Oncofusion

start / graph / hyp-therapy-installed-impairment-index

Hypothesisspeculative confidence 1 of 5: Our own inference or analogy. Not shown to be true.Q12-analogy

Hypothesis: prior therapy itself installs menin-inhibitor resistance, and the dose of prior therapy predicts it better than any mutation does

The treatment a child has already had may be what makes the next drug stop working. If so, you could predict who will not respond just by counting what they have already received, and the right answer is to give the drug earlier.

The same thing, for a scientist

In serial patient-derived xenografts, heavily pretreated KMT2A::AFF1 samples were far less menin-inhibitor-sensitive than earlier samples from the same patients, with no prior menin exposure and with sustained on-target transcriptional effect; emergent RAS and TP53 co-mutations were individually insufficient to cause resistance in vitro [E1]. We hypothesise that this is causal and dose-dependent: ex vivo menin-inhibitor sensitivity in KMT2A-rearranged samples declines monotonically with cumulative prior therapy, and a simple index built from treatment history (number of lines, cumulative anthracycline and nucleoside-analogue exposure, prior chromatin-directed agent, prior immunotherapy, prior transplant) predicts ex vivo sensitivity and clinical response better than MEN1 status or any single co-mutation. The competing explanation — that repeatedly relapsed disease is intrinsically different biology and therapy is merely a marker of it — is not separable in observational data and requires the controlled arm described in the experiment.

Go deeper
Separating the causal claim from the predictive one

Predictive claim (prior 0.6). Cumulative prior therapy predicts menin-inhibitor sensitivity. This needs only an association and is testable on banked serial samples. It would be immediately useful for trial stratification even if the mechanism is confounding rather than causation.

Causal claim (prior 0.35). Exposure to particular therapy classes installs the resistant state. This requires the controlled experiment: passage matched KMT2A::AFF1 models through defined chemotherapy exposures in vivo and measure menin sensitivity before and after, against time-matched untreated controls. Only that design separates therapy from disease evolution and from time.

The distinction matters because the two claims imply different actions. The predictive claim says stratify. The causal claim says change the order of therapy, which is a much bigger step and needs the stronger evidence.

Why the controlled arm is not optional

Every observational version of this comparison is confounded three ways: patients who receive more therapy have disease that resists therapy, more therapy means more elapsed time for clonal evolution, and later samples are drawn from patients selected for having survived to be sampled. The PDX serial-sample observation 1 is subject to all three, on n=2.

The controlled version is straightforward and nobody appears to have run it: take an isogenic KMT2A::AFF1 PDX, split it, expose one arm to a clinically representative chemotherapy course in vivo and hold the other in parallel untreated, then harvest both at the same time point and measure menin-inhibitor sensitivity, transcriptome and MEN1 genotype. Two arms, one time point, one readout. The result would either establish the causal claim or retire it.

What follows if the causal claim holds
  1. Sequencing. Menin inhibitors should be moved earlier, not because they work better in fitter patients but because the target state is destroyed by the therapy given while waiting. This converges with hyp-menin-inhibitors-belong-frontline-in-infants and with hyp-menin-value-dominated-by-bridge-rate from a third independent direction.
  2. Bridging. Cytotoxic bridging given between the decision to use a menin inhibitor and its start would be actively harmful, which is testable and currently unexamined.
  3. Trial interpretation. Every single-agent menin-inhibitor trial to date enrolled heavily pretreated patients — 86% intensively pretreated and 77% venetoclax-exposed in one post-failure cohort 3 — which would mean the class has only ever been measured in its worst-case population, and the frontline efficacy is unknown in a systematically favourable direction.
  4. Which class to avoid. If the effect is class-specific, the index identifies which prior therapies to avoid before a planned menin inhibitor, which is actionable in protocol design.
Facts and figures
statementCumulative prior therapy causally reduces menin-inhibitor sensitivity in KMT2A-rearranged leukemia independently of MEN1 mutation and of elapsed time, and a treatment-history impairment index (lines of therapy, cumulative anthracycline and nucleoside-analogue dose, prior chromatin-directed agent, prior immunotherapy, prior transplant) predicts ex vivo menin-inhibitor sensitivity and clinical response better than MEN1 status or any single co-mutation.
originanalogy
source analogyanalogy-finance-pre-existing-impairment-diligence
prior probability0.45
prior reasoningThe predictive form — that treatment history correlates with sensitivity — we put at about 0.6, since it requires only association and there is a directional observation already. The causal form we put at about 0.35, because the observation is n=2 and is confounded by time, by selection and by the intrinsic biology of repeatedly relapsed disease. The composite reflects the hypothesis as stated, which asserts causation.
decision valueBears directly on whether menin inhibitors should be moved frontline, on whether cytotoxic bridging before a menin inhibitor is harmful, and on whether all existing efficacy data systematically understate the drug.
predictions
[
  "In matched serial samples from the same patients, ex vivo menin-inhibitor IC50 rises monotonically with cumulative prior therapy.",
  "In a controlled PDX experiment, an arm passaged through chemotherapy in vivo is less menin-sensitive than a time-matched untreated arm from the same starting material.",
  "The impairment index outperforms MEN1 status in predicting clinical response in menin-inhibitor cohorts where treatment history is recorded.",
  "The effect is class-specific rather than uniform, with chromatin-directed prior therapy contributing more per unit than anthracycline exposure.",
  "Frontline or early-line menin-inhibitor cohorts, when they report, will show markedly higher complete remission rates than the 22.8% seen in relapsed disease, beyond what fitness and disease burden explain."
]
experiments
[
  "exp-serial-sample-menin-sensitivity-ladder"
]
analyses
[
  "ana-menin-resistance-cell-state-signature"
]
falsifiers
[
  "In the controlled PDX arm, chemotherapy-exposed and time-matched untreated cells are equally menin-sensitive — therapy does not install the state and the observation was confounding.",
  "Ex vivo sensitivity across matched serial samples shows no monotone relationship with cumulative therapy.",
  "The impairment index adds nothing to MEN1 status and disease burden in predicting clinical response.",
  "Frontline menin-inhibitor cohorts show remission rates consistent with the relapsed setting after adjustment, removing the main clinical signature of the effect."
]
How this could bite us back
  • If confirmed, every existing menin-inhibitor efficacy figure understates the drug in the frontline setting and overstates what can be expected from salvage use — which cuts in opposite directions for two different clinical decisions.
  • An impairment index derived from a two-patient observation could be used to deny a child their last option. It must not leave the research setting without prospective validation.
  • If therapy installs resistance to chromatin-directed agents generally, then the whole obligate-partner-interface strategy has a shelf life that starts running at diagnosis, whatever the target.
Still open
  1. Which prior therapy classes contribute most, and is prior venetoclax or prior immunotherapy as impairing as cytotoxic chemotherapy?
  2. Is the installed state reversible if therapy is withheld for a period, which would make the impairment index time-dependent rather than cumulative?
  3. Do the two menin-sensitive infant PDX samples in the source study represent protection by youth, by early disease, or by chance?
Every source behind this entry (4)
  1. E1moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    Heavily pretreated KMT2A::AFF1 ALL samples were much less sensitive to menin inhibition than earlier samples from the same patients despite no prior menin-inhibitor exposure, with sustained on-target transcriptional effects, and with RAS-pathway and TP53 co-mutations that were individually insufficient to induce resistance in vitro; two infant ALL PDX samples were sensitive.

    Upfront menin-inhibitor resistance in multiply pretreated leukemias · Mahdavi L, Bernt KM, et al. · Experimental Hematology 2025
    PMID 41027490DOI
    What this evidence cannot show

    Two patients, observational, and confounded by time and by selection. It is the entire empirical basis for this hypothesis and it is thin, which is why the hypothesis leads with a controlled experiment rather than with a claim.

    Evidence that argues against this (1)
    • BOUNDS THE EFFECT SIZE HARD. If cumulative prior therapy causally and dose-dependently reduces menin-inhibitor sensitivity, a relapsed/refractory cohort with a median of two prior lines, of whom 52% had already received venetoclax, should respond poorly to a revumenib-containing regimen. In SAVE that cohort achieved composite complete remission of 71%, CR/CRh of 60%, flow-MRD negativity in 80% of responders and median CR/CRh duration of 10.5 months. Whatever the two-patient serial PDX observation is measuring, it is not a general therapy-installed block on menin-directed benefit; at most it identifies a minority phenotype. This does not refute the predictive claim (an index might still rank patients within such a cohort) but it substantially deflates the causal claim and, with it, the sequencing argument that menin inhibitors must be moved earlier because prior therapy destroys the target state. Issa GC et al. 2026
  2. E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    KMT2A::AFF1 B-ALL cells driven to acquired DOT1L-inhibitor resistance became largely independent of H3K79 methylation while still requiring DOT1L, HOXA9 and the fusion, abandoning some fusion target genes and upregulating myeloid genes including CD33 and LILRB4 — showing that drug exposure alone can install a durable adapted state in these cells.

    Modelling acquired resistance to DOT1L inhibition exhibits the adaptive potential of KMT2A-rearranged acute lymphoblastic leukemia. · Schneider P, Crump NT, Arentsen-Peters STCJM, et al. · Experimental Hematology and Oncology 2023
    PMID 37740239DOI
    What this evidence cannot show

    Demonstrates that exposure to a chromatin-directed drug installs adaptation; it does not show that conventional cytotoxic chemotherapy does the same, which is the specific claim here.

    Evidence that argues against this (1)
    • Searched for a controlled study that exposes isogenic KMT2A-rearranged models to clinically representative cytotoxic chemotherapy in vivo and then measures menin-inhibitor sensitivity against time-matched untreated controls — the design this hypothesis says is needed and that nobody appears to have run. None was found. Note also the mismatch in the cited model: acquired DOT1L-inhibitor resistance is sustained, target-directed epigenetic pressure, which is not the same exposure class as anthracycline or nucleoside-analogue chemotherapy; using it as the precedent for 'drug exposure installs a durable adapted state' overstates the transfer. Recorded so the absence is known to have been checked rather than assumed. Q12-analogy-counter (search record) 2026
  3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In 84 adults with relapsed or refractory AML after menin-inhibitor failure, 86% had been intensively pretreated and 77% venetoclax-exposed, with 63% KMT2A-rearranged; response to next therapy was 32% and median overall survival 4.4 months.

    Outcomes of relapsed or refractory acute myeloid leukemia after menin inhibition failure · Multicentre retrospective consortium · Blood Advances 2026
    PMID 41351513DOI
    What this evidence cannot show

    Characterises the population in which the class has been tested; it is the denominator for the trial-interpretation argument and not evidence of causation.

    Evidence that argues against this (1)
    • This cohort is defined by menin-inhibitor failure, so 86% intensive pretreatment and 77% venetoclax exposure describe who reaches that point rather than what prior therapy installs. Building an impairment index on it would fit the selection, and the graph's own finding of resistance in patients with NO prior menin exposure is the cleaner test of whether impairment is therapy-installed. Nature 2023
  4. E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    AUGMENT-101 enrolled patients from 30 days of age with relapsed or refractory disease, producing CR+CRh of 22.8% and overall response 63.2%, so all published single-agent efficacy for this class in KMT2A-rearranged leukemia comes from a relapsed, pretreated setting.

    Menin Inhibition With Revumenib for KMT2A-Rearranged Relapsed or Refractory Acute Leukemia (AUGMENT-101) · Issa GC, Aldoss I, Thirman MJ, et al. · Journal of Clinical Oncology 2024
    PMID 39121437DOI
    What this evidence cannot show

    This supports the claim that frontline efficacy is unmeasured; it says nothing about whether pretreatment caused the shortfall.

    Evidence that argues against this (1)
    • The frontline-versus-relapsed contrast this entry rests on is now measurable, and it is confounded in a way the entry does not allow for. Frontline azacitidine plus venetoclax plus revumenib in newly diagnosed NPM1-mutant or KMT2A-rearranged AML gave ORR 88.4% and composite CR 81.4% (88.9% in KMT2A-rearranged) with universal centralised flow-MRD negativity — far above relapsed/refractory monotherapy. But that comparison changes the line of therapy *and* the regimen *and* the age distribution simultaneously, so it cannot be used as evidence that prior therapy is what impairs response; it is exactly the confounded observational comparison the entry itself says is not separable. Any version of the impairment index validated on cross-trial data will inherit this confound, which strengthens the entry's own insistence that the controlled two-arm PDX experiment is not optional. Zeidner JF et al. 2025
Connected to