interface mutants alter menin's engagement of clients other than MLL1 — JUND in particular — in a way that could impose a cost the biochemistry of the MLL1 interaction alone would miss?
- Is there any real-time disease-burden signal in acute leukemia precise enough to guide adaptive scheduling, independent of the fitness question?
- Is containment rather than eradication ever an acceptable objective in a potentially curable childhood leukemia, and if not, does that alone rule out adaptive therapy here regardless of the biology?
- E1moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
An evolutionary model and pilot trial of adaptive abiraterone dosing in metastatic castrate-resistant prostate cancer maintained stable oscillations of tumour burden in 10 of 11 patients with median time to progression of at least 27 months at 47% of standard cumulative dose, against a median of about 16.5 months under standard continuous dosing.
Integrating evolutionary dynamics into treatment of metastatic castrate-resistant prostate cancer · Zhang J, Cunningham JJ, Brown JS, Gatenby RA · Nature Communications 2017What this evidence cannot show
Eleven patients, non-randomised, in a disease with a continuous serum biomarker and a containment rather than cure objective. Every one of those features is absent in infant acute leukemia.
Evidence that argues against this (1)
- REFUTES THE INFERENCE, NOT THE MEASUREMENT. The entry's first clause (MEN1 interface mutants are near-neutral off drug) is a reasonable prediction. The second clause — 'consequently adaptive or intermittent scheduling will not suppress resistant clones and will not extend durability' — does not follow. Reviewing modelling work on precisely this point, Wodarz concludes that a clinical benefit from adaptive therapy 'can be achieved even in the absence of a fitness cost for resistant cells, which broadens the applicability of adaptive therapy': competitive suppression can operate through space and shared resources between populations of equal intrinsic fitness. So a null result in `exp-men1-mutant-fitness-cost-competition` would license the first clause and not the second, and the entry as titled asserts more than its evidence can carry. Recommend narrowing the statement to the fitness-cost prediction and treating the scheduling conclusion as a separate, weaker hypothesis that needs its own model. Wodarz D 2021
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Vemurafenib-resistant melanoma xenografts became drug-dependent, regressing when drug was withdrawn, and discontinuous dosing exploiting the fitness disadvantage of resistant cells off drug forestalled lethal resistant disease.
Modelling vemurafenib resistance in melanoma reveals a strategy to forestall drug resistance · Das Thakur M, Salangsang F, Landman AS, et al. · Nature 2013What this evidence cannot show
Resistance there ran through over-expression of the drug target, a mechanism with obvious cost. A point mutation preserving the physiological interaction is a structurally different and much less promising case.
Evidence that argues against this (1)
- Vemurafenib drug-dependence is BRAF-mutant melanoma in xenografts, where resistance arises through MAPK-pathway hyperactivation that becomes toxic without drug. That is a specific mechanism of drug addiction tied to a signalling pathway with a tolerable activity window; menin inhibition has no equivalent described. Nature 2023
- E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
MEN1 interface mutations confer clinical resistance by attenuating small-molecule binding through structural perturbation while preserving the interaction with the natural ligand MLL1 and preventing inhibitor-induced eviction of menin and MLL1 from chromatin.
MEN1 mutations mediate clinical resistance to menin inhibition · Perner F, Stein EM, Wenge DV, et al. · Nature 2023What this evidence cannot show
Preservation of MLL1 binding is shown biochemically; it does not by itself establish that the mutant cell is fully fit, since menin has other clients whose engagement was not assessed. That residual possibility is what the experiment measures.
Evidence that argues against this (1)
- Even if MEN1 interface mutants are cost-free, adaptive scheduling would address only part of the failure, so the hypothesis's prediction of outright failure is over-strong in one direction and under-specified in the other. PRIMARY, pre-existing resistance to menin inhibitors is demonstrable in MLL-fusion AML models independently of any MEN1 mutation and is overcome by adding KAT6/KAT7 catalytic inhibition, with KAT7 remaining a dependency in both genetic and non-genetic acquired resistance. A schedule designed around the competitive dynamics of MEN1-mutant clones has no purchase on disease that never responded, which in the pooled clinical data is the majority. Gordon SJV et al. 2025
- E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
In competitive repopulation assays in primary murine bone marrow in vitro and in a murine CML transplantation model, BCR-ABL T315I and Y253H conferred no growth advantage over unmutated BCR-ABL in the absence of imatinib, with similar kinase activity and substrate phosphorylation, while imatinib clearly selected for the mutants.
The Bcr-Abl mutations T315I and Y253H do not confer a growth advantage in the absence of imatinib · Miething C, Feihl S, Mugler C, et al. · Leukemia 2006What this evidence cannot show
Absence of advantage is not the same as absence of cost; the study was not designed to detect a small cost, so it supports neutrality rather than proving it.
Evidence that argues against this (1)
- BCR-ABL is a constitutively active KINASE and MEN1 encodes a scaffold; a kinase-domain mutation changes catalysis and drug contact at the same site, whereas a MEN1 interface mutation attenuates drug binding while preserving the natural ligand. There is no reason the fitness arithmetic should transfer. The competitive repopulation assays also measure fitness of transduced murine haematopoietic cells, which is not the fitness of an established human leukaemic clone under treatment. Nature 2023
- E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Among five common BCR-ABL kinase-domain mutants, M351T and H396P were less potent transformers than unmutated BCR-ABL while Y253F and E255K were more potent and T315I was assay-dependent, showing that fitness effects of resistance mutations are mutation-specific.
Kinase domain mutants of Bcr-Abl exhibit altered transformation potency, kinase activity, and substrate utilization, irrespective of sensitivity to imatinib · Griswold IJ, MacPartlin M, Bumm T, et al. · Molecular and Cellular Biology 2006What this evidence cannot show
Transformation potency rather than competitive fitness, and a kinase rather than a chromatin adaptor. Cited to justify measuring each MEN1 mutant separately.
Evidence that argues against this (1)
- BCR-ABL is a constitutively active KINASE and MEN1 encodes a scaffold; a kinase-domain mutation changes catalysis and drug contact at the same site, whereas a MEN1 interface mutation attenuates drug binding while preserving the natural ligand. There is no reason the fitness arithmetic should transfer. The competitive repopulation assays also measure fitness of transduced murine haematopoietic cells, which is not the fitness of an established human leukaemic clone under treatment. The mutation-specific and assay-dependent result is the finding, and it argues that a fitness cost cannot be assumed for a class. Nature 2023
- E6moderate confidence 3 of 5: One strong study, or several smaller consistent ones.review
Acquired MEN1 mutations were described in 39% of post-revumenib relapses and have not been identified following ziftomenib or bleximenib.
The promise of menin inhibitors: from approval to triplet regimens · Thakur RK, Wang ES · Hematology, American Society of Hematology Education Program 2025What this evidence cannot show
A review; the figure sets how much of the failure space a mutant-specific adaptive strategy could address if one mutant did prove costly.
Evidence that argues against this (1)
- A review's observation that MEN1 mutations have not been identified after ziftomenib or bleximenib is an absence from a much smaller and less sequenced relapse experience. Using it to infer a chemotype-dependent resistance landscape runs ahead of the data. Journal of Clinical Oncology 2025
counter-evidence pass: adaptive-therapy benefit is achievable without a fitness cost (Wodarz 2021), so a null result in this experiment does not license the entry's scheduling conclusion
the stated prerequisite of this analogy — that adaptive therapy needs a fitness cost to resistance — is contradicted in the mathematical-oncology literature
- proposed by →Competitive release and adaptive therapy — what ecology says happens when you remove the dominant species
generated by the fitness-cost prerequisite of adaptive therapy
denies intermittent dosing its evolutionary justification, leaving only the toxicity argument
- related to →MEN1 drug-interface mutations — the first proof that a chromatin drug can drive Darwinian escape in patients
predicts near-neutral fitness off drug for all four recurrent interface mutants
the selection-coefficient measurement on which the negative prediction rests