Hypothesis: only a mechanistically orthogonal partner pre-empts menin-inhibitor resistance; a second chromatin agent will not
also called orthogonality hypothesis · menin plus immune combination
Every way this leukemia escapes a menin inhibitor works by rearranging how genes are switched on. So adding a second gene-switching drug should fail, while adding something that kills by a completely different route — an immune therapy or an apoptosis drug — should work.
The same thing, for a scientist
Every characterised resistance mechanism to menin inhibition — MEN1 interface mutation, PRC1.1 loss with MYC reactivation, KMT2C/D inactivation, H3K79me-independent maintenance, adaptive transcriptional drift — restores or preserves the leukemic transcriptional output. We hypothesise that combination partners acting within the same transcriptional/chromatin layer (DOT1L, BET, LSD1, other menin inhibitors) will show additive short-term cytoreduction but will not extend durability, whereas partners acting outside it (CD19/CD3 engagement, BCL-2 inhibition, CDK9-mediated collapse of MYC and MCL1) will extend durability. The prediction is discriminating, quantitative and testable in existing models and in samples from trials already enrolling.
Why we think this1 deeper
Line up the resistance mechanisms characterised so far and ask what each one does to the cell:
| Mechanism | What the cell does | Layer |
|---|---|---|
| MEN1 interface mutation | keeps menin and MLL1 on chromatin | transcriptional |
| PRC1.1 (PCGF1/BCOR) loss | reactivates non-canonical menin targets including MYC | transcriptional |
| KMT2C/D inactivation | alters enhancer landscape | transcriptional |
| H3K79me-independent maintenance (DOT1L precedent) | keeps the program without the drugged enzymatic step | transcriptional |
| Pre-existing resistance in pretreated disease | sustains output despite on-target drug effect | transcriptional |
| RAS/FLT3 co-mutation | grows despite the program being off | signalling |
Five of six are the same kind of solution. A leukemia cell facing a chromatin drug solves the problem by rewiring chromatin. That is not a coincidence: it is what a cell with a plastic epigenome and a low mutation burden can most cheaply do.
The corollary is uncomfortable for the combination strategies the field currently favours most. Menin plus DOT1L is strongly synergistic in short-term assays 1. Menin plus BET, menin plus LSD1, and menin-inhibitor switching are all proposed on the same logic. But if the escape is 'rearrange the transcriptional layer', then adding a second lever within that layer gives the cell a harder problem, not a different kind of problem — and cells solve harder versions of problems they already know how to solve.
The one apparent counter-example strengthens the hypothesis rather than weakening it. KAT6/KAT7 inhibition reportedly overcomes both genetic and non-genetic menin resistance 2. That is a chromatin agent. But its mechanism is to physically evict the fusion protein from chromatin by an independent route — it does not modulate the program, it removes the driver. If it holds up in peer review, it defines a third category: not orthogonal in layer, but orthogonal in point of attack.
The supporting clinical signal
In 84 adults failing menin inhibitors, every complete remission achieved with non-trial therapy came from a venetoclax-containing regimen or from menin-inhibitor switching 3. Venetoclax is the orthogonal agent. And the PRC1.1 mechanism explains why it should be: PRC1.1-deficient resistant cells lose their monocytic signature and become BCL-2-dependent 4. The orthogonal partner is not just different, it is specifically enabled by the resistance state.
The corresponding immune argument has no clinical test yet in this combination, but it is the strongest a priori case: a CD19×CD3 engager or a CAR T cell is indifferent to MEN1 genotype, to PRC1.1 status, to MYC levels and to transcriptional state, provided the antigen is present. The one thing that defeats it — lineage switch — is the one thing menin-driven differentiation might either cause or prevent, which is itself the most interesting unresolved question in this beat.
What would kill this hypothesis
This is not a safe hypothesis and it should not be treated as one. Three results would falsify or badly damage it:
- A menin-plus-DOT1L or menin-plus-BET combination that produces durable responses in a resistance-competent in vivo model, not just deeper short-term cytoreduction.
- Menin-inhibitor switching producing durable remissions at a meaningful rate — the retrospective data show 1 response in 2 patients, which is uninformative but not encouraging for the hypothesis if it replicates.
- A resistance mechanism that defeats venetoclax and T-cell engagement as readily as it defeats menin inhibition, for example a shared apoptotic-threshold phenotype, which would mean the layers are not independent after all.
| statement | Combination partners that act outside the transcriptional/chromatin layer (BCL-2 inhibition, CD19/CD3 engagement, cellular therapy, CDK9-mediated MYC/MCL1 collapse) will extend the durability of menin inhibition in KMT2A-rearranged leukemia, whereas partners acting within that layer (DOT1L, BET, LSD1, alternative menin inhibitors) will deepen short-term cytoreduction without extending durability. The exception is an agent that physically evicts the fusion from chromatin by an independent route, such as KAT6/KAT7 inhibition. |
|---|---|
| origin | data-pattern |
| prior probability | 0.55 |
predictions
[ "In a resistance-competent in vivo model under continuous menin-inhibitor pressure, time-to-resistance is extended by venetoclax or a T-cell engager but not by a DOT1L or BET inhibitor, despite the latter producing greater initial log-kill.", "Resistant clones emerging under menin plus a second chromatin agent will be transcriptionally similar to those emerging under menin alone; those emerging under menin plus venetoclax will show apoptotic-threshold changes (MCL1 or BCL-xL dependence) instead.", "In patients on menin plus venetoclax phase 3 trials, MEN1 interface mutations at relapse will be no less frequent than under monotherapy — because venetoclax does not prevent the mutation, it kills the cell that has it.", "Menin-inhibitor switching will produce short responses without durable remissions.", "KAT6/KAT7 inhibition will behave like an orthogonal partner despite being a chromatin agent, because its mechanism is driver eviction rather than program modulation." ]
experiments
[ "exp-orthogonality-time-to-resistance-panel" ]
analyses
[ "ana-menin-resistance-cell-state-signature" ]
falsifiers
[ "A menin plus DOT1L or menin plus BET combination extends time-to-resistance in a resistance-competent in vivo model as much as or more than menin plus venetoclax.", "Menin-inhibitor switching produces durable remissions in a prospective series, showing that resistance is largely compound-specific rather than pathway-level.", "Resistant clones arising under menin plus venetoclax are transcriptionally indistinguishable from those arising under menin monotherapy, showing that the apoptotic layer is not independent.", "A single resistance mechanism is found that confers cross-resistance to menin inhibition, venetoclax AND CD19-directed therapy simultaneously." ]
- If the hypothesis is right, the field's most popular combination strategies (menin+DOT1L, menin+BET, inhibitor switching) are being pursued on a reasoning error, and patients will be exposed to added toxicity for deeper but not longer responses.
- If the hypothesis is wrong in the specific case of KAT6/KAT7, the layer framing is too coarse and could cause the lab to dismiss the single most promising resistance-overcoming agent in the literature.
- Orthogonal partners have their own bite-back: venetoclax selects MCL1 dependence, T-cell engagers select lineage switch. Orthogonality buys independence of resistance mechanisms, not absence of them.
- Combining orthogonal agents in an infant stacks differentiation syndrome, tumour lysis, cytokine release and infection risk in a patient with no physiological reserve — the mechanistic argument and the tolerability argument point in opposite directions.
- Two of this hypothesis's four supporting evidence items argue against it. E1 and E2 both show NON-orthogonal, transcriptional combinations working - menin plus DOT1L synergising, KAT6/KAT7 inhibition overcoming acquired menin resistance - which is what the hypothesis predicts should fail. They were filed as support.
- The orthogonality premise appears to be backwards in this disease. KMT2A-rearranged leukaemia's documented escape from CD19-directed therapy is lineage switch, a transcriptional identity change - so the engager shares an escape route with the menin inhibitor rather than supplying a second one.
- Is 'time to resistance' even measurable in the models this field uses? Almost all preclinical combination work reports short-term synergy indices, which are the wrong readout for this question, and changing that readout may be the most valuable methodological contribution the lab could make.
- Does menin-driven differentiation raise or lower CD19 surface density? If it lowers it, the orthogonal partner this hypothesis most favours is undermined by the very drug it is meant to protect.
- Where does KAT6/KAT7 inhibition actually sit? Resolving whether driver eviction is functionally orthogonal would sharpen the hypothesis from a layer claim into a mechanism claim.
- If non-orthogonal transcriptional pairs (menin+DOT1L, menin+KAT6/7) are the combinations with evidence behind them, and the engager's escape route is the same lineage-switch mechanism, should this hypothesis be inverted rather than tested as written?
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Menin inhibition combined with DOT1L inhibition is markedly synergistic in KMT2A-rearranged leukemia models, enhancing differentiation and cell killing while sparing normal haematopoiesis and non-KMT2A leukemias, and revumenib plus pinometostat is significantly synergistic in primary KMT2A-rearranged ALL.
Complementary activities of DOT1L and Menin inhibitors in MLL-rearranged leukemia · Dafflon C, Craig VJ, Méreau H, et al. · Leukemia 2017What this evidence cannot show
Cited here as the strongest evidence AGAINST this hypothesis. All synergy readouts are short-term; no study has tested whether the combination extends durability in a resistance-competent in vivo model, which is precisely the distinction this hypothesis rests on. Synergy in a cytotoxicity assay and prevention of resistance are different claims and the literature routinely conflates them.
Evidence that argues against this (2)
- This evidence item refutes the hypothesis it is filed under. The hypothesis predicts that a second gene-switching drug should FAIL, because every escape route is transcriptional. Menin and DOT1L inhibitors are both transcriptional, and combining them is markedly synergistic, with a second group independently reporting synergy with DOT1L inhibition in KMT2A-rearranged leukaemia. If non-orthogonal pairs are the ones that work, the central premise is wrong as stated. Leukemia 2017
- Independently reported, which removes the single-group explanation for the previous objection: distinct responses to menin inhibition across KMT2A-rearranged models come with synergy on adding DOT1L inhibition. Two transcriptional agents again. International Journal of Molecular Sciences 2024
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Combined catalytic inhibition of KAT6A/B and KAT7 rapidly evicts the MLL fusion protein from chromatin, represses oncogenic transcription, and overcomes primary resistance as well as acquired genetic and non-genetic resistance to menin inhibition.
Catalytic Inhibition of KAT6/KAT7 Enhances the Efficacy and Overcomes Primary and Acquired Resistance to Menin Inhibitors in MLL Leukemia. · Cancer Discovery 2025What this evidence cannot show
If confirmed, it would define a third category of partner (chromatin-layer but driver-evicting) and would require this hypothesis to be restated in terms of point of attack rather than layer. Treated here as a named risk to the hypothesis rather than as support for anything. REGRADED weak->strong 2026-09-20: this evidence item restates what PMID 40536420 (Gordon et al., Cancer Discov 2025, 'Catalytic Inhibition of KAT6/KAT7 Enhances the Efficacy and Overcomes Primary and Acquired Resistance to Menin Inhibitors in MLL Leukemia') reports in its own models, and two other entries carry the same claim at strong. Per concept-the-confidence-grading-rule the grade attaches to the CLAIM, so identical claims must carry identical grades; a weaker grade here was expressing how load-bearing the paper is for THIS entry's argument, which belongs in the caveat rather than in the confidence field. The hedging this entry needs is that the work is preclinical and from one group.
Evidence that argues against this (1)
- The strongest case against orthogonality in this entry is the entry's own E2. KAT6A/B and KAT7 are chromatin acetyltransferases - squarely the class the hypothesis says should be redundant with menin inhibition - and their catalytic inhibition overcomes PRIMARY AND ACQUIRED menin-inhibitor resistance by evicting the fusion from chromatin. That is a non-orthogonal combination defeating exactly the resistance the orthogonal one was proposed to pre-empt. CORRECTED 2026-09-19: this note previously flagged the work as a The peer-reviewed version exists and is Cancer Discovery 2025, PMID 40536420, so the objection stands WITHOUT that discount. Cancer Discovery 2025
- E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
Among 50 adults treated after menin-inhibitor failure, all complete remissions with non-trial therapy occurred with venetoclax-containing regimens (HMA/VEN 2 of 13, intensive chemotherapy plus venetoclax 4 of 6) or with menin-inhibitor switching (1 of 2), and no patient with FLT3-ITD responded to gilteritinib.
Outcomes of relapsed or refractory acute myeloid leukemia after menin inhibition failure · See PubMed record PMID 41351513 for the full author list (not transcribed here) · Blood Advances 2026What this evidence cannot show
Retrospective, adults only, tiny numbers per regimen (2 of 13, 4 of 6, 1 of 2). This is a pattern consistent with the hypothesis, not a test of it. Venetoclax-containing regimens also include intensive chemotherapy, so the orthogonality attribution is not clean.
Evidence that argues against this (2)
- The immune arm is not orthogonal in this disease, which inverts the hypothesis rather than qualifying it. KMT2A-rearranged leukaemia escapes CD19-directed therapy by LINEAGE SWITCH - post-immunotherapy lineage switch is now described as an emergent entity, and 64.3% of a 70-case international cohort carried a KMT2A rearrangement, with 81.4% of switches inside six months. Lineage switch is a transcriptional identity change: the same escape class the hypothesis assigns to epigenetic drugs. Adding an engager does not add a second failure mode, it re-uses the first one. Blood 2025
- Mechanistically documented, not only epidemiological: MLL-rearranged B-ALL acquires a CD19-negative myeloid phenotype and escapes CD19 CAR-T that way. The engager's own resistance mechanism is a change in cell identity, so a menin-plus-engager combination puts two drugs in front of one escape route. Blood 2016
- E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
PRC1.1-deficient, menin-inhibitor-resistant cells show reduced monocytic gene signatures, are susceptible to BCL-2 inhibition, and are resensitised to menin inhibition by venetoclax.
Epigenetic regulation of noncanonical menin targets modulates menin inhibitor response in acute myeloid leukemia · See PubMed record PMID 39158067 for the full author list (not transcribed here) · Blood 2024What this evidence cannot show
Preclinical AML models only; no patient data. This is the mechanistic underpinning of the hypothesis's strongest prediction and it is also the case where the 'orthogonal' partner is specifically enabled by the resistance state, which is a stronger and narrower claim than general orthogonality.
Evidence that argues against this (1)
- The orthogonality this hypothesis proposes is being pre-empted by a different pairing that already has strong human data, and the venetoclax-resensitisation result argues for that rival rather than for an engager. In newly diagnosed NPM1-mutant or KMT2A-rearranged AML in patients 60 and over, azacitidine + venetoclax + revumenib produced a composite CR rate of 81.4% (95% CI 66.6-91.6; KMT2A-r 88.9%), ORR 88.4%, no refractory disease after 1-2 cycles, and MRD negativity in all 37 patients evaluated by central flow, with differentiation syndrome in 19% and no maximum tolerated dose. If PRC1.1-deficient, menin-inhibitor-resistant cells are resensitised by BCL-2 inhibition, the obvious clinical instantiation is the menin + venetoclax + hypomethylating triplet, not a menin inhibitor plus a T-cell engager — and the triplet is already generating the efficacy data an engager combination would have to beat. Zeidner JF et al. 2025
puts a magnitude and a possible negative sign on the orthogonality argument
common-cause failure is the expected dominant failure mode of any redundant design, not a residual risk
pursues that hypothesis's own listed falsifier as a named adversarial hypothesis
the same experiment can vindicate the orthogonality strategy if survivors diverge
raised antigen presentation and reduced donor T-cell exhaustion give an independent mechanistic reason to pair menin inhibition with a T-cell-directed agent
orthogonality and ordering are different questions about the same combination
pairing the interface drug with an orthogonal mechanism to pre-empt escape
aims to remove the selection pressure that creates these mutations
the G2 gate is a direct test of whether a second chromatin agent pre-empts resistance
- tests →Experiment: time-to-resistance, not synergy index, as the readout for menin-inhibitor combinations
time-to-resistance panel is the discriminating experiment
- proposed by →Adaptive, non-genetic resistance — the leukemia keeps the fusion and rewires how it is maintained
the hypothesis is generated by the observation that five of six resistance routes act in the same layer
predicts venetoclax extends durability where a second chromatin agent would not
- ← refutesPinometostat (EPZ-5676) — the DOT1L inhibitor, and the most instructive failure in this field
strong menin+DOT1L synergy in short-term assays is the main evidence against the orthogonality hypothesis
- ← supportsBlinatumomab in infant KMT2A-rearranged ALL — the largest single outcome gain in thirty years
the orthogonal immune partner the hypothesis predicts should extend durability