Experiment: measure the two missing time constants — how fast the KMT2A::AFF1 program collapses on drug, and how fast it comes back off drug
Add a menin inhibitor, watch how quickly the leukemia program shuts down, then wash the drug out and watch how quickly it comes back. Nobody has ever done the second half.
The same thing, for a scientist
A pulse-chase design with time-resolved readouts on both arms. Arm A measures suppression kinetics from drug addition; Arm B measures recovery kinetics from washout, which has never been reported for any KMT2A fusion - bounded 2026-09-19: PubMed `menin[Title/Abstract] AND (washout OR withdrawal OR "drug holiday" OR intermittent OR rebound OR reversib*)[Title/Abstract]` returns 11 records, none of them a drug-withdrawal recovery time course in a KMT2A-rearranged model. Readouts are staged from fastest to slowest layer: nascent transcription, steady-state mRNA, chromatin occupancy of menin and the fusion, H3K79me2, and HOXA9/MEIS1 protein. The output is a pair of half-times whose ratio determines whether an intermittent schedule is worth testing in vivo. Arm C is the in vivo schedule comparison at matched cumulative dose, which is only run if the ratio exceeds a pre-specified threshold.
Design
Cells. SEM and MV4-11 (both KMT2A::AFF1) plus at least two KMT2A::AFF1 infant patient-derived xenograft lines ex vivo; MOLM-13 (KMT2A::MLLT3) as a cross-partner comparator; an NPM1-mutant line (OCI-AML3) to test whether hysteresis is principal-specific; a KMT2A-germline line as a specificity control.
Arm A, collapse. Revumenib at 5x the cellular IC90. Sample at 0, 15, 30, 60 min, 2, 4, 8, 24, 48, 72 h.
Arm B, recovery. Treat to steady-state suppression (72 h), then wash out with confirmed free-drug clearance by LC-MS/MS on the medium. Sample at the same intervals after washout, extended to 7 days.
Arm C, in vivo. Only if the Arm B half-time exceeds the Arm A half-time by the pre-specified factor of 3. KMT2A::AFF1 PDX mice randomised to continuous dosing versus an intermittent schedule at matched cumulative dose (for example 4 days on, 3 days off), with survival as the primary endpoint and serial marrow sampling for program suppression and for emergent MEN1 mutations.
Readouts, layered so the lag structure is visible
- Nascent transcription (TT-seq or PRO-seq) at HOXA cluster, MEIS1, PBX3, MEF2C — the fastest layer, where the 15-minute effect was seen with a degron 1.
- Steady-state mRNA by RNA-seq, which lags nascent by transcript half-life.
- Chromatin occupancy of menin, KMT2A N-terminus and LEDGF by CUT&RUN, to confirm that washout restores occupancy before it restores output.
- H3K79me2 by CUT&Tag over HOXA gene bodies — the layer whose decay half-life is currently unreported anywhere.
- HOXA9 and MEIS1 protein by capillary immunoassay, the slowest and the one that actually governs the phenotype.
Fitting each layer separately is the point: the hypothesis is about the ordering and spacing of layers, not about a single number.
Controls and the ways this can go wrong
Free-drug clearance after washout must be measured, not assumed; residual intracellular drug would fake hysteresis and is the single most likely artefact. Include a drug-free mock-washout arm. Use a second, structurally distinct menin inhibitor (ziftomenib) to separate compound pharmacology from target biology — this matters because the class is already known to differ in resistance-mutation profile between agents 2. Confirm that cells are not simply dying and regrowing, by tracking viability and by performing the analysis on a sorted viable population.
Go/no-go: if Arm A does not reproduce rapid program suppression in the KMT2A::AFF1 lines, stop and fix the model before interpreting Arm B.
| design | Pulse-chase kinetics of menin-inhibitor effect in KMT2A::AFF1 cells and PDX-derived cells: Arm A collapse kinetics from drug addition, Arm B recovery kinetics from confirmed washout, Arm C conditional in vivo comparison of continuous versus intermittent schedule at matched cumulative dose. |
|---|---|
| readout | Layered half-times for nascent transcription (TT-seq/PRO-seq), steady-state mRNA, menin/fusion/LEDGF chromatin occupancy (CUT&RUN), H3K79me2 (CUT&Tag), and HOXA9/MEIS1 protein; primary estimand is the ratio of recovery half-time to collapse half-time. |
| controls | LC-MS/MS confirmation of free-drug clearance after washout; mock-washout arm; second structurally distinct menin inhibitor (ziftomenib); KMT2A-germline line as specificity control; viability tracking and sorted-viable-cell analysis to exclude kill-and-regrow artefacts. |
| cost tier | med |
| timeline | Arms A and B, 4-6 months. Arm C, a further 6-9 months, and only if the pre-specified ratio threshold is met. |
| go no go | Proceed to Arm C only if recovery half-time exceeds collapse half-time by a factor of 3 or more in at least two independent KMT2A::AFF1 models. |
| decision value | Produces the H3K79me2 decay half-life, a number the graph records as absent from the entire indexed literature, and determines whether a paediatric schedule study is justified. |
| why nobody has done it | Washout kinetics are unglamorous and were not needed for approval. The collapse arm answers a mechanism question and got published; the recovery arm answers a dosing question and had no owner. |
tests hypotheses
[ "hyp-agency-revocation-hysteresis-pulsed-menin" ]
collaborators needed
[ "a group with KMT2A::AFF1 infant PDX lines", "a nascent-transcription facility for TT-seq or PRO-seq" ]
bounded searches
[
{
"query": "menin[Title/Abstract] AND (washout OR withdrawal OR \"drug holiday\" OR intermittent OR rebound OR reversib*)[Title/Abstract]",
"records": 11,
"relevant": 0,
"date": "2026-09-19",
"source": "PubMed eutils esearch"
}
]- A negative result in Arm B would mean continuous occupancy is required, which strengthens the case for a degrader over an occupancy inhibitor and weakens the whole reversible-inhibitor strategy.
- Residual intracellular drug is the obvious way to get a false positive here, and a false positive would lead to a schedule change in infants on the basis of an artefact.
- Is ex vivo washout in PDX-derived cells a fair model of a drug holiday in a patient, where clearance is governed by CYP3A4 and the trough is never truly zero?
- Should the intermittent schedule be defined in days or in cycles, and does the marrow microenvironment change the recovery constant relative to suspension culture?
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Degron-mediated MLL::AF9 degradation changes transcriptional elongation at core target genes within 15 minutes, with loss of the active chromatin landscape occurring subsequently, establishing that a rapid nascent-transcription readout is the right first layer.
MLL::AF9 degradation induces rapid changes in transcriptional elongation and subsequent loss of an active chromatin landscape · Olsen SN, Godfrey L, Healy JP, et al. · Molecular Cell 2022What this evidence cannot show
A degron is a cleaner and more complete perturbation than a small molecule occupying one interface; the collapse arm with a drug may be slower and less complete.
Evidence that argues against this (1)
- The kinetic paper this design rests on measured the *onset* limb, not the recovery limb, and its finding argues against hysteresis rather than for it. Degron-mediated MLL::AF9 removal changed transcriptional elongation at core targets within 15 minutes, with chromatin-landscape loss following. A system whose transcriptional output responds within minutes of losing the driver is a system with a short memory at the transcriptional layer; hysteresis, if it exists, would have to live in the slower chromatin layer, and the paper does not measure whether that layer stays changed after the driver returns. The experiment should therefore state explicitly that no published washout/recovery kinetics exist for this system and that it is measuring the recovery limb for the first time — currently the entry reads as though the reference is established in both directions. Olsen SN et al. 2022
- E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.review
Menin inhibitors differ from each other in safety, QTc, cytopenia and pharmacokinetic profiles, and acquired MEN1 mutations described in 39% of post-revumenib relapses have not been identified following ziftomenib or bleximenib, so compound identity is not interchangeable in a schedule experiment.
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
Review-level; the absence of MEN1 mutations after other agents may reflect shorter follow-up and fewer sequenced relapses rather than a true compound difference.
Evidence that argues against this (1)
- Not having been identified is not absence. The post-revumenib relapse cohorts are larger and much more comprehensively sequenced than those for ziftomenib or bleximenib, so the contrast between 39% and none is consistent with differential ascertainment. The safety and pharmacokinetic differences are real and are separate from the resistance claim. Journal of Clinical Oncology 2025
- E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Revumenib in AUGMENT-101 was administered every 12 hours continuously in 28-day cycles, a schedule carried forward from dose escalation rather than derived from a schedule comparison.
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 2024What this evidence cannot show
The trial was not designed to evaluate schedule, so the continuous schedule is a convention rather than an optimum.
Evidence that argues against this (1)
- The clinical case for exploring intermittent schedules has weakened while this entry stood. Continuous twice-daily revumenib inside an all-oral triplet with decitabine/cedazuridine and venetoclax produced composite complete remission of 71% and median CR/CRh duration of 10.5 months in relapsed/refractory AML; the frontline triplet reached composite CR of 81.4% with universal flow-MRD negativity. There is also a specific hazard in pulsing: emergent menin-binding-site mutations occurred in 13% of SAVE patients under *continuous* exposure, and intermittent dosing that spends time at sub-saturating concentrations is the classic condition for selecting partially resistant mutants. A hysteresis experiment remains worth doing, but the schedule question it feeds into should be framed against combination therapy rather than against monotherapy, and should include a mutation-emergence readout. Issa GC et al. 2026
- related to →MEN1 drug-interface mutations — the first proof that a chromatin drug can drive Darwinian escape in patients
add a mutation-emergence readout: intermittent dosing spends time at sub-saturating concentrations, the classic condition for selecting interface mutants, which arose in 13% under continuous triplet exposure
measures collapse and recovery half-times across five molecular layers
would produce the H3K79me2 decay half-life, currently absent from the literature