Oncofusion

start / graph / fusion-kmt2a-usp2

Fusionmoderate confidence 3 of 5: One strong study, or several smaller consistent ones.Q-usp2-maml2-second-bcr

KMT2A::USP2, inv(11)(q23.3q23.3) - the minor-BCR fusion that keeps the PHD fingers

also called MLL-USP2 · KMT2A-USP2 · MLL::USP2 · inv(11)(q23q23) KMT2A-USP2

KMT2A::USP2 is made when a piece of chromosome 11 flips over and joins the KMT2A gene to a neighbouring gene about a million letters away. It is the ninth most common KMT2A fusion, it mostly affects babies and toddlers, and ordinary chromosome tests often cannot see it at all.

The same thing, for a scientist

KMT2A::USP2 arises by intrachromosomal inversion at 11q23.3, because USP2 lies ~1 Mbp telomeric to KMT2A and is transcribed toward the centromere [E1]. It is the ninth most frequent KMT2A partnership in the 3401-patient DCAL recombinome (30 patients) and occurs exclusively in the minor breakpoint cluster region [E2]. Two consequences follow and neither is shared with the classical elongation-complex fusions. First, the KMT2A moiety retains the complete PHD1-3 cluster, bromodomain and ePHD4, modules that suppress transformation when added back to MLL-AF9 [E3]. Second, the partner contributes a catalytic deubiquitinase domain rather than an elongation module [E1]. Clinically the 17-patient founding series was entirely paediatric (median 17 months, range 3-120), predominantly B-ALL (12) with MPAL (4) and one AML, with CNS disease in 29%, day-33 MRD positivity in 76% and poor prednisone response in 5 of 12 assessed [E1]. Roughly half of cases are unresolvable by standard cytogenetics and FISH, so the fusion is systematically under-ascertained [E4]. No cell line, mouse model or xenograft of KMT2A::USP2 exists, and no chromatin, localisation or menin-dependence experiment has ever been performed on it.

Go deeper
The geometry: an inversion, not a translocation

Every other high-frequency KMT2A fusion in this graph is an interchromosomal translocation. KMT2A::USP2 is not. USP2 sits about 1 Mbp telomeric to KMT2A on 11q23.3 and is transcribed toward the centromere, so joining 5'-KMT2A to 3'-USP2 requires the intervening segment to invert 1. The reciprocal USP2 breakpoints in the founding series were scattered over a larger region of 11q23.3, involving the upstream gene C2CD2L and the downstream USP2-AS1 1.

This is why the karyotype is so often normal and why the fusion is cytogenetically cryptic: there is no derivative chromosome to see, and a break-apart FISH probe set spanning KMT2A may show either a 3'-signal loss or a completely normal wild-type pattern 14.

What is in the protein1 deeper

On the KMT2A side, a minor-BCR break at intron 20 to exon 24 leaves the AT-hooks, the menin-binding region, the LEDGF/PSIP1-binding region and CXXC in place and additionally retains the complete PHD1-PHD2-ePHD3 cluster, the bromodomain and ePHD4 - everything that a major-BCR fusion throws away up to roughly exon 24 2. On the USP2 side, the founding report states that every USP2 and USP8 rearrangement fused the conserved UCH domain to the extended 5'-MLL portion 1. The UCH domain occupies USP2 residues 267-599 with catalytic Cys276 and His557 5.

Caveat that should not be lost: neither paper gives the exon-level junction for individual KMT2A::USP2 patients, so the exact first USP2 residue in the fusion is not published. The domain lists below are derived from the stated BCR2 interval and the stated UCH retention, not from a per-patient junction sequence, and are flagged as such in data.

The PHD paradox, stated without resolving it

Adding two or more PHD fingers back to MLL-AF9 blocked immortalisation of murine haematopoietic progenitors, reduced association with the Hoxa9 locus and suppressed Hoxa9 upregulation 3. KMT2A::USP2 retains all of them and causes leukaemia in thirty people. Something in this fusion either bypasses or neutralises the block. The 2019 authors propose that the partner's UCH domain counteracts the PHD-dependent polyubiquitination and proteasomal degradation of MLL 1 - carried here as hyp-uch-domain-rescues-phd-mediated-mll-degradation, not as fact, because no experiment has tested it. The graph should not state that PHD retention prevents transformation. Thirty patients say otherwise.

Clinical picture2 deeper

From the 17-patient founding series, the only cohort-level clinical description that exists 1:

featurevalue
sex8 male, 9 female
ageall children; median 17 months (range 3-120 months)
WBC at diagnosismedian 30.4 x 10^9/L (range 3.4-324.0)
phenotypeB-ALL 12, MPAL (B/myeloid) 4, AML 1
CNS disease5 of 17 (29%)
MRD positive at day 3313 of 17 (76%)
poor prednisone response5 of 12 assessed (42%)
follow-upmedian 1.2 years (range 0.1-11.1)
deaths2, at 5 and 9 months

An independent three-case series described KMT2A-USP2 and KMT2A-USP8 as high-risk with a high incidence of CNS involvement, poor initial prednisone response and poor event-free survival, and noted that approximately half of cases cannot be resolved by standard cytogenetic approaches 4. The two descriptions agree, which is worth something given how small both are - but they are not independent in the way they look, because both are descriptive series without a comparator arm.

It is not exclusively a B-lineage infant disease

One AML case is in the founding series 1, and the first dedicated report of KMT2A-USP2 in AML was a 9-month-old boy whose fusion was found only by whole-transcriptome sequencing after conventional screening failed 6. A 2025 case report describes an infant B-ALL carrying both KMT2A::USP2 and a CBL::GRIK4 fusion 7, which is the only report of a co-occurring second fusion in this genotype.

Treatment: what has actually been tried

There is no trial. What exists is case-level experience, all of it in relapsed or refractory disease.

  • Blinatumomab. Two refractory paediatric KMT2A-USP2 B-ALL cases treated with blinatumomab 8. A 2026 case used blinatumomab to clear MRD, then blinatumomab as a bridge to haploidentical HSCT, then three further cycles plus two prophylactic donor-lymphocyte infusions post-transplant, with grade 1 CRS and sustained remission 9.
  • Venetoclax. In a 22-patient R/R paediatric ALL venetoclax series, one KMT2A::USP2 patient achieved a morphological leukaemia-free state 10.
  • Menin inhibitors. Nothing. A PubMed search for (MLL OR KMT2A) AND USP2 AND (menin OR MEN1 OR revumenib) returns zero records 11. Given that the fusion retains the menin-binding motifs by position, this is the single largest actionable gap in the entry.
What has never been measured

Stated explicitly, because absence of a statement reads as absence of an effect:

  • Cell of origin / cell type: not determined. The age distribution is consistent with a fetal or early-postnatal progenitor, by analogy only.
  • Cell state: no differentiation-block characterisation.
  • Subcellular localisation: never imaged. USP2 itself is annotated to cytoplasm, perinuclear cytoplasm, nucleus and membrane 5, so the fusion's localisation is not safely predictable.
  • Subnuclear localisation: never imaged.
  • Epigenetic readout: no ChIP-seq, CUT&RUN, H3K79me2, H3K4me3 or HOXA/MEIS1 expression data for any KMT2A::USP2 sample. The only transcriptional handle is that LAMP5 expression tracked the USP2 cases in one classifier 12.
  • Menin dependence: never tested, in any system.
Analogy: the other deubiquitinase fusion, and what the pair can and cannot decide

Closest analogue, and the axis. fusion-kmt2a-usp8, on the axis of the class of the 3-prime module - a USP/UCH-family deubiquitinase catalytic domain - combined with a shared 5-prime extent, since both break in the KMT2A minor BCR and therefore both retain the complete PHD1-3 cluster, the bromodomain and ePHD4. They are the only two deubiquitinase partners in the recombinome, at 30 and 3 patients, and both are minor-BCR-restricted. That double coincidence is the whole basis of hyp-uch-domain-rescues-phd-mediated-mll-degradation.

What the pair predicts, and where the nearest test sits. The prediction is that the retained PHD cassette should block transformation - adding two or more PHD fingers back into MLL-AF9 abolishes immortalisation, reduces association with the Hoxa9 locus and suppresses Hoxa9 upregulation 3 - and that the UCH domain is what overrides the block. The block half has been tested, in a different fusion; the override half has not been tested at all. The counter-observation is already published and is in the human data: 47 minor-BCR patients exist, so the PHD-retaining configuration is plainly leukaemogenic, and two of those cases have AFF1 and one has MLLT3 - no deubiquitinase anywhere.

The fusion geometries differ, which is informative rather than incidental. USP2 arises by intrachromosomal inversion because USP2 lies ~1 Mbp telomeric to KMT2A and is transcribed toward the centromere; USP8 arises by a balanced t(11;15)(q23;q21). Two different mechanisms of formation converging on the same domain class is an argument that the domain is selected for, not a by-product of local genome architecture - which is the strongest thing the pair says.

Where the analogy breaks, and it is a break the UCH-rescue hypothesis does not currently account for: the two deubiquitinases have unrelated substrate biology.

USP2USP8
canonical proteinO75604, 605 aa, USP domain 267-599P40818, 1118 aa, USP domain 777-1109
best-characterised substratesMDM2 and MdmX/MDM4, i.e. the p53 axisEGFR and endosomal receptor trafficking
disease role outside leukaemiaoverexpression stabilises MDM2 and promotes p53 degradationsomatic gain-of-function mutations in the 14-3-3-binding motif cause Cushing's disease by sustaining EGF signalling
fusion geometryintrachromosomal inversion at 11q23.3balanced t(11;15)(q23;q21)

And in both cases the fusion delivers the catalytic domain stripped of its N-terminal regulatory region, which is the one thing the two really do share at the protein level. For USP8 the Cushing hotspot - the 14-3-3-binding motif around residue 718 - sits in a disordered segment N-terminal to the USP domain, so the region whose mutation activates the enzyme in pituitary adenoma is most likely not in the fusion at all. For USP2 the region annotated as necessary for the MDM4 interaction spans residues 1-200, which a minor-BCR fusion carrying the conserved UCH domain does not retain. So the honest statement is that neither partner brings its known substrate biology into the fusion; both bring an unregulated catalytic module, and that is a cleaner statement of what the UCH-rescue hypothesis actually requires than 'a deubiquitinase is fused in'.

The discriminating experiment the pair makes available, which nobody has run. exp-usp2-uch-catalytic-dead-fusion-transformation is designed as a three-arm USP2 experiment. Adding a fourth arm - the matched KMT2A portion fused to the USP8 catalytic domain - converts it from a test of one fusion into a test of the hypothesis. If catalytic class is all that matters, the USP2 and USP8 arms should behave identically and both should be abolished by the active-site mutation. If they differ, the UCH-rescue model is incomplete and something partner-specific is doing work. Neither fusion has a cell line, a mouse or a xenograft, so this is buildable from constructs alone.

A caution on how much weight the pair can carry. n = 30 and n = 3 in the same recombinome, no protein-level evidence for either fusion, no published per-patient junction, and no western blot of either chimera. The pair is a good hypothesis-generator and a poor evidence base, and the existing counter-observation stands: two AFF1 cases and one MLLT3 case also break in the minor BCR and supply no deubiquitinase at all, so the PHD-retaining configuration is leukaemogenic without a UCH domain.

Facts and figures
gene5KMT2A
gene3USP2
cytobandinv(11)(q23.3q23.3)
mechanism classintrachromosomal inversion; USP2 lies ~1 Mbp telomeric to KMT2A and is transcribed toward the centromere
uniprot5Q03164
uniprot3O75604
uniprot3 verifiedO75604 = UBP2_HUMAN, PRIMARY gene name USP2, 605 aa, Homo sapiens (UniProt REST, fetched 2026-09-19; primary gene name and length read back from the record, not assumed from a synonym match). Distinct from Q9UPT9 = USP22.
prognosisDescribed as high-risk in both series that comment on it, with frequent CNS involvement, poor prednisone response and high day-33 MRD. No survival curve, no comparator arm and median follow-up of 1.2 years in the only cohort, so no quantitative prognostic statement is defensible.
diseases
[
  "paediatric B-ALL",
  "mixed-phenotype acute leukaemia (B/myeloid)",
  "infant AML"
]
frequency
{
  "dcal_recombinome_n_2023": 30,
  "dcal_rank_2023": 9,
  "dcal_n_2019": 17,
  "dcal_rank_2019": 14,
  "share_of_bcr2": "17 of 21 in the 2019 discovery cohort; the 2023 series reports 47 BCR2 patients in total",
  "pubmed_records_on_intersection": 11,
  "pubmed_records_title_checked_on_topic": 11,
  "note": "Rank and count are from a single referral centre and rose sharply once targeted NGS replaced PCR, so 30 should be read as a floor."
}
breakpoints
[
  {
    "id": "bp-kmt2a-bcr2-usp2-uch",
    "population": "KMT2A::USP2 as reported in the 2019 discovery cohort and the 2023 recombinome; the KMT2A break lies in the minor BCR (intron 20 - exon 24 per 2023; intron 19 - exon 24 per 2019, preference introns 21-23) and the USP2 contribution is the conserved UCH domain. NO PER-PATIENT EXON-LEVEL JUNCTION IS PUBLISHED, so the retained/lost domain lists below are inferred from the stated interval and the stated UCH retention, not from a junction sequence.",
    "domains_retained_5": [
      "AT-hook 1",
      "AT-hook 2",
      "AT-hook 3",
      "SNL1/2",
      "MBM1",
      "MBM2",
      "CXXC",
      "PHD1",
      "PHD2",
      "PHD3",
      "Bromodomain",
      "PHD4"
    ],
    "domains_lost_5": [
      "FYRN",
      "TAD",
      "FYRC",
      "WIN",
      "SET"
    ],
    "domains_retained_3": [
      "UCH (ubiquitin carboxyl-terminal hydrolase catalytic domain)"
    ],
    "domains_lost_3": [
      "MDM4-interaction region (N-terminal)"
    ],
    "phenotype_notes": "Retention of the complete PHD1-3 cluster and bromodomain is the defining anomaly: those modules block transformation when added back to MLL-AF9, yet this fusion causes leukaemia in at least 30 people."
  }
]
clinical series 2019
{
  "n": 17,
  "sex": "8 male, 9 female",
  "age_median_months": 17,
  "age_range_months": [
    3,
    120
  ],
  "wbc_median_x10e9_per_L": 30.4,
  "wbc_range_x10e9_per_L": [
    3.4,
    324
  ],
  "phenotype": {
    "B-ALL": 12,
    "MPAL (B/myeloid)": 4,
    "AML": 1
  },
  "cns_disease": "5/17 (29%)",
  "mrd_positive_day33": "13/17 (76%)",
  "poor_prednisone_response": "5/12 assessed (42%)",
  "follow_up_median_years": 1.2,
  "follow_up_range_years": [
    0.1,
    11.1
  ],
  "deaths": "2, at 5 and 9 months"
}
never measured
[
  "cell of origin and cell type",
  "cell state / differentiation block",
  "subcellular localisation of the fusion protein",
  "subnuclear localisation of the fusion protein",
  "any epigenetic readout (ChIP-seq, CUT&RUN, H3K79me2, H3K4me3)",
  "HOXA9/MEIS1 expression",
  "menin dependence",
  "catalytic activity of the fusion's UCH domain",
  "any cell line, mouse model or xenograft"
]
How this could bite us back
  • Menin inhibitors are prescribed on the basis of KMT2A rearrangement, not partner. If KMT2A::USP2 turns out not to be menin-dependent, those patients will receive an ineffective drug with real toxicity, and nothing in current practice would detect that, because the partner is usually not even determined.
  • CD19-directed therapy has been used in at least three reported KMT2A::USP2 patients. Four of 17 in the founding series were already B/myeloid MPAL, which is the phenotype most prone to myeloid lineage switch under CD19 pressure - a documented failure mode in KMT2A-rearranged disease.
  • The fusion is cryptic to karyotype and FISH in roughly half of cases. Any trial that enrols on cytogenetics will systematically exclude these patients, so the genotype will stay unstudied for as long as enrolment criteria are written that way.
  • If the UCH domain does rescue the fusion from degradation, then a deubiquitinase inhibitor would be a degrader-by-proxy for this fusion - but USP2's known substrate MDM2 sits upstream of p53, so inhibiting USP2 systemically has an obvious on-target effect in every dividing normal cell.
Still open
  1. Is KMT2A::USP2 menin-dependent? It retains the menin-binding motifs by position and nobody has ever tested it. If it is, every one of these patients is a candidate for an approved drug; if it is not, it is the first well-populated KMT2A partnership that escapes the menin axis, which would matter far beyond 30 patients.
  2. Does the retained PHD1-3/bromodomain cluster in this fusion bind CYP33/PPIE and recruit a Polycomb repressor complex to CXXC, as it does in wild-type KMT2A? If so, the fusion is a repressor-capable molecule and the HOXA-activation model may simply not apply to it.
  3. Why is the phenotype so heavily B-lineage and MPAL (16 of 17) when the retained KMT2A modules are the same ones present in myeloid-dominant fusions? Breakpoint position, not partner identity, is the candidate explanation and it is testable in the existing DCAL breakpoint set.
  4. Is the intervening ~1 Mbp segment between KMT2A and USP2 a single topologically associating domain? If so this inversion is the cleanest available natural test of whether spatial proximity, rather than local fragility, sets partner choice.
  5. Does the KMT2A::USP2 chimera retain USP2 residues 1-200? UniProt annotates that segment as necessary for the MDM4 interaction, and if it is absent the fusion carries catalysis without USP2's characterised substrate-recruitment region - which decides whether the UCH-rescue hypothesis needs USP2 to be a specific deubiquitinase or merely an active one. No per-patient junction has been published at protein level for any of the 30 patients, and a single RT-PCR across the junction in banked material would settle it.
Every source behind this entry (13)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    KMT2A::USP2 arises by intrachromosomal inversion at 11q23.3, accounted for 17 of 21 minor-BCR cases in the discovery cohort, fuses the conserved USP2 UCH domain to an extended 5'-MLL portion, and in those 17 children presented at median 17 months with B-ALL (12), MPAL (4) and AML (1), 29% CNS disease, 76% day-33 MRD positivity and 42% poor prednisone response.

    Human MLL/KMT2A gene exhibits a second breakpoint cluster region for recurrent MLL-USP2 fusions. · Meyer C, Lopes BA, Caye-Eude A, Cavé H, Arfeuille C, et al., Marschalek R · Leukemia 2019
    PMID 30899083DOI
    What this evidence cannot show

    n = 17 for every clinical figure, so each percentage moves by 6 points per patient. Patients were treated on diverse protocols with no comparator arm, median follow-up only 1.2 years, and the cohort was assembled from samples referred to one centre after other assays failed. No outcome comparison against major-BCR KMT2A-r disease is offered. Full text read (PMC6756029).

    Evidence that argues against this (1)
    • The series is ascertained from cases that standard methods could not resolve, which plausibly selects for cryptic, karyotypically normal disease and could bias the clinical picture in either direction - toward worse outcomes if diagnosis was delayed, or toward better if these are patients whose disease lacked an overt complex karyotype. Nothing in the paper controls for this. Blackburn PR et al. 2020
  2. E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    USP2 is the ninth most frequent KMT2A partner at 30 patients in 3401, occurs exclusively in the minor BCR (intron 20 - exon 24), and a BCR2 break places the complete PHD1-3 cluster, bromodomain and ePHD4 into the direct fusion protein.

    The KMT2A recombinome of acute leukemias in 2023. · Meyer C, Larghero P, Almeida Lopes B, Burmeister T, Gröger D, et al., Marschalek R · Leukemia 2023
    PMID 37019990DOI
    What this evidence cannot show

    Single diagnostic centre; rank is a rank within referrals. 'Exclusively BCR2' is a statement about 30 observed patients, not a proven constraint. The domain-retention claim follows from the interval and the KMT2A domain map rather than from protein sequencing of any patient's fusion product. Counts and BCR coordinates read from the published article in this session.

    Evidence that argues against this (1)
    • No independent cohort of comparable size has reproduced the BCR2 partner spectrum, so USP2's dominance of BCR2 could partly reflect which samples reach Frankfurt. The 2019 letter's own rank for USP2 was fourteenth, four years earlier. Meyer C et al. 2019
  3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Inclusion of two or more PHD fingers in MLL-AF9 blocked immortalisation of haematopoietic progenitors, reduced association with the Hoxa9 locus and suppressed Hoxa9 upregulation.

    The PHD fingers of MLL block MLL fusion protein-mediated transformation. · Muntean AG, Giannola D, Udager AM, Hess JL · Blood 2008
    PMID 18796627DOI
    What this evidence cannot show

    Murine retroviral system, MLL-AF9 only, methylcellulose immortalisation readout, no in vivo leukaemia endpoint and no minor-BCR fusion tested. Abstract and central claim read in this session.

    Evidence that argues against this (1)
    • Thirty human patients with a PHD-retaining KMT2A::USP2 fusion are a direct counter-observation. Either the PHD block is partner-dependent, or dosage-dependent, or something in this fusion neutralises it. The murine result must not be generalised into a rule about human minor-BCR fusions. Meyer C et al. 2023
  4. E4moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    Three cases (two KMT2A-USP2, one KMT2A-USP8) were resolvable only by mate-pair sequencing; one had a normal karyotype with atypical KMT2A FISH, one had a normal KMT2A FISH result, and the authors state approximately half of such cases cannot be resolved by standard cytogenetic approaches and are likely under-recognised.

    Cryptic and atypical KMT2A-USP2 and KMT2A-USP8 rearrangements identified by mate pair sequencing in infant and childhood leukemia. · Blackburn PR, Smadbeck JB, Znoyko I, Webley MR, Pitel BA, Vasmatzis G, Xu X, Greipp PT, Hoppman NL, Ketterling RP, Baughn LB, Lindsey KG, Schandl CA, Wolff DJ, Peterson JF · Genes, Chromosomes & Cancer 2020
    PMID 32196814DOI
    What this evidence cannot show

    Three patients, two institutions, retrospective. The 'approximately half' figure is the authors' summary of the prior literature rather than a measurement in this series. The prognostic descriptors (high-risk, CNS involvement, poor prednisone response, poor EFS) are restated from the earlier Frankfurt series, not independently derived. Abstract read in full.

    Evidence that argues against this (1)
    • Diagnostic-detectability claims age quickly. A 2025 four-laboratory Australian exchange found three of four laboratories detected KMT2A::USP2 correctly by targeted RNA sequencing, so the fusion is no longer cryptic to a modern RNA-based assay even though it remains cryptic to karyotype and FISH. Li EW et al. 2025
  5. E5established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    USP2 is UniProt O75604, 605 aa, with the catalytic UCH domain at residues 267-599 (Cys276, His557) and annotated localisation to cytoplasm, perinuclear cytoplasm, nucleus and membrane. UniProt additionally annotates residues 1-200 and 403-503 as necessary for interaction with MDM4, so the N-terminal half of the region required for USP2's best-characterised substrate interaction lies outside the conserved UCH domain that the minor-BCR fusion retains.

    What this evidence cannot show

    Database annotation for the wild-type canonical 605-aa isoform, not for the fusion protein. Which USP2 residues actually begin the fusion is not published for any patient. Accession resolved by reading back the primary gene name (USP2) and length (605 aa) from the fetched record. The MDM4-interaction regions were added on 2026-09-21 from a fresh fetch of the same record. Whether residues 1-200 are truly absent from the chimera cannot be stated, because no per-patient junction has been published at protein level for any KMT2A::USP2 case; the inference is from the founding report's statement that the conserved UCH domain is what is fused.

    Evidence that argues against this (1)
    • Because the per-patient junction is unpublished, the assertion that the fusion contains an intact, correctly folded UCH domain rests on the founding letter's sequence-level statement rather than on any protein-level evidence. No western blot of a KMT2A::USP2 protein has been published. Meyer C et al. 2019
  6. E6moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    The first reported case of AML with a KMT2A-USP2 fusion was a 9-month-old boy in whom the fusion was found only by whole-transcriptome sequencing after conventional screening for known KMT2A partners failed.

    Whole transcriptome sequencing reveals a KMT2A-USP2 fusion in infant acute myeloid leukemia. · Ikeda J, Shiba N, Tsujimoto SI, Yoshida M, Nakabayashi K, Ogata-Kawata H, Okamura K, Takeuchi M, Osumi T, Tomizawa D, Hata K, Kiyokawa N, Ito S, Kato M · Genes, Chromosomes & Cancer 2019
    PMID 30869817DOI
    What this evidence cannot show

    Single patient. No outcome follow-up long enough to inform prognosis, and the authors' framing is explicitly that partner identification in KMT2A-r AML is needed precisely because outcomes by partner are unknown. Abstract read in full.

    Evidence that argues against this (1)
    • The Frankfurt series published a month earlier already contained one AML case among 17, so 'first report' is a claim about a dedicated case report rather than about first observation. Two independent AML cases in about 47 reported patients makes AML a real but minority phenotype for this fusion. Meyer C et al. 2019
  7. E7weak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.paper

    A case of infant-type B-ALL carrying both a KMT2A::USP2 and a CBL::GRIK4 fusion has been reported.

    Dual KMT2A::USP2 and CBL::GRIK4 Fusions in Infant-Type B-ALL: A Case Report. · Du C, Zhang N, Xia M, Jiang L, Weng W · Pediatric Blood & Cancer 2025
    PMID 41189082DOI
    What this evidence cannot show

    A Letter with no indexed abstract, so only the title-level claim is carried here; the body was not read. One patient. Nothing about causality, clonal structure or outcome can be inferred, and a second fusion in one case does not establish co-operation.

    Evidence that argues against this (1)
    • No other KMT2A::USP2 report describes a co-occurring fusion, and the founding series explicitly reported infrequent concurrent alterations in the analogous KMT2A-MAML2 thymomas. A single dual-fusion case is at present an outlier rather than a pattern. Meyer C et al. 2019
  8. E8weak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.paper

    Two cases of refractory childhood B-ALL positive for KMT2A-USP2 have been treated with blinatumomab and reported.

    [Two cases of refractory childhood acute B-lymphoblastic leukemia with positive KMT2A-USP2 treated with Belintouximab]. · Sun Y, Ni YA, Xu HJ, Wang LZ, Yang J, Jiang J, Zhong R · Zhonghua Er Ke Za Zhi (Chinese Journal of Pediatrics) 2023
    PMID 37803862DOI
    What this evidence cannot show

    Chinese-language case report with no indexed English abstract; only title-level information is carried here and the body was not read. Two patients, no comparator, no durable-outcome data available to this entry. The transliteration 'Belintouximab' in the indexed title is blinatumomab.

    Evidence that argues against this (1)
    • CD19-directed therapy in KMT2A-rearranged infant leukaemia carries a documented risk of myeloid lineage switch, which this graph holds as risk-lineage-switch-cd19-directed-therapy. Given that 4 of 17 KMT2A::USP2 patients in the founding series were already B/myeloid MPAL, blinatumomab in this genotype should not be reported as straightforwardly favourable. Meyer C et al. 2019
  9. E9weak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.paper

    A 5-year-old boy with KMT2A::USP2-positive early pro-B ALL and persistent MRD cleared flow and RQ-PCR MRD after one cycle of blinatumomab, was bridged to haploidentical HSCT, and received three post-transplant blinatumomab cycles plus two prophylactic donor-lymphocyte infusions with only grade 1 CRS and sustained remission.

    PMID 42389523DOI
    What this evidence cannot show

    One patient. Follow-up duration after the last DLI is not stated in the abstract. Blinatumomab plus prophylactic DLI is a combination whose contributions cannot be separated in a single case, and the authors themselves say larger prospective trials are needed. Abstract read in full; body not read.

    Evidence that argues against this (1)
    • A single favourable case cannot be weighed against the cohort-level picture, in which 76% of KMT2A::USP2 patients were MRD-positive at day 33 and the disease is described as high-risk with poor event-free survival. Publication bias toward successful salvage is the expected direction for single case reports. Blackburn PR et al. 2020
  10. E10weak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.paper

    In a 22-patient series of venetoclax-based therapy for relapsed or refractory paediatric ALL, one KMT2A::USP2-positive child achieved a morphological leukaemia-free state.

    Venetoclax-Based Regimen in Refractory or Relapsed Pediatric Acute Lymphoblastic Leukemia. · Zhang L, Zhang Z, Lu A, Jia Y, Huang L, Zhang L, Zeng H · Acta Haematologica 2025
    PMID 40628245DOI
    What this evidence cannot show

    n = 1 for this genotype inside a 22-patient single-centre retrospective series. MLFS is a weaker endpoint than CR or MRD-negative CR. Venetoclax was combined with chemotherapy, so no single-agent inference is possible. Abstract read in full.

    Evidence that argues against this (1)
    • The same series reports the strongest responses in ETP-ALL rather than in KMT2A-rearranged disease, and the KMT2A::USP2 patient reached only MLFS while ETP patients reached MRD-negative remission. Read as written, this is a weak signal for this genotype, not a positive one. Li YT et al. 2026
  11. E11strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

    No published work connects KMT2A::USP2 to menin or to menin inhibitors: (MLL OR KMT2A) AND USP2 AND menin returns 0 PubMed records, as does the same query with revumenib or MEN1 added.

    PubMed E-utilities esearch counts for KMT2A/USP2/menin intersection queries, run 2026-09-19 · National Center for Biotechnology Information (NCBI), PubMed · PubMed (NCBI E-utilities) 2026
    What this evidence cannot show

    A counted-absence claim, and the standing warning in this project is that such claims are only as good as the symbol set. This one was run over USP2, MEN1, menin and revumenib but not over every menin-inhibitor trade or code name, and it would miss a trial enrolling a KMT2A::USP2 patient without naming the partner in the abstract. It is a statement about the indexed literature, not about what has been observed in clinics.

    Evidence that argues against this (1)
    • Menin-inhibitor trials enrol on KMT2A rearrangement, not on partner identity, so KMT2A::USP2 patients have almost certainly received revumenib without appearing in any partner-indexed search. The absence in the literature is real; an absence of exposure is not established. Meyer C et al. 2023
  12. E12weak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.paper

    High LAMP5 expression estimated the occurrence of all KMT2A-USP2 fusions in a random-forest/LightGBM classifier trained to predict KMT2A rearrangement from expression data.

    Novel Diagnostic and Therapeutic Options for KMT2A-Rearranged Acute Leukemias. · Lopes BA, Poubel CP, Teixeira CE, Caye-Eude A, Cavé H, Meyer C, Marschalek R, Boroni M, Emerenciano M · Frontiers in Pharmacology 2022
    PMID 35734412DOI
    What this evidence cannot show

    'All KMT2A-USP2 fusions' means all such cases in the training data, a very small number that the abstract does not state. The reported AUCs are for KMT2A-r overall (SKIDA1 0.839, LAMP5 0.746, CSPG4/NG2 0.722), not for the USP2 subset. Shared authorship with the group that defined the fusion. Full-text abstract read in this session; the model was not re-run.

    Evidence that argues against this (1)
    • LAMP5 overexpression is a general feature of KMT2A-rearranged leukaemia rather than a USP2 marker, so perfect separation of USP2 cases in one training set is most parsimoniously a small-sample coincidence. Li EW et al. 2025
  13. E13strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    USP2's best-characterised biology is the p53 axis rather than anything chromatin-related: USP2a binds MDM2 and deubiquitinates it without reversing MDM2-mediated p53 ubiquitination, so overexpression accumulates MDM2 and promotes p53 degradation; MdmX/MDM4 is also a USP2a substrate, associating independently of MDM2, with wild-type but not catalytically dead USP2a preventing MDM2-mediated MdmX degradation, and siRNA knockdown of endogenous USP2a destabilising MdmX in NTERA-2 and MCF7 cells.

    MdmX is a substrate for the deubiquitinating enzyme USP2a. · Allende-Vega N, Sparks A, Lane DP, Saville MK · Oncogene 2010
    PMID 19838211DOI
    What this evidence cannot show

    Testicular embryonal carcinoma and breast carcinoma cell lines with overexpression and siRNA, nothing haematopoietic and nothing to do with a fusion protein; USP2a is one isoform of USP2 and the isoform used is not the object this entry is about. Cited here only to establish what USP2's characterised substrate biology IS, as the input to a disanalogy - not as evidence that any of it operates in KMT2A::USP2.

    Evidence that argues against this (1)
    • This substrate biology may not be transferred into the fusion at all. UniProt annotates residues 1-200 of the 605-residue protein as necessary for the interaction with MDM4, alongside 403-503; a minor-BCR KMT2A::USP2 chimera retains the conserved UCH catalytic domain (267-599) and not the N-terminus, so the fusion plausibly delivers catalysis without the substrate-recruitment region that the p53-axis work depends on. That makes the reading in this entry's analogy section - both USP fusions deliver an UNREGULATED catalytic module rather than their partner's known substrate specificity - the better-supported one, and it is why the disanalogy with USP8 is about regulation rather than about which proteins get deubiquitinated. UniProt Consortium; InterPro (EMBL-EBI) 2026
Connected to