Oncofusion

start / graph / fusion-zmynd11-mbtd1

Fusionstrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.Q121-zmynd11-mbtd1-the-architecture-from-other-parts

ZMYND11::MBTD1, t(10;17)(p15;q21) - a chromatin reader welded to the NuA4/TIP60 acetyltransferase, maintaining Hoxa/Meis1/Myb/Myc/Sox4 in AML: the KMT2A-fusion architecture built from different parts

also called ZMYND11-MBTD1 · ZMYND11/MBTD1 · ZM · BS69-MBTD1 · t(10;17)(p15;q21) · t(10;17)(p15.3;q21.33)

A rare leukaemia swap joins the front half of a gene that reads a chemical tag on packaged DNA to a gene that plugs into a machine that loosens that packaging, and the joined protein drags the loosening machine to the wrong genes and holds them switched on. The genes it holds on are the same ones a KMT2A fusion holds on - HOXA, MEIS1, MYB, MYC - which makes it the closest thing in another disease to the fusion this project is about.

The same thing, for a scientist

**ZMYND11::MBTD1 is the clearest published example of the KMT2A-fusion architecture assembled from entirely different parts.** The recurrent t(10;17)(p15;q21) fuses ZMYND11 exons 1-12 (10p15.3) to MBTD1 exons 3-17 (17q21.33), producing a 1,054-amino-acid in-frame chimera; the reciprocal product is predicted to lack a productive start codon [E4]. The retained ZMYND11 portion carries the PHD finger, bromodomain and the PWWP module that reads H3K36me3 on the H3.3 variant; the MBTD1 portion carries the FCS zinc finger and all four MBT repeats, and is the part that docks the fusion into the NuA4/TIP60 histone acetyltransferase complex through EPC1 [E1][E2][E7]. **In primary murine haematopoietic stem/progenitor cells and in transplanted mice**, the fusion confers indefinite ex vivo self-renewal and causes AML with an average latency of 154 days in 8 of 9 animals; it binds 16,114 sites, predominantly promoter-proximal regions and gene bodies overlapping H3K36me3 and devoid of H3K27me3, recruits NuA4/Tip60, builds super-enhancers at proto-oncogenes, and maintains high expression of *Hoxa*, *Meis1*, *Myb*, *Myc* and *Sox4*. Both the Tip60-interaction surface and the H3.3K36me3-binding PWWP are required for transformation, and the fusion-driven AML is sensitive to the BET inhibitor I-BET151 in vitro and in vivo [E1]. **Clinically it is a distinct and poorly-behaved entity.** Across twelve assembled cases the blasts are uniformly CD7+, CD33+, CD56-bright and CD117+, HLA-DR-negative in 8 of 10 and MPO-negative in 9 of 9, with minimal differentiation (FAB M0); the karyotype is simple or non-complex and co-occurring mutations are rare; most adults relapse early, all four transplanted patients relapsed, and salvage responses were poor. Two paediatric patients achieved long-term remission [E3]. A 2026 case report argues the immunophenotype is that of myeloid/NK cell precursor leukaemia and that earlier cases were misclassified [E6]. **The point of this entry for the Iqbal Lab** is the architecture, not the disease: a chromatin-reading N-terminal module fused to a complex-recruiting C-terminal module, mis-targeting a coactivator complex to a locus it then holds open, producing HOXA/MEIS1/MYB/MYC - through NuA4/TIP60 rather than through SEC/DOT1L, and with a completely different reader chemistry. See `analogy-zmynd11-mbtd1-is-the-kmt2a-architecture-from-other-parts` for the point-by-point mapping and for where the analogy breaks.

Go deeper
What the translocation makes, resolved at exon level

ZMYND11 has 14 exons and MBTD1 has 17 3. The fusion joins ZMYND11 exons 1-12 (less often exons 1-11) to MBTD1 exons 3-17, in frame, encoding a protein of 1,054 amino acids; the reciprocal MBTD1::ZMYND11 product is predicted to lack a productive start codon and is presumed non-functional 43. Optical genome mapping in one of the 2026 MD Anderson cases refined the karyotype to t(10;17)(p15.3;q21.33)(244532;51223617) and independently detected the fusion 3.

The construct used in the founding functional study is informative about domain content, because it states the residue boundaries the karyotype does not: the ZM cDNA was assembled from ZMYND11 amino acids 1-409 (NM_006624) joined to a 16-residue 5'-UTR-derived segment plus the entire MBTD1 open reading frame, amino acids 1-628 (NM_017643) 1. That is 409 + 16 + 628 = 1,053 residues, one short of the 1,054 reported from the patient transcript 4 - a construct-versus-transcript difference of one residue that neither paper comments on and which this entry records rather than resolves.

Domain arithmetic, stated as a derivation and not as a published fact. Taking the construct boundaries above, MBTD1 residue n is fusion residue n + 425. UniProt Q05BQ5 places the four MBT repeats at 141-245, 253-350, 351-456 and 464-560 and the FCS-type zinc finger at 45-80 11; on a 1-409 ZMYND11 segment, UniProt Q15326's PHD (100-148), bromodomain (149-255) and PWWP (280-331) are retained and the MYND zinc finger (563-598) is lost 11. Both readings agree with the Haematologica group's statement that the fusion protein retains the PHD, bromodomain and PWWP of ZMYND11 3. The offset arithmetic is mine; the domain retention is published.

Mechanism: which modules are required, and the one that is not

Systematic mutagenesis in murine HSPCs separates the fusion into modules that matter and modules that do not 1:

modulemutation testedtransformation
ZMYND11 PWWP (H3.3K36me3 cage)W294A; F310Aabolished
ZMYND11 PHDdeletionabolished (yet only mildly impairs H3K36me3 binding in vitro - the authors call this a possible H3K36me3-binding-independent function)
MBTD1 MBT repeats (TIP60 docking)delta-MBTx4abolished
MBTD1 EPC1 contactA661D/L684D/L688D (ALL>DDD, fusion numbering)abolished
MBTD1 FCS zinc fingerdeletionretained (reduced, not lost)
MBTD1 MBT4 H4K20me1/2 cageW954A; F951A/W954A/Y958A (FWY>AAA)RETAINED - all transformed, all still bound Tip60

The last row is the load-bearing negative. The fusion carries an intact methyl-lysine reader for H4K20me1/me2 and does not need it. What it needs is the scaffolding half of MBTD1 - the surface that EPC1 grips - and the reading half of ZMYND11. See mark-h4k20me1 for what that costs the H4K20 story and interface-mbtd1-epc1-nua4-tip60 for the structure of the contact.

An artificial ZMYND11-TIP60 chimera produces an equivalent transformation phenotype 1. That is the cleanest statement of modularity available in any fusion family: the identity of the 3' partner is dispensable so long as it delivers the acetyltransferase.

Where the two mechanism papers disagree, and how far

Li et al. (murine HSPC/AML, ChIP-seq of an HA-tagged fusion) report ZM peaks predominantly promoter-proximal and extending into gene bodies, overlapping H3K36me3 and not H3K27me3, with super-enhancers called at proto-oncogenes, and frame the mechanism as recruitment of NuA4/Tip60 to cis-regulatory elements sustaining Hoxa, Meis1, Myb, Myc, Sox4 1. Devoucoux et al. (human cells, biochemistry plus genomics, side-by-side with each wild-type partner) report the fusion stably incorporated into the endogenous NuA4/TIP60 complex and mislocalised onto the BODY of genes normally bound by ZMYND11, with increased chromatin acetylation, altered transcription most notably at MYC, and altered alternative splicing; their headline gene is MYC and their headline phenotype is Myc-driven pluripotency plus HSPC self-renewal 2.

These are not flatly contradictory - Li's own peak distribution includes gene bodies, and both report NuA4/TIP60 mistargeting and self-renewal. They differ on where the decisive mistargeting happens (regulatory elements versus transcription units) and on which target carries the phenotype (a HOXA/MEIS1 stem programme versus MYC). This graph records both, at the same confidence, and does not pick. The 2026 clinical review cites both and merges them into a single sentence about "ZMYND11-bound loci" 3, which is the safe reading and also the one that loses the disagreement.

Why it matters here: the KMT2A::AFF1 literature has the identical unresolved axis - promoter-proximal menin/LEDGF-directed targeting versus gene-body elongation - and here it appears in a fusion with no menin, no CXXC domain and no DOT1L.

Clinical picture, prognosis, and how it is detected

Twelve cases have been assembled 3: six confirmed in the literature before 2026 54789, one immunophenotypically suggestive but unconfirmed, and five new MD Anderson cases; a further 2026 paediatric-oncology case report adds one more and reviews five 6. Immunophenotype across the assembled series: CD7 12/12, CD33 12/12, CD56 (bright) 11/11, CD117 11/11, HLA-DR negative 8/10, MPO negative 9/9 3. Blasts are primitive and agranular; in two cases they formed cohesive clusters mimicking a solid tumour in the marrow 38.

Outcomes in adults are poor. Of the five MD Anderson patients: one achieved CR, relapsed at 23 months and died at 34 months; one had persistent disease on venetoclax/cladribine/LDAC; one relapsed three months after transplant and died four months later; one relapsed after each of two transplants and died at 57 months; one had persistent disease on azacitidine and died at 11 months. All four transplanted patients across the assembled cohort relapsed. Two paediatric patients achieved long-term remission 3. The authors propose recognition as a separate entity under "other rare recurrent genetic abnormalities".

Detection is the practical problem. The translocation is visible on conventional karyotype, but neither gene is typically included in targeted RNA-sequencing fusion panels and FISH probes are not routinely available; published detections have used BAC FISH probes RP11-10D13 and RP11-379D19, RT-PCR, RNA sequencing, and - in the 2026 series - optical genome mapping 3. A fusion that a standard panel cannot see is a fusion whose true incidence is unknown, and every frequency statement in this entry is therefore a floor.

Cooperating lesions: the mouse and the patients point opposite ways

In mice, ZM alone causes AML only after long latency (mean 154 days) and with incomplete penetrance; co-expressed NrasG12D shortens latency to 41 days, and whole-exome sequencing of a ZM-alone murine AML recovered a spontaneously acquired activating Ptpn11 S502L mutation. The authors conclude that activated RAS signalling is a cooperating event 1.

In patients, the 2026 series found simple or non-complex karyotypes, one TET2 mutation among five cases, and no mutations at all in four, and concluded that ZMYND11::MBTD1 "is a very potent leukemogenic driver ... and does not require additional co-operating mutations" 3.

These two inferences are in tension and both are in print. They can be reconciled - a targeted myeloid panel does not see everything, mouse latency is not human latency, and one TET2 mutation is not zero - but the reconciliation is an argument, not a measurement. Recorded here as an open disagreement rather than resolved, because it is exactly the kind of statement a later cycle would otherwise quote from one side.

Facts and figures
gene5ZMYND11
gene3MBTD1
cytobandt(10;17)(p15;q21); refined by optical genome mapping to t(10;17)(p15.3;q21.33)
gene5 cytoband10p15.3
gene3 cytoband17q21.33
uniprot5Q15326
uniprot3Q05BQ5
length aa5602
length aa3628
ensembl5ENSG00000015171
ensembl3ENSG00000011258
fusion protein length AA1054
detection methodConventional G-banded karyotype detects t(10;17)(p15;q21). Fusion confirmation has used BAC FISH probes RP11-10D13 and RP11-379D19, RT-PCR with Sanger sequencing, nested PCR, RNA sequencing, and optical genome mapping. NEITHER GENE IS TYPICALLY ON A TARGETED RNA-SEQ FUSION PANEL and clinical FISH probes are not routinely available, so ascertainment is by karyotype plus a deliberate follow-up.
immunophenotypeCD7+ 12/12, CD33+ 12/12, CD56-bright 11/11, CD117+ 11/11, CD34 variably positive; HLA-DR negative 8/10 (minimally expressed in the other 2); MPO negative 9/9 by flow and negative or <3% by cytochemistry. CD4 negative and CD123 not bright, which is what separates it from blastic plasmacytoid dendritic cell neoplasm.
prognosisPoor in adults. Most patients relapse early after initial remission; all four transplanted patients in the assembled cohort relapsed; salvage responses were generally poor. Reported survivals in the 2026 series: 11, 34 and 57 months in three patients who died, one lost to follow-up with persistent disease, one relapsed 3 months post-transplant and dead 4 months later. TWO PAEDIATRIC PATIENTS ACHIEVED LONG-TERM REMISSION, so the adult qualifier in the source title is load-bearing. No survival curve, hazard ratio or confidence interval exists for this entity.
co occurring lesionsSimple or non-complex karyotype is the rule, including at relapse. Among five MD Anderson cases: one TET2 mutation, one del(16q), one acquired inv(2)(p11.2q21) at second relapse, and four with no mutations identified. In mice, NrasG12D accelerates the fusion's AML from 154 to 41 days and a Ptpn11 S502L activating mutation was acquired spontaneously in a ZM-alone AML.
mechanism summaryReader-to-coactivator mistargeting. The retained ZMYND11 PWWP reads H3.3K36me3 on transcribed chromatin; the retained MBTD1 MBT repeats dock the whole chimera into NuA4/TIP60 through EPC1; the result is acetyltransferase activity delivered to ZMYND11-bound loci, an active chromatin state refractory to PRC2/H3K27me3, and maintained expression of Hoxa, Meis1, Myb, Myc and Sox4.
therapeutic leadBET/BRD4 inhibition. I-BET151 blocked proliferation, elevated cell death and triggered terminal myeloid differentiation of murine ZM-transformed AML cells, and extended survival in vivo. BRD4 acts downstream of the fusion's own acetylation output, which is why an inhibitor of the READER of the mark the fusion deposits works where an inhibitor of the fusion does not exist.
shown for fusionThis entry IS the fusion. Its relationship to the index fusion is analogical only - nothing in this literature has been tested in KMT2A::AFF1 or in any KMT2A-rearranged system, and no co-occurrence of ZMYND11::MBTD1 with a KMT2A rearrangement has been reported.
cell typeHuman: AML blasts with minimal differentiation, CD7+/CD56-bright/HLA-DR-/MPO-, in adults and children. Mouse: primary bone-marrow haematopoietic stem/progenitor cells (the transforming target), mouse embryonic stem cells in the Devoucoux arm.
cell stateSelf-renewal with a differentiation block. The fusion confers indefinite serial replating ex vivo, blocks terminal myeloid differentiation, and its PWWP- and TIP60-defective mutants fail to block differentiation - so the block is the readout of the mechanism and not an incidental phenotype. BET inhibition releases the block into terminal differentiation.
subcellular localizationNuclear and chromatin-associated by inference from the ChIP-seq and from both parent proteins' UniProt localisations; NO dedicated imaging of the fusion protein was found and none is asserted.
subnuclear localizationNOT MEASURED. No speckle, PML-body, nucleolar or nuclear-matrix assignment exists for ZMYND11::MBTD1. The chromatin-compartment assignment (promoter-proximal plus H3K36me3-marked gene bodies, super-enhancers at proto-oncogenes) is the only spatial statement available and it is a genomic, not a morphological, one.
epigenetic readoutReads H3.3K36me3 through the retained ZMYND11 PWWP; writes histone H4 and H2A acetylation through the recruited NuA4/TIP60 catalytic subunit KAT5/TIP60; the chromatin state at its targets is acetylation-rich, H3K36me3-positive and H3K27me3-depleted. It carries but does not require an H4K20me1/me2 reader (MBTD1 MBT4).
absence classnever-measured
absence class noteThe absence this entry records is the ANALOGY arm, not the fusion's own biology, which is measured. Nothing has ever been measured jointly on ZMYND11::MBTD1 and a KMT2A fusion: no side-by-side occupancy comparison, no shared-target-set analysis, no test of whether an artificial reader swap into a KMT2A backbone reproduces the programme. That needs new material (constructs, cells, ChIP), so it is never-measured rather than measured-not-analysed. A weaker sub-arm IS measured-not-analysed - the two published ZM genomic datasets could be compared against this project's KMT2A::AFF1 occupancy data as pure compute - but no accession is named here because the GEO series for PMID 33594072 and PMID 35705031 was NOT looked up this session, and naming one from memory is exactly the failure this field exists to prevent.
salvaged cycle 82Q121 was killed by the account session limit after writing this entry's evidence but before its counter_evidence. The coordinator added 6 reasoned methodological counter-arguments during salvage, each derived from the claim and its citation type and each labelled as such. NO counter-citation was fetched and none is asserted - inventing one would have been the fabrication this project has a rule against. Every note also states what a stronger, literature-sourced counter would be, so the gap is recorded rather than closed.
diseases
[
  "acute myeloid leukaemia with minimal differentiation (FAB M0)",
  "myeloid/NK cell precursor leukaemia (proposed reclassification of the same immunophenotype)",
  "therapy-related AML (one reported case)"
]
breakpoints
[
  {
    "id": "bp-zmynd11-exon12-mbtd1-exon3",
    "gene5_exon_last": 12,
    "gene3_exon_first": 3,
    "population": "the commonest reported junction; ZMYND11 exons 1-11 reported in at least one case",
    "domains_retained_5": [
      "PHD-type zinc finger",
      "Bromo (bromodomain)",
      "PWWP (H3.3K36me3 reader)"
    ],
    "domains_lost_5": [
      "MYND-type zinc finger",
      "adenovirus E1A-interaction region"
    ],
    "domains_retained_3": [
      "FCS-type zinc finger",
      "MBT repeat 1",
      "MBT repeat 2",
      "MBT repeat 3",
      "MBT repeat 4 (H4K20me1/2 cage)"
    ],
    "domains_lost_3": [],
    "phenotype_notes": "In-frame, 1,054 aa. The reciprocal MBTD1::ZMYND11 product is predicted to lack a productive start codon. The functional construct used by Li et al. was ZMYND11 aa 1-409 (NM_006624) + 16 UTR-derived residues + MBTD1 aa 1-628 (NM_017643) = 1,053 aa, one short of the transcript-derived 1,054; the discrepancy is unexplained in either source."
  },
  {
    "id": "bp-zmynd11-mbtd1-ogm-grch38",
    "population": "one MD Anderson case, optical genome mapping",
    "phenotype_notes": "t(10;17)(p15.3;q21.33)(244532;51223617) - the two coordinates quoted in the paper. THE ASSEMBLY IS NOT STATED in the paper and was not confirmed; 51,223,617 falls inside the GRCh38 MBTD1 gene body (51,177,419-51,261,114) and 244,532 falls inside the GRCh38 ZMYND11 gene body (130,088-254,771), which is consistent with GRCh38 but is a consistency check and not a statement from the source."
  }
]
frequency
{
  "reported_cases_total": 12,
  "reported_cases_note": "Twelve assembled by the 2026 Haematologica series (5 new + 6 confirmed in the literature + 1 immunophenotypically suggestive but unconfirmed), plus one further 2026 paediatric case report. No population-based incidence figure exists for this fusion in any registry, and because standard fusion panels miss it every count is a floor."
}
required modules
{
  "ZMYND11 PWWP (H3.3K36me3 cage)": "REQUIRED - W294A and F310A each abolish transformation of murine HSPCs",
  "ZMYND11 PHD": "REQUIRED - deletion abolishes transformation, though it only mildly impairs H3K36me3 binding in vitro",
  "MBTD1 MBT repeats (TIP60 docking)": "REQUIRED - delta-MBTx4 abolishes both Tip60 binding and transformation",
  "MBTD1-EPC1 contact": "REQUIRED - A661D/L684D/L688D (fusion numbering) abolishes transformation",
  "MBTD1 FCS zinc finger": "NOT REQUIRED - deletion still transforms, with a reduced phenotype",
  "MBTD1 MBT4 H4K20me1/2 cage": "NOT REQUIRED - W954A and F951A/W954A/Y958A (fusion numbering) both bind Tip60 and both transform"
}
analogous fusion families
[
  "KMT2A::AFF1 and the KMT2A fusion family - same output genes (HOXA, MEIS1, MYB, MYC), same reader-plus-complex architecture, different reader (PWWP/H3.3K36me3 versus CXXC/unmethylated CpG), different complex (NuA4/TIP60 versus SEC and DOT1L), different disease lineage",
  "NUP98 fusions - a 5' module that provides chromatin/condensate targeting fused to a partner that provides the effector, also maintaining HOXA",
  "SS18::SSX and EWSR1::FLI1 - fusions that redirect a chromatin-remodelling or coactivator complex to new sites; already compared in this graph",
  "MBTD1::CXorf67 in low-grade endometrial stromal sarcoma - the SAME gene with the polarity reversed, MBTD1 as the 5' partner"
]
How this could bite us back
  • BET inhibition is the published lead and BET inhibitors have repeatedly under-delivered in the clinic. The paper's own discussion says single-agent BET inhibitors show only limited efficacy in trials and points to bivalent inhibitors, PROTAC degraders and combinations as the route forward. A lead that works in a murine model of a twelve-case disease, with a drug class that has already disappointed, is a hypothesis about biology and not a treatment plan.
  • The drug acts DOWNSTREAM of the lesion, on BRD4 reading the acetylation the fusion causes. That means every normal cell that depends on BRD4 reading acetylation is in the blast radius - and unlike a menin inhibitor, there is no interface here that exists only in the diseased cell. The selectivity argument that makes the menin-KMT2A interface attractive has no counterpart in this fusion.
  • MBTD1's normal job is to hold adult haematopoietic stem cells quiescent: conditional Mbtd1 knockout increases HSC and progenitor numbers, drives hyperactive cycling and breaks the response to exogenous stress, through direct binding at the FoxO3a promoter. Any agent that disables the MBTD1-EPC1 arm of NuA4/TIP60 to break the fusion would hit that function in normal HSCs. See gene-mbtd1.
  • KAT5/TIP60 is a near-pan-essential dependency in DepMap (mean gene effect -0.607 across 1,258 lines, 70.2% below -0.5, release 26/06), so 'inhibit the hijacked acetyltransferase' is an essential-function toxicity proposal, not a targeted one. EP400 is worse (mean -0.932, 94.3% below -0.5). The druggable node in this pathway is the reader downstream, which is precisely why the published lead is BET inhibition.
  • Two paediatric patients achieved long-term remission with conventional therapy while adults did not. If that survives a larger series, the entity's real problem may be age-dependent tolerance of intensive therapy rather than intrinsic drug resistance - and a novel-agent programme aimed at the fusion would be answering the wrong question for half the patients.
Still open
  1. Do ZMYND11::MBTD1 and KMT2A::AFF1 occupy the same HOXA sub-elements, or the same genes by different routes? Both deposited datasets are public and this project already holds binned KMT2A::AFF1 occupancy; the comparison is compute, and it is the single cheapest test of whether 'the same programme' means the same coordinates or only the same gene list.
  2. Is the reader module interchangeable? Li et al. showed an artificial ZMYND11-TIP60 chimera transforms equivalently, which tests the EFFECTOR half. The untested half is the opposite swap: does a CXXC domain in place of the PWWP, or a PWWP in place of KMT2A's CXXC, retarget the programme? That experiment would say whether the address or the effector sets the target list.
  3. Why does PHD deletion abolish leukaemogenesis when it barely affects H3K36me3 binding? The authors flag this themselves. If the PHD is doing something other than assisting the cage, the 'PWWP-mediated chromatin association' headline is incomplete and the analogy to KMT2A's CXXC is narrower than it looks.
  4. Does the fusion's chromatin state exclude PRC2 actively or passively? Li et al. propose the H3K36me3-plus-acetylation state is refractory to PRC2 invasion. This graph holds a PRC2 arm and an H3K27me3 entry; whether the same exclusion operates at KMT2A fusion targets is asked nowhere.
  5. Is ZMYND11::MBTD1 AML menin-independent? Nobody has asked. It has no CXXC domain and no menin-binding motif, so the expectation is yes - but a HOXA/MEIS1 programme that is menin-independent would be the cleanest available demonstration that the programme is not owned by the menin interface, and the reagents are commercial.
Every source behind this entry (11)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In primary murine haematopoietic stem/progenitor cells and syngeneic transplant recipients, ZMYND11-MBTD1 confers indefinite ex vivo self-renewal and causes AML (mean latency 154 days, 8/9 mice); it binds 16,114 sites concentrated at promoter-proximal regions and gene bodies overlapping H3K36me3 and devoid of H3K27me3, recruits the NuA4/Tip60 acetyltransferase complex and establishes super-enhancers at proto-oncogenes, maintains high expression of Hoxa, Meis1, Myb, Myc and Sox4, requires both the Tip60-interaction surface and an intact H3.3K36me3-binding PWWP domain for transformation, and yields AML sensitive to the BET inhibitor I-BET151 in vitro and in vivo.

    PMID 33594072DOI
    What this evidence cannot show

    MURINE THROUGHOUT for the functional arm: primary mouse HSPCs, retroviral overexpression of a human ZM cDNA, syngeneic transplant. No human ZMYND11::MBTD1 leukaemia cell line exists and none was used. The ChIP-seq is on an HA- or GFP-tagged construct, not an endogenous fusion allele. ZM alone gave incomplete penetrance and long latency, so the transforming phenotype is not a clean single-hit result. The I-BET151 efficacy is in murine ZM-transformed AML, not in a patient. Full text read this session via PMC7886901.

    Evidence that argues against this (2)
    • Devoucoux et al. locate the decisive mistargeting on the BODY of genes rather than at cis-regulatory elements, and identify MYC rather than the Hoxa/Meis1 stem programme as the headline target - a different emphasis from the same fusion in human cells with the endogenous complex. The two are reconcilable but the divergence is real and is not discussed in either paper. Devoucoux M et al. 2022
    • The paper's own discussion concedes that PHD deletion abolishes leukaemogenesis while only mildly impairing H3K36me3 binding in vitro, which means the clean 'PWWP-mediated chromatin association' headline does not account for the whole N-terminal requirement. The authors state this as an unsolved question. Li J et al. 2021
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In human cells, ZMYND11-MBTD1 is stably incorporated into the endogenous NuA4/TIP60 complex and mislocalises it onto the body of genes normally bound by ZMYND11, increasing chromatin acetylation and altering transcription - most notably of MYC - and alternative splicing; expression favours Myc-driven pluripotency during embryonic stem cell differentiation and self-renewal of haematopoietic stem/progenitor cells.

    PMID 35705031DOI
    What this evidence cannot show

    READ AT ABSTRACT LEVEL ONLY. Europe PMC returns no PMCID for this record and the efetch ArticleIdList carries no pmc id either, so routes 1-4 of the unreadable test all fail; the DOI route was not attempted this session. The systems are human cell lines, mouse embryonic stem cells and HSPCs with an ectopically expressed fusion; this is not a patient-derived system either. No numbers (peak counts, fold-changes) are quoted here because the abstract gives none.

    Evidence that argues against this (1)
    • Li et al. report the majority of fusion peaks as promoter-proximal, with super-enhancer formation at proto-oncogenes, and place the phenotype on a Hoxa/Meis1/Myb/Sox4 stem programme rather than on MYC alone. Neither paper's genomic data has been re-analysed side by side, so the disagreement is currently between two abstracts and one full text. Li J et al. 2021
  3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Across twelve assembled human cases of AML with t(10;17)(p15;q21)/ZMYND11::MBTD1, blasts are CD7+ (12/12), CD33+ (12/12), CD56-bright (11/11) and CD117+ (11/11), HLA-DR-negative (8/10) and MPO-negative (9/9), with minimal differentiation, simple or non-complex karyotypes and rare co-occurring mutations; most adults relapsed early, all four transplanted patients relapsed, salvage responses were poor, and two paediatric patients achieved long-term remission.

    PMID 41262054DOI
    What this evidence cannot show

    Case series, n=5 new cases plus 7 assembled from the literature, single institution for the new cases, retrospective, no control group and no survival statistic with a confidence interval. PubMed returns an EMPTY abstract for this record; the full text was read this session through PMC12951177 after efetch supplied the PMCID that Europe PMC's core record omitted. Denominators differ by marker because not every reported case was tested for every marker - 12/12, 11/11, 10 and 9 are all stated in the paper and are reproduced here unchanged. The classification proposal ('separate entity') is the authors' recommendation, not a WHO or ICC decision.

    Evidence that argues against this (1)
    • The same series concludes that ZMYND11::MBTD1 'does not require additional co-operating mutations' from the observed paucity of mutations, while the founding mouse study found incomplete penetrance, long latency, and a spontaneously acquired Ptpn11 activating mutation, and required NrasG12D to accelerate disease. The clinical inference rests on the sensitivity of the sequencing panels used, which the paper does not enumerate. Li J et al. 2021
  4. E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In paediatric minimally differentiated (FAB M0) AML, t(10;17)(p15;q21) produces an in-frame fusion of ZMYND11 exon 12 to MBTD1 exon 3 encoding a 1,054-amino-acid protein, the reciprocal product is predicted to lack a productive start codon, and gene expression profiling of the leukaemic cells showed high HOXA expression.

    Recurrent translocation t(10;17)(p15;q21) in minimally differentiated acute myeloid leukemia results in ZMYND11/MBTD1 fusion · de Rooij JD, van den Heuvel-Eibrink MM, Kollen WJ, et al. · Genes, Chromosomes & Cancer 2016
    PMID 26608508DOI
    What this evidence cannot show

    Read at abstract level only; no PMCID via either Europe PMC or efetch. The HOXA-high finding is expression profiling of patient leukaemic cells with no stated n in the abstract and no comparator group named there; it is an association in a rare karyotype, not a functional result. The 1,054-residue length is the transcript-derived prediction, not a measured protein mass.

    Evidence that argues against this (1)
    • The functional study that followed used a construct of ZMYND11 aa 1-409 plus the full 628-residue MBTD1 ORF plus a 16-residue linker, which sums to 1,053 rather than 1,054. Neither paper reconciles the one-residue difference, so the exact junction residue of the patient protein is not established by either source. Li J et al. 2021
  5. E5moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    The translocation (10;17)(p15;q21) was described as recurrent in acute poorly differentiated myeloid leukaemia and attributed to a probable ZMYND11-MBTD1 fusion in 2014, two years before the fusion was molecularly confirmed.

    PMID 23915195DOI
    What this evidence cannot show

    TITLE ONLY. PubMed returns an empty abstract for this record (a letter to the editor), there is no PMCID by either route, and the DOI route was not attempted. The word 'likely' in the authors' own title is doing real work: this is a cytogenetic and positional inference, and the molecular confirmation is E4. Nothing about its content is asserted here beyond what the title states.

    Evidence that argues against this (1)
    • [ADDED BY THE COORDINATOR DURING SALVAGE, 2026-09-23 cycle 82. Q121 was killed by the account session limit before it wrote this field. This is a reasoned methodological limitation derived from the claim and its citation type, NOT a literature-sourced rebuttal - no counter-citation was fetched, and none is asserted.] This is a cytogenetic case description and the fusion was ATTRIBUTED AS PROBABLE, not demonstrated at the transcript level in that report. A t(10;17)(p15;q21) karyotype is compatible with breakpoints that do not produce an in-frame ZMYND11-MBTD1 transcript, and 'recurrent' in a small cytogenetic series is a statement about a referral population, not an incidence. STRONGER COUNTER: a series that karyotyped t(10;17) cases and failed to recover the fusion transcript in some of them. Not searched for.
  6. E6moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    A 2026 case report of a patient initially diagnosed as aggressive NK cell leukaemia found an in-frame ZMYND11::MBTD1 fusion by RNA sequencing and nested PCR, argued the immunophenotype (CD7+, CD13+, CD33+, CD34+, CD56+, CD117+, MPO-negative) is that of myeloid/NK cell precursor leukaemia, and on reviewing five published ZMYND11::MBTD1 leukaemias concluded that prior cases had probably been misclassified.

    PMID 41454826DOI
    What this evidence cannot show

    n=1 plus a five-case literature review, read at abstract level only. 'Myeloid/NK cell precursor leukaemia' is not a WHO-recognised entity with settled criteria, and the reclassification argument is an immunophenotypic-pattern argument, not a molecular or outcome one. It overlaps but does not agree in framing with E3, which calls the same immunophenotype AML with minimal differentiation.

    Evidence that argues against this (1)
    • The larger 2026 series reaches the same immunophenotypic description and keeps the diagnosis inside AML with minimal differentiation, proposing recognition under 'other rare recurrent genetic abnormalities' rather than reassignment to a myeloid/NK precursor category. The two 2026 papers agree on the markers and disagree on the label. Wei Q et al. 2026
  7. E7moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    A novel ZMYND11/MBTD1 fusion transcript was reported in CD7-positive, CD56-positive acute myeloid leukaemia with t(10;17)(p15;q21).

    PMID 29911449DOI
    What this evidence cannot show

    Single case, title-level plus the exon-boundary summary reported by E3 (which states this case used ZMYND11 exons 1-11 rather than 1-12). The abstract was not retrieved this session; the case is counted here only as one of the twelve assembled in E3.

    Evidence that argues against this (1)
    • [ADDED BY THE COORDINATOR DURING SALVAGE, 2026-09-23 cycle 82. Q121 was killed by the account session limit before it wrote this field. This is a reasoned methodological limitation derived from the claim and its citation type, NOT a literature-sourced rebuttal - no counter-citation was fetched, and none is asserted.] A single case report. The CD7/CD56 immunophenotype is a property of one patient's blasts and cannot be read as a phenotype of the entity; with fewer than ten records in the whole literature for this fusion, any immunophenotypic association is at the mercy of which cases happened to be written up. STRONGER COUNTER: an unselected cohort reporting immunophenotype across multiple fusion-positive cases. Does not appear to exist.
  8. E8moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    A case of AML with ZMYND11-MBTD1 fusion presented with cannibalistic blasts.

    PMID 31000516DOI
    What this evidence cannot show

    A Blood 'Images in Hematology' single-page item; one patient; the morphological descriptor is not a defining feature of the entity and appears in one case of twelve. Title-level only.

    Evidence that argues against this (1)
    • [ADDED BY THE COORDINATOR DURING SALVAGE, 2026-09-23 cycle 82. Q121 was killed by the account session limit before it wrote this field. This is a reasoned methodological limitation derived from the claim and its citation type, NOT a literature-sourced rebuttal - no counter-citation was fetched, and none is asserted.] A single case report, and 'cannibalistic blasts' is a morphological curiosity that is reported BECAUSE it is unusual - the clearest possible publication-bias channel. Nothing here supports cannibalism being characteristic of the fusion, and this entry should not be read as associating the two. STRONGER COUNTER: morphology reported in fusion-positive cases ascertained without regard to how striking they looked.
  9. E9moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    A ZMYND11::MBTD1 fusion AML arose following chemotherapy and radiotherapy for breast cancer, giving the entity a therapy-related presentation.

    PMID 39258225DOI
    What this evidence cannot show

    Single case, title-level only. One therapy-related case among twelve does not establish the entity as therapy-related; note separately that patient 2 of E3 also had a prior breast cancer treated by mastectomy and hormonal therapy - hormonal therapy is not a leukaemogenic exposure, so these are two different stories that happen to share an organ.

    Evidence that argues against this (1)
    • [ADDED BY THE COORDINATOR DURING SALVAGE, 2026-09-23 cycle 82. Q121 was killed by the account session limit before it wrote this field. This is a reasoned methodological limitation derived from the claim and its citation type, NOT a literature-sourced rebuttal - no counter-citation was fetched, and none is asserted.] A single therapy-related case. Any fusion with fewer than ten reported cases will acquire a therapy-related example by chance once enough cases accumulate, and t-AML after breast cancer chemoradiotherapy is common enough that co-occurrence is weak evidence of a causal or characteristic link. STRONGER COUNTER: the proportion of fusion-positive cases that are therapy-related, against the background t-AML rate. Not computable from 9 records.
  10. E10strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

    Bounded literature counts for this fusion on 2026-09-23: ZMYND11[tiab] AND MBTD1[tiab] returns 9 records; "t(10;17)"[tiab] AND leukemia[tiab] returns 15; ZMYND11[tiab] AND leukemia[tiab] returns 11; MBTD1[tiab] returns 35 in total.

    What this evidence cannot show

    Titles and abstracts only. Every query was run through pubmed_safe.py, which refuses to return a zero on an empty querytranslation; all counts above are non-zero so the false-zero failure mode does not apply here. The whole primary literature on this fusion is nine records, of which two are mechanism papers and the rest are case reports, a case series and a review; nothing here bounds what may sit in supplementary figures of papers about other fusions.

    Evidence that argues against this (1)
    • [ADDED BY THE COORDINATOR DURING SALVAGE, 2026-09-23 cycle 82. Q121 was killed by the account session limit before it wrote this field. This is a reasoned methodological limitation derived from the claim and its citation type, NOT a literature-sourced rebuttal - no counter-citation was fetched, and none is asserted.] These are TITLE/ABSTRACT counts and rule 80 applies with full force: a fusion described only in a figure, a supplementary table or a cytogenetics appendix is invisible to every one of these queries. The counts bound the indexed abstract literature and nothing else. They also cannot distinguish a genuinely rare entity from a poorly-named one - if some reports call it MBTD1-ZMYND11, or give only the karyotype, they are missed. STRONGER COUNTER: the same queries against full text, plus a karyotype-only search.
  11. E11established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    ZMYND11 is Q15326 (602 aa, ENSG00000015171, HGNC:16966, 10p15.3, GRCh38 chr10:130,088-254,771, + strand) with PHD 100-148, bromodomain 149-255, PWWP 280-331 and MYND 563-598; MBTD1 is Q05BQ5 (628 aa, ENSG00000011258, HGNC:19866, 17q21.33, GRCh38 chr17:51,177,419-51,261,114, - strand) with an FCS-type zinc finger at 45-80 and four MBT repeats at 141-245, 253-350, 351-456 and 464-560.

    What this evidence cannot show

    All four accessions fetched live on 2026-09-23. gene_exact:MBTD1 AND organism_id:9606 AND reviewed:true returns exactly ONE reviewed record with MBTD1 as the primary gene name, so the several-rows trap does not fire; the same is true for ZMYND11. HGNC lists no previous official symbol for either gene (MBTD1 aliases SA49P01 and FLJ20055; ZMYND11 aliases BS69 and BRAM1). The ZMYND11 domain boundaries are reproduced from the existing gene-zmynd11 entry's fetched UniProt block and were re-checked against the live record this session. The cytoband in the karyotype (10p15;17q21) and the cytobands of the two genes (10p15.3, 17q21.33) agree.

    Evidence that argues against this (1)
    • [ADDED BY THE COORDINATOR DURING SALVAGE, 2026-09-23 cycle 82. Q121 was killed by the account session limit before it wrote this field. This is a reasoned methodological limitation derived from the claim and its citation type, NOT a literature-sourced rebuttal - no counter-citation was fetched, and none is asserted.] UniProt and HGNC records are CURATION SNAPSHOTS, not experiments. Domain boundaries in particular are predictions or inferences from structure papers and are revised between releases; this project has already recorded a case where a feature table omitted a domain a paper had characterised. Coordinates are assembly-specific and the strand and span quoted are GRCh38 only. STRONGER COUNTER: the same boundaries from the primary structural literature, checked for disagreement with the feature table.
Connected to