GMP-like state — the destination of therapy-driven lineage switch and, separately, the state most sensitive to menin inhibition
also called GMP-like · GMP-like AML · granulocyte-monocyte progenitor-like programme
Leukaemia cells can take on the character of a particular normal bone-marrow cell — the one that makes neutrophils and monocytes — without actually being that cell. Two separate studies found that this is where cells go when therapy forces them to change identity, and that this is also where the newest class of drug for this disease works best.
The same thing, for a scientist
The GMP-like state is a transcriptional and proteomic programme occupied by leukaemic blasts, not a cell type: the blast scores as a granulocyte-monocyte progenitor against a normal haematopoietic reference while retaining its founding lesion. Two independent lines of evidence give it standing in KMT2A-rearranged disease. In lineage-switching samples the MLL oncoprotein's level drops and it preferentially activates **GMP genes**; in one sample that switched during revumenib treatment, oncoprotein and menin fell to undetectable levels while **ENL persisted** on oncoprotein-target loci including the GMP-like lineage-switching programme [E1]. Independently, a large-scale drug screen integrating genetics and quantitative proteomics with sensitivity across primary AML found **GMP-like AML to be the most bleximenib-sensitive class** [E2]. The two are recorded as a paradox in `hyp-gmp-destination-paradox-menin`; this entry defines the state that hypothesis is about and states plainly the reason the paradox may dissolve — the two studies may not be describing the same programme.
State, not cell type — and why the distinction is load-bearing here
cell-type-gmp is a normal cell that can be sorted out of a marrow aspirate with four antibodies. The GMP-like state is a scoring result: a leukaemic blast projected onto a normal haematopoietic reference lands nearest the GMP. The blast still carries its fusion; it may express CD19 or CD34 that no GMP would; it may be arrested where no GMP arrests.
The distinction matters because the two pieces of evidence that make this state interesting were generated by different methods against different references. E1 is a chromatin method — CUT&Tag oncoprotein occupancy over gene sets. E2 is a proteomic clustering of primary AML samples. Both produce a label that reads 'GMP-like'. Nothing yet shows that they identify the same cells.
This is the most likely resolution of the apparent paradox, and it is recorded as the second falsifier in hyp-gmp-destination-paradox-menin: if the chromatin-defined GMP-like programme and the proteomic GMP-like class do not overlap when scored on the same cells, the two literatures are describing different states and there was never a paradox.
The clinical question this state decides
A clinician watching a marrow turn myeloid while a child is on a menin inhibitor has to choose between stopping the drug and continuing it. The two readings of the GMP-like state give opposite answers.
Stop. Lineage switch in KMT2A-rearranged leukaemia is a recognised escape mechanism with poor outcomes; a marrow changing lineage under a targeted agent is the classic signature of selection for a resistant population.
Continue. If the cell is moving into the state where menin inhibition has the greatest effect 2, stopping removes the pressure at the moment it would finally bite, and the switch is transitional rather than terminal.
The one profiled patient who switched on revumenib did badly, and the chromatin data offer a mechanism that is compatible with 'continue': the oncoprotein and menin were gone, but ENL was still sitting on the target loci 1. On that reading the cell escaped not by changing lineage but by keeping the transcriptional programme running through a cofactor the drug does not touch — an epigenetic memory. Lineage switch would then be a correlate of escape rather than the escape mechanism.
There is no evidence base for this decision and paediatric trials are enrolling now.
What the GMP-like state implies metabolically
A GMP is the immediate upstream of the monocyte, and the monocytic programme carries a distinctive apoptotic and metabolic configuration: loss of BCL2, dependence on MCL1 and oxidative phosphorylation, and consequent venetoclax resistance 3. A blast in a GMP-like state is partway along that path.
That creates a specific and testable prediction with immediate clinical relevance: a cell moving into the GMP-like state under menin inhibition is moving toward venetoclax resistance at the same time as it moves toward menin sensitivity. Menin inhibitors are increasingly being combined with venetoclax. If this prediction holds, the combination's two arms pull in opposite directions in exactly the patients whose disease is changing state on therapy, and the apparent synergy measured in unselected models would not be present in them.
Consistent with the direction of this argument, PRC1.1-deficient KMT2A-rearranged cells that are resistant to menin inhibition show reduced monocytic gene signatures and become susceptible to BCL2 inhibition 4 — the reciprocal of the prediction.
| ontology | CL:0000557 (reference cell type against which the state is scored) |
|---|---|
| state definition | A leukaemic transcriptional/proteomic programme that projects nearest to the normal granulocyte-monocyte progenitor on a haematopoietic reference, while retaining the founding genetic lesion |
| overlap between definitions | untested; this is the principal falsifier of hyp-gmp-destination-paradox-menin |
| location | bone marrow |
two operational definitions
[
{
"method": "CUT&Tag oncoprotein occupancy over GMP gene sets",
"source_pmid": "39472576",
"context": "therapy-associated lineage switch"
},
{
"method": "quantitative proteomic clustering of primary AML",
"source_pmid": "39704147",
"context": "ex vivo bleximenib sensitivity"
}
]markers
[ "GMP gene-set enrichment score", "CD34+CD38+CD123+CD45RA+ where surface phenotype is retained", "loss of CD19 in post-switch B-ALL cases" ]
predicted drug behaviour
{
"menin_inhibitor": "most sensitive class ex vivo",
"venetoclax": "predicted to become less sensitive as the state moves toward monocytic; untested"
}atlas refs
[
{
"label": "Oncoprotein-level-dependent lineage identity across 36 MLL-r samples",
"pmid": "39472576",
"title": "MLL oncoprotein levels influence leukemia lineage identities.",
"authors": "Janssens DH, Duran M, Otto DJ, et al.",
"venue": "Nat Commun",
"year": 2024,
"title_provenance": "Title/authors/venue backfilled from PubMed esummary on 2026-09-20 because this citation stored a bare PMID with no title, which made it unscoreable by verify-citations.py. BACKFILLING IS NOT VERIFICATION: it confirms the PMID resolves to a real paper, not that this is the right paper for the claim. That still requires reading."
},
{
"label": "Proteomic GMP-like class with greatest bleximenib sensitivity",
"pmid": "39704147",
"title": "Bleximenib, the novel menin-KMT2A inhibitor JNJ-75276617, impairs long-term proliferation and immune evasion in acute myeloid leukemia.",
"authors": "Hogeling SM, Lê DM, La Rose N, et al.",
"venue": "Haematologica",
"year": 2025,
"title_provenance": "Title/authors/venue backfilled from PubMed esummary on 2026-09-20 because this citation stored a bare PMID with no title, which made it unscoreable by verify-citations.py. BACKFILLING IS NOT VERIFICATION: it confirms the PMID resolves to a real paper, not that this is the right paper for the claim. That still requires reading."
}
]- If the GMP-like state is both the most menin-sensitive and the gateway to monocytic venetoclax resistance, then the standard combination backbone is antagonistic in the subgroup it was most intended to help.
- A state defined by projection onto a normal reference can be shifted by treatment, so a state-based stratification measured at diagnosis will not describe the cell that relapses.
- ENL persistence at target loci after the oncoprotein and menin are both undetectable means the interface can be fully disengaged and the programme still run, which is the most direct published threat to the menin-interface hypothesis in this graph.
- Two research communities are using the same label for possibly different states, so a clinical decision rule built on either literature may be built on a construct the other does not share.
- Do the chromatin-defined and proteomically defined GMP-like labels identify the same cells? Scoring both on one cohort is a single-experiment question and would either establish or dissolve the central paradox in this area.
- Does a blast that has moved into a GMP-like state under menin inhibition lose BCL2 and gain MCL1 dependence, as the monocytic end of the axis does? If so, menin-plus-venetoclax is internally self-defeating in exactly the patients whose disease is changing state.
- Is ENL retention at oncoprotein-target loci present in switched samples that respond as well as those that progress? That is the only proposed explanation for why the one profiled revumenib switch did badly, and it has not been tested against a responder.
- Can a GMP-like score measured at diagnosis be used prospectively to assign menin-inhibitor therapy, or does it only describe states reached on treatment?
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
In MLL-rearranged leukaemia samples that lineage-switched in response to therapy, oncoprotein levels were reduced and the oncoproteins preferentially activated granulocyte-monocyte progenitor genes; in a sample that lineage-switched during revumenib treatment, the oncoprotein and menin were reduced to undetectable levels while ENL persisted on numerous oncoprotein-target loci including genes in the GMP-like lineage-switching programme.
MLL oncoprotein levels influence leukemia lineage identities · Janssens DH, Duran M, Otto DJ, Wu W, Xu Y, Kirkey D, et al., Henikoff S · Nature Communications 2024What this evidence cannot show
Three lineage-switching samples out of 36, and exactly one of those switched on a menin inhibitor. The GMP-like programme is defined by oncoprotein occupancy over GMP gene sets, which is a chromatin readout rather than a demonstration of GMP-like cell identity. Read: abstract only.
Evidence that argues against this (1)
- A chromatin-occupancy definition of a cell state can diverge from a functional one. The same authors' preprint framed the identical data as resistance rather than as a transitional state, titling it 'KMT2A oncoproteins induce epigenetic resistance to targeted therapies'. The same observation therefore supports two opposite clinical readings depending on framing, which is a warning against inferring prognosis from state assignment. Read: abstract only. Janssens DH et al. 2024
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
A large-scale drug screen comparing genetic alterations and quantitative proteomics with drug sensitivity in primary AML patient cells revealed that GMP-like AML display the greatest sensitivity to the menin-KMT2A inhibitor JNJ-75276617 (bleximenib); NPM1c/DNMT3A-mutant AML were also sensitive, and some NPM1-wild-type subtypes without KMT2A-MLLT3 rearrangement benefited.
Bleximenib, the novel menin-KMT2A inhibitor JNJ-75276617, impairs long-term proliferation and immune evasion in acute myeloid leukemia · Hogeling SM, Le DM, La Rose N, Kwon MC, Wierenga ATJ, Van den Heuvel FAJ, et al., Schuringa JJ · Haematologica 2025What this evidence cannot show
Ex vivo sensitivity in primary AML, predominantly adult; 'GMP-like' is a proteomic cluster defined within the study. Sensitivity in NPM1c AML means GMP-like status is not specific to KMT2A rearrangement and may be a proxy for a shared HOXA/MEIS1 dependency. Read: abstract only.
Evidence that argues against this (1)
- Menin-inhibitor sensitivity is set by chromatin context independently of differentiation state, which undercuts reading GMP-like status as a mechanistic predictor. Loss of PRC1.1 subunits (PCGF1, BCOR) confers resistance by epigenetically reactivating non-canonical menin targets including MYC, in models whose differentiation state was not the variable manipulated. A GMP-like sample carrying PRC1.1 loss would be predicted sensitive by state and resistant by chromatin. Read: abstract only. Zhou X et al. 2024
- E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Monocytic AML, the mature end of the GMP-monocyte axis, is resistant to venetoclax-azacitidine: it loses expression of the venetoclax target BCL2 and relies on MCL1 to mediate oxidative phosphorylation and survival, with monocytic subpopulations selected at relapse.
Monocytic Subclones Confer Resistance to Venetoclax-Based Therapy in Patients with Acute Myeloid Leukemia · Pei S, Pollyea DA, Gustafson A, Stevens BM, Minhajuddin M, Fu R, et al., Jordan CT · Cancer Discovery 2020What this evidence cannot show
Cited here for the monocytic end-state, not for the GMP-like state itself; the extrapolation that a GMP-like blast is partway to the monocytic metabolic configuration is an inference this entry makes, not a result the paper reports. Read: abstract only.
Evidence that argues against this (1)
- The inference fails if monocytic drug behaviour is a property of mature monocytic cells rather than of the axis. In an independent cohort, monocytic differentiation was not clinically predictive of azacitidine/venetoclax response and the predictive variable was the BCL2/BCL-xL/MCL1 ratio in leukaemic stem cells; a GMP-like blast is upstream of the monocytic stage and need not have made the BCL2-to-MCL1 switch at all. Read: abstract only. Waclawiczek A et al. 2023
- E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
KMT2A-rearranged leukaemia cells made resistant to menin inhibition by loss of PRC1.1 subunits exhibit reduced monocytic gene signatures and are susceptible to BCL2 inhibition, so that combined treatment with venetoclax overcomes the resistance.
Epigenetic regulation of noncanonical menin targets modulates menin inhibitor response in acute myeloid leukemia · Zhou X, Zhang L, Aryal S, Veasey V, Tajik A, Restelli C, et al., Lu R · Blood 2024What this evidence cannot show
Engineered PRC1.1 loss in human and mouse KMT2A-rearranged AML models; the coupling between reduced monocytic signature and restored BCL2 sensitivity is correlative within those models, and no patient cohort has tested whether PRC1.1-deficient KMT2A-rearranged AML is less monocytic and more venetoclax-sensitive. Read: abstract only.
Evidence that argues against this (1)
- The reciprocal prediction this entry draws — that moving toward a GMP-like and then monocytic state trades menin sensitivity for venetoclax resistance — is not what the lineage-switch literature observes clinically. Switched cases do badly on everything, not selectively badly on venetoclax, and the single profiled revumenib switch showed persistence of ENL at target loci rather than a metabolic reconfiguration. The trade-off framing may be too tidy. Read: abstract only. Janssens DH et al. 2024
the third exit: reduced oncoprotein preferentially activates GMP genes, and in a revumenib lineage-switch sample ENL persisted on the GMP-like programme after menin became undetectable
- related to →Granulocyte-monocyte progenitor (GMP) — the committed cell an MLL fusion can turn back into a stem cell
the state is scored against this normal cell type, and is not the same thing
defines the state the hypothesis is about and names the definitional escape route
most sensitive proteomic class ex vivo
the one profiled patient who switched into this state on revumenib retained ENL at target loci
the state entered at switch
predicted trajectory toward the monocytic metabolic configuration and venetoclax resistance
the GMP-like programme is not detectable at promoter or interval scale in this deposit
the AML state of the same patient
the relapse clone activates the GMP-like program the authors describe
ENL holds its signal at the GMP-like program peaks
the GMP-like chromatin class is not fusion-enriched outside the pair that defined it
records the measured support for that state's chromatin operational definition
CANDIDATE PARALLEL ONLY, not a finding: Daxx loss shifts normal progenitors B-to-myeloid, and therapy-driven lineage switch in KMT2A-rearranged B-ALL lands in a GMP-like state. Nothing connects them experimentally
- ← related toRELA (p65) — menin binds it and represses it, and it is required to keep MLL oncoproteins on chromatin
the LSC compartment in the murine MLL-AF9 system this result comes from is GMP-like; the RELA requirement is recorded against that state rather than against a lineage