Oncofusion

start / graph / resistance-prc1-1-loss-menin-inhibitor-nup98

Resistancemoderate confidence 3 of 5: One strong study, or several smaller consistent ones.Qn-nup98-family

Evicting the fusion is not enough: PRC1.1 has to arrive, and losing it causes menin-inhibitor resistance in vivo

also called PRC1.1 menin inhibitor resistance · NUP98 PRC1.1 · KDM2B menin resistance

Menin drugs push the cancer protein off the genes it is holding open, but that alone does not switch those genes off. A repressive complex has to move in afterwards — and leukaemia cells that have lost that complex keep the genes on and survive the drug.

The same thing, for a scientist

The standard account of menin inhibition is eviction: the drug detaches the oncoprotein from chromatin and the leukaemic programme collapses. The follow-up work shows that account is **incomplete for both fusion families**. NUP98 fusion proteins and KMT2A–menin **antagonise PRC1.1**, and eviction of the fusion–menin–KMT2A complex is **not sufficient** to silence target genes; silencing requires **PRC1.1 accumulation** at those loci, and **PRC1.1 loss causes menin-inhibitor resistance in vivo** [E1]. This changes the resistance landscape in a way no current assay looks for. The escape routes already catalogued in this graph are **MEN1 binding-site mutation** — 13% of patients in the one trial that reports genotyped resistance [E2] — and, in the only published NUP98 complete remission, **RAS-pathway co-mutation with no MEN1 mutation at all** [E3]. PRC1.1 loss is a third route, and it is invisible to a menin-pocket sequencing panel, invisible to a RAS panel, and would present as primary rather than acquired resistance if the complex is already absent at diagnosis. It also supplies a fourth: **MEIS1 re-expression abrogates sensitivity** downstream of the tether entirely [E4].

Go deeper
Four escape routes, only one of which anyone screens for
routemechanismdetectable byevidence
MEN1 binding-site mutationdrug cannot bind menintargeted MEN1 sequencing13% of patients in SAVE 2
RAS-pathway co-mutationbypass downstream of the programmemyeloid NGS panelthe one published NUP98 CR relapsed this way 3
PRC1.1 losseviction happens but genes stay onnothing in routine usein vivo resistance 1
MEIS1 re-expressionbypass downstream of the tetherRNA, not DNAsensitivity abrogated in models 4

Only the first is a mutation in the drug target and only the first is routinely looked for. Two of the four are transcriptional states rather than mutations, which means a DNA-based resistance panel would report 'no resistance mechanism identified' in a patient whose resistance mechanism is perfectly well defined.

For a NUP98-rearranged population this matters more than for KMT2A-rearranged disease, because the observed remissions are short (median 5.9 months in the only reported subgroup, trial-save-revumenib-nup98-subgroup) and the reason is unknown. If a substantial share of that shortness is PRC1.1-state-driven rather than mutation-driven, the resistance story for this family is being told with the wrong instrument.

Why 'eviction is not sufficient' is the important half

It reframes the mechanism of the entire drug class. Menin inhibition has been described as removing the oncoprotein; this work says removal creates an opportunity for silencing that a repressive complex must then take 1.

Two consequences follow that are specific to this beat:

  1. The borrowed-cofactor model gets more fragile. In NUP98-rearranged disease the drug removes a cofactor the fusion borrowed (mechanism-nup98-fusion-menin-mll1-dependency), and now that removal also depends on a second, separate event happening afterwards. Two sequential requirements are a weaker therapeutic proposition than one.
  2. Combination logic changes. If the limiting step is PRC1.1 recruitment rather than fusion eviction, then the productive partner for a menin inhibitor is something that promotes Polycomb-mediated silencing, not a second inhibitor of the same axis. Nothing in the current NUP98 combination literature — CDK6 and FLT3 inhibitors — acts on that step.

The entry does not claim PRC1.1 loss occurs in patients. It has been shown in vivo in models 1 and has not, as far as this pass found, been looked for in a patient sample.

Facts and figures
kindresistance
severityhigh for NUP98-rearranged disease, where observed remissions are the shortest of any menin-susceptible genotype
mechanismEviction of the fusion-menin-KMT2A complex is not sufficient to silence target genes; PRC1.1 must accumulate. Loss of PRC1.1 leaves target genes active despite eviction and causes menin-inhibitor resistance in vivo.
why dna panels miss itTwo of the four routes are transcriptional or complex-composition states, not mutations, so a DNA-based resistance panel reports no mechanism identified.
combination implicationIf the limiting step is PRC1.1 recruitment rather than fusion eviction, the productive partner for a menin inhibitor promotes Polycomb-mediated silencing; the current NUP98 combination literature (CDK6, FLT3 inhibitors) acts on neither step.
mitigationAssay PRC1.1 component integrity and MEIS1 expression at baseline and at progression in any NUP98-rearranged menin-inhibitor cohort; neither is currently collected.
affects
[
  "drug-revumenib",
  "drug-ziftomenib",
  "drug-bleximenib",
  "drug-icovamenib"
]
escape routes catalogued
[
  {
    "route": "MEN1 binding-site mutation",
    "detectable_by": "targeted MEN1 sequencing",
    "clinical_evidence": "13% of SAVE patients"
  },
  {
    "route": "RAS-pathway co-mutation",
    "detectable_by": "myeloid NGS panel",
    "clinical_evidence": "the single published NUP98 complete remission"
  },
  {
    "route": "PRC1.1 loss",
    "detectable_by": "nothing in routine clinical use",
    "clinical_evidence": "none; in vivo model only"
  },
  {
    "route": "MEIS1 re-expression",
    "detectable_by": "RNA, not DNA",
    "clinical_evidence": "none; model only"
  }
]
How this could bite us back
  • A resistance-monitoring programme built around MEN1 pocket sequencing will report 'no mechanism identified' for most failures, and that silence will be misread as idiopathic relapse rather than as a mechanism nobody assayed.
  • If eviction requires a second, separate silencing event, then menin inhibition in NUP98-rearranged disease has two sequential points of failure rather than one, which is a structural reason to expect shorter remissions than in KMT2A-rearranged disease - exactly what is observed.
  • Promoting Polycomb silencing as a combination strategy would repress a great deal more than the fusion's target genes, and Polycomb loss-of-function is itself leukaemogenic in other contexts, so the obvious combination partner carries an obvious risk.
Still open
  1. Does PRC1.1 loss occur in patients relapsing on a menin inhibitor? Nobody has looked, and the assay would be a straightforward addition to an existing correlative programme.
  2. Are MEIS1 re-expression and PRC1.1 loss the same route seen from two ends? MEIS1 is Polycomb-regulated, so the two may not be independent.
  3. What fraction of the 87% of SAVE patients without a binding-site mutation fail by a transcriptional mechanism? The samples exist; the assay does not.
  4. Is there a drug that promotes PRC1.1 recruitment, and would it be a better partner for a menin inhibitor than a kinase inhibitor?
Every source behind this entry (4)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    NUP98 fusion proteins and KMT2A-MENIN antagonise PRC1.1 to drive gene expression in AML; eviction of the fusion-menin-KMT2A complex is not sufficient to silence target genes, silencing requires PRC1.1 accumulation, and PRC1.1 loss causes menin-inhibitor resistance in vivo.

    PMID 39475509DOI
    What this evidence cannot show

    Mouse and cell-line models from the same group that established the menin dependency, so this is an internal refinement rather than independent confirmation. 'PRC1.1 loss causes resistance in vivo' is a genetic manipulation result; no patient sample has been shown to lose PRC1.1 under menin-inhibitor treatment. Details were obtained in a delegated research pass in this session and the primary figures were not read.

    Evidence that argues against this (1)
    • The prospective trial that genotyped resistance found emergent MEN1 binding-site mutations in 13% of patients and did not report PRC1.1-related events, so the clinically observed resistance so far is on-target mutation rather than the mechanism described here - which may mean PRC1.1 loss is a model phenomenon, or simply that nobody assayed for it. Issa GC et al. 2026
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Emergent mutations in the menin-binding site occurred in 13% of patients treated with a revumenib-containing triplet in a cohort that included 21% NUP98-rearranged patients.

    All-Oral Combination of Revumenib, Decitabine, and Venetoclax for Relapsed or Refractory AML (SAVE) · Issa GC, et al. · Journal of Clinical Oncology 44(22):2110-2120 2026
    PMID 42272166DOI
    What this evidence cannot show

    The 13% is not broken down by genotype in the abstract, so whether NUP98-rearranged patients acquire binding-site mutations at the same rate as KMT2A-rearranged patients is unknown. Patients received three drugs, so selection pressure on the menin interface was applied alongside two other agents.

    Evidence that argues against this (1)
    • 13% leaves 87% of patients whose eventual failure is not explained by a binding-site mutation, which is consistent with most resistance in this setting running through mechanisms that a MEN1 sequencing panel does not see. Heikamp EB et al. 2024
  3. E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    The only published complete remission of NUP98-rearranged AML on a menin inhibitor ended in progression with emergent CEBPA, KRAS G13D, NRAS G13D, WT1 and RUNX1 variants and no MEN1 binding-site mutation.

    PMID 39633542DOI
    What this evidence cannot show

    A single patient. Absence of a reported MEN1 mutation depends on the sequencing panel used, which is not described in detail. Whether the RAS-pathway subclones were acquired or selected from below the detection limit at baseline cannot be determined from one time point.

    Evidence that argues against this (1)
    • One patient's escape route cannot be generalised, and RAS-pathway subclones emerge under many selective pressures in AML including chemotherapy, so this may be a general relapse mechanism rather than a menin-inhibitor-specific one. Issa GC et al. 2026
  4. E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Retroviral Meis1 overexpression or CRISPR activation at the endogenous Meis1 promoter abrogates menin-inhibitor sensitivity in NUP98-fusion leukaemia cells.

    The menin-MLL1 interaction is a molecular dependency in NUP98-rearranged AML · Heikamp EB, et al. · Blood 139(6):894-906 2022
    PMID 34582559DOI
    What this evidence cannot show

    Enforced overexpression and CRISPR activation both drive MEIS1 above physiological levels, so this shows that high MEIS1 is sufficient for resistance, not that MEIS1 elevation is a route cells actually take under drug pressure. Read via a delegated full-text pass in this session.

    Evidence that argues against this (1)
    • No patient has been reported to relapse on a menin inhibitor with MEIS1 upregulation as the identified mechanism, so this remains a model-derived possibility. It is also not separable from PRC1.1 loss, since MEIS1 is a Polycomb-regulated locus and losing PRC1.1 could produce MEIS1 re-expression as its readout rather than as an independent route. Heikamp EB et al. 2024
Connected to