Oncofusion

start / graph / disease-kmt2a-r-sarcoma

Diseasestrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.Q92-kmt2a-sarcoma

KMT2A-rearranged soft tissue sarcoma - the same gene as infant B-ALL, in a young adult's leg, as a solid tumour

also called KMT2A-rearranged sarcoma · YAP1::KMT2A-rearranged sarcoma · MUC4-negative sclerosing epithelioid fibrosarcoma · KMT2A (MLL) fusion sarcoma · MLL-rearranged sarcoma

A rare soft-tissue cancer of young adults is driven by a rearrangement of KMT2A - the same gene whose rearrangement causes infant leukaemia - but it grows as a solid lump in muscle or limb rather than in blood, and it often spreads. There are two forms: a dense, scar-like one with a YAP1 partner, and a more uniformly cellular one with a vimentin partner.

The same thing, for a scientist

KMT2A-rearranged soft tissue sarcoma is an emerging entity first reported in 2019-2020 and still absent from most solid-tumour sequencing panels. In a 14,680-sarcoma comprehensive genomic profiling database, KMT2A fusions were found in 34 cases (0.2%); YAP1 was the partner in 16 (47%), VIM in 2, and 16 had unique non-recurrent partners of unclear significance. The YAP1 arm presents as a MUC4-negative, EWSR1/FUS-fusion-negative sclerosing epithelioid fibrosarcoma (SEF)-like tumour; the VIM arm as a uniformly cellular spindle-to-round-cell tumour without sclerosis. Median age is 38-40 years with a reported range of 10-91, male:female about 1.4:1, predominantly somatic soft tissue with rare bone and dural primaries. The behaviour is aggressive: in the 22-case series with follow-up available in 14, local recurrence occurred in 21% and metastasis in 50%. DNA methylation profiling places the YAP1 arm in a cluster distinct from both conventional SEF and low-grade fibromyxoid sarcoma, supporting a separate entity rather than a SEF subtype. FISH is unreliable - break-apart probes for KMT2A and YAP1 are frequently negative in sequencing-confirmed cases - so the diagnosis depends on RNA-based NGS. There is no standard therapy, no cell line, no xenograft and no clinical trial specific to the entity.

Go deeper
Two morphologies, two partners
YAP1 armVIM arm
share of KMT2A-rearranged sarcomas16/34 (47%) 12/34 (6%) 1
morphologySEF-like: monomorphic epithelioid-to-spindle cells in dense collagenous/sclerotic stroma 12uniformly cellular fascicular spindle cells, no extensive sclerosis 3
MUC4negative in all tested 24not the defining marker
immunoprofilenon-specific; YAP1 and EMA often positive; cyclin D1 a possible clue 4CD99+, BCL2+, variable S100 3
architectureinsertional YAP1::KMT2A::YAP1 1conventional VIM(ex4)::KMT2A(ex2) 3
HOXA outputnot upregulated (n=3) 5reported upregulated (secondary attribution, weak) 5

A third and a fourth arrangement exist as single case reports - CBX6::KMT2A::PYGO1 and LDB1::KMT2A - each with its own spindle-cell morphology 6.

Who gets it

Median age 38 in the largest series (22 cases) and median 40 in the pan-sarcoma screen's YAP1 arm (range 20-66) 14. But the range across series is much wider than 'young adults' suggests: 10 to 86 in one nine-case series (including a 12 cm frontal dural tumour in a 10-year-old girl) and 11 to 91 in another eight-case series 25. Male:female about 1.4:1 4. Sites are predominantly somatic soft tissue - lower extremity most often - with trunk, upper extremity, acral, neck, dura and one primary intraosseous case reported 245.

The age contrast with the index disease is the point. Infant KMT2A::AFF1 B-ALL peaks in the first year of life and its rearrangement is largely prenatal. This entity's median is around four decades later, in a mesenchymal lineage, with no reported congenital or infant case. Same gene, different lineage, different age, and - by the breakpoint architecture - a different molecule.

Is it one entity? The honest answer is 'the YAP1 arm is'

DNA methylation profiling of 22 YAP1::KMT2A cases formed a cluster clearly separate from conventional SEF and from LGFMS 4, and gene expression profiling separately distinguished YAP1/KMT2A tumours from LGFMS/SEF 2. That is good evidence for the YAP1 arm.

The other 16 cases in the pan-sarcoma screen are a different matter: diverse histologies (leiomyosarcoma, atypical lipomatous tumour, undifferentiated pleomorphic sarcoma, high-grade endometrial stromal sarcoma), median age 57, TP53 mutation in 44%, KMT2A as the 3' partner in eight and the 5' partner in seven, no reciprocal transcripts, and the authors call them 'of unclear significance' 1. Those look like passenger rearrangements in genomically complex sarcomas. 'KMT2A-rearranged sarcoma' as an entity means the YAP1 and VIM arms, not every sarcoma with a KMT2A break - and this distinction matters for the graph, because the 16 are the group that would inflate any prevalence figure.

What it implies for the index fusion, by contrast

Three contrasts, in descending order of how much they should change this project's mind.

  1. The molecule is different, not just the lineage. In both defined arms the fusion lacks KMT2A exon 1 and therefore the menin-binding motif. This is not a KMT2A::AFF1-like protein in a fibroblast; it is a different chimera. So the entity is weak evidence about lineage permissiveness and strong evidence about which parts of KMT2A are needed for which kind of transformation.
  2. A KMT2A rearrangement can transform without a HOXA program. The YAP1 arm shows no HOXA9/MEIS upregulation 5. The leukaemic program's dependence on HOXA is therefore not a property of KMT2A rearrangement; it is a property of the leukaemic molecule in a haematopoietic cell.
  3. The partner class is unprecedented. Neither partner is elongation machinery. If the index fusion's mechanism is 'partner supplies elongation competence', these fusions cannot work that way, and what they do instead has never been measured.
Facts and figures
entity statusemerging entity; described 2019-2020 and referred to in the 2025 literature as 'the emerging entity of KMT2A-rearranged sarcomas'
prevalence34 of 14,680 sarcomas submitted for comprehensive genomic profiling (0.2%); a testing-cohort figure and a lower bound, because KMT2A is often absent from solid-tumour panels and FISH is insensitive
age median years38
sex ratio m f1.4
behaviourlocal recurrence 3/14 (21%), metastasis 7/14 (50%) at median 25 months follow-up in the 22-case series
diagnostic pitfallbreak-apart FISH for KMT2A and YAP1 is frequently negative in sequencing-confirmed cases; RNA-based NGS is required, and hypocellular 'fibroma-like' areas invite a benign diagnosis
distinguishing markersMUC4 negative and EWSR1/FUS fusion negative, which separates it from conventional sclerosing epithelioid fibrosarcoma
methylation22 YAP1::KMT2A cases form a cluster distinct from conventional SEF and from LGFMS; the dataset is by-request and not deposited
standard therapynone established; surgery with or without radiation in reported cases
models availablenone - no cell line, no xenograft and no patient-derived model of this entity exists in the literature searched
contrast with index diseaseinfant KMT2A::AFF1 B-ALL is a haematopoietic disease of the first year of life with a largely prenatal rearrangement retaining KMT2A exon 1; this is a mesenchymal solid tumour with a median age of 38 whose fusions lack KMT2A exon 1
arms
[
  "YAP1::KMT2A::YAP1 (SEF-like)",
  "VIM::KMT2A (uniformly cellular spindle-to-round cell)",
  "CBX6::KMT2A::PYGO1 (n=1)",
  "LDB1::KMT2A (n=1)"
]
age range years
[
  10,
  91
]
sites
[
  "lower extremity (commonest)",
  "trunk",
  "upper extremity",
  "acral",
  "neck",
  "dura",
  "bone (1 case)"
]
How this could bite us back
  • There is no standard therapy and no model system, which makes this exactly the setting where a speculative targeted recommendation is most dangerous: a clinician reading 'KMT2A-rearranged' will think of menin inhibitors, and by the breakpoint architecture there is no menin binding site in either defined arm.
  • The entity is currently defined by methylation or expression clustering rather than by morphology, and the defining dataset is not deposited. A patient assigned to this entity on a cluster nobody can re-run is assigned on an unauditable basis.
  • 50% metastasis at a median 25 months means any single-arm response signal in a basket trial will be confounded by the natural history, and with four-to-thirty published cases per arm there is no realistic path to a randomised comparison.
  • Because FISH is unreliable here, an institution that screens for this entity by FISH will report an absence and conclude the tumour is something else - the same class of instrument-generated negative this project has been repeatedly bitten by.
Still open
  1. What is the cell of origin? Ultrastructural fibroblastic differentiation was noted in one tumour and the tumours make collagen, but no lineage-tracing or single-cell work exists - and the index disease's whole cell-of-origin literature has no counterpart here.
  2. Why is there no infant or congenital case, when the same gene's rearrangement in blood is overwhelmingly a disease of infancy? Is the mesenchymal lineage permissive only later, or is the rearrangement mechanism different?
  3. Is the 50% metastasis rate real, or an artefact of series assembled from consultation practice and commercial profiling? A population-based cohort through a sarcoma registry would answer it and none exists.
  4. Do the 16 non-recurrent-partner KMT2A sarcomas share anything at all, or are they passenger rearrangements? Their 44% TP53 rate and median age of 57 suggest the latter, and a methylation profile of that group would settle it.
Every source behind this entry (6)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In a comprehensive genomic profiling database of 14,680 human sarcomas plus one index patient, 34 KMT2A-fusion-positive sarcomas (0.2%) were identified: YAP1 was the partner in 16 (47%) usually as a complex YAP1-KMT2A-YAP1 rearrangement, VIM in 2, and 16 had unique non-recurrent partners with diverse histologies, median age 57 and TP53 mutation in 44%, which the authors describe as of unclear significance.

    PMID 32461620DOI
    What this evidence cannot show

    A commercial profiling database, which the authors themselves name as a limitation because it is enriched for aggressive and advanced disease where genetic testing is sought. The denominator of 14,680 is sarcomas SUBMITTED for profiling, not incident sarcomas, so 0.2% is a testing-cohort prevalence and not a population incidence. Histologic review was limited to a representative section in most cases. KMT2A is under-represented on solid-tumour panels, so this is a floor. Read from the open-access full text.

    Evidence that argues against this (1)
    • Because the rearrangement is frequently missed by FISH and KMT2A is often not covered on solid-tumour panels, the true prevalence among MUC4-negative SEF-like tumours is expected to be higher than any current figure - the authors of a companion series say so explicitly. A prevalence estimate from a panel that often lacks the gene is a lower bound of unknown depth. Kao YC et al. 2020
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In nine human MUC4-negative, EWSR1/FUS-fusion-negative sclerosing epithelioid fibrosarcomas with YAP1 and/or KMT2A rearrangements, patients ranged from 10 to 86 years (mean and median 45), four female and five male, with tumours in extremities, trunk, neck and dura; break-apart FISH was negative for KMT2A in all three sequencing-confirmed index cases and positive for YAP1 in only one, and of six cases with follow-up two recurred locally and two metastasised with one death.

    PMID 31592798DOI
    What this evidence cannot show

    Nine cases assembled from consultation practice, so referral bias toward diagnostically difficult tumours is heavy. Twenty-four control fibromyxoid spindle cell tumours were tested and none carried the abnormality, which is a useful specificity control but a small one. Follow-up available in only six. Read from the PMC full text via efetch db=pmc after Europe PMC fullTextXML returned HTTP 500.

    Evidence that argues against this (1)
    • These authors read the tumours as a subset of MUC4-negative SEF rather than as a separate entity, and titled the paper that way. The later methylation evidence favours a distinct entity, so the classification in this founding series has been superseded - but the disagreement is about entity status, not about the molecular findings. Warmke LM et al. 2025
  3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Two human VIM::KMT2A-rearranged sarcomas in adult males' deep soft tissues were hypercellular fascicular spindle-cell neoplasms with scant delicate collagen, diffusely CD99- and BCL2-positive and variably S100-positive, lacking the extensively sclerotic background of the KMT2A::YAP1 tumours, with the fusion at VIM exon 4 and KMT2A exon 2 in both.

    VIM::KMT2A-rearranged sarcomas: A report of two new cases confirming an entity with distinct histologic features. · Almohsen SS, Griffin AM, Dickson BC, et al. · Genes Chromosomes Cancer 2023
    PMID 36959690DOI
    What this evidence cannot show

    Two case reports. The authors state the tumours have no specific immunohistochemical profile and overlap morphologically with other monomorphic spindle cell neoplasms, so cases cannot be found without molecular testing and the morphological definition rests on four published cases in total. Read from the abstract.

    Evidence that argues against this (1)
    • The pan-sarcoma screen described its two VIM-KMT2A tumours as sheets of spindle-to-round BLUE cells and the founding case expressed BCOR, WT1 and NKX2-2 - a small-round-cell immunoprofile. Whether 'uniformly cellular spindle cell' and 'small round blue cell' describe the same morphology across four cases is not established. Massoth LR et al. 2020
  4. E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In the largest series of human YAP1::KMT2A-rearranged sarcomas (22 cases), median age was 38 years with a male:female ratio of 1.4:1, tumours predominantly involved somatic soft tissue with the first reported primarily intraosseous case, YAP1 and EMA were often expressed and MUC4 never was, local recurrence occurred in 3 of 14 patients with follow-up (21%) and metastasis in 7 (50%), and DNA methylation analysis formed a cluster clearly separate from conventional SEF and from low-grade fibromyxoid sarcoma.

    PMID 39641785DOI
    What this evidence cannot show

    Retrospective multi-institutional collection with referral bias; median follow-up only 25 months (range 1-170), so the 50% metastasis figure is over a short and uneven window and the denominator is 14 of 22. Several cases had initially false-negative FISH and solid-tumour panel results, i.e. the series is of cases that survived a diagnostic gauntlet. The methylation dataset is available from the corresponding author on request and was not deposited, so the clustering cannot be independently re-run.

    Evidence that argues against this (1)
    • One reported YAP1::KMT2A sarcoma could not be classified even after the fusion was found: it showed diffuse ERG, CD31, CD34, D2-40 and TFE3 positivity and epithelioid haemangioendothelioma-like blister cells, and only RNA-seq clustering placed it with the SEF-like group. A methylation or expression cluster is doing the entity definition where morphology fails, which means the entity is currently defined by its assay. Fumagalli C et al. 2025
  5. E5moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    In three human YAP1-KMT2A sclerosing epithelioid fibrosarcomas, RNA sequencing showed no upregulation of the KMT2A leukaemia targets HOXA9 and MEIS (and no upregulation of YAP1 or KMT2A itself), whereas the same authors record that upregulation of KMT2A and HOXA genes had been noted in the previously reported VIM-KMT2A sarcoma.

    PMID 31592798DOI
    What this evidence cannot show

    n = 3, reported as 'data not shown' in a discussion, with no deposited data - elink from this PMID to GEO returns zero datasets (checked 2026-09-21). The VIM-arm half is a secondary attribution to a paper whose full text is unreadable through all five of this project's routes (no PMCID; Wiley DOI returns HTTP 403), so nobody here has seen either measurement. A two-gene negative in three bulk samples cannot exclude HOXA engagement in a subpopulation.

    Evidence that argues against this (1)
    • If the VIM arm does upregulate HOXA, the claim that this entity runs a non-HOXA program applies only to the YAP1 arm, and the entity would then split on precisely the axis that defines the leukaemic program. The graph should not generalise the HOXA-negative result across both arms. Yoshida A et al. 2019
  6. E6moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    Two further arrangements have been reported as single human cases: CBX6::KMT2A::PYGO1 in a 47-year-old man's subcutaneous chest spindle cell tumour, disease-free at 2 years 8 months without postoperative treatment, and LDB1::KMT2A (LDB1 exons 1-10 to KMT2A exons 4-36) in a 19-year-old man's erector spinae mass with a concurrent pathogenic BCOR frameshift, disease-free at 5 months.

    KMT2A-rearranged sarcoma with unusual fusion gene CBX6::KMT2A::PYGO1. · Nakamura H, Kukita Y, Wakamatsu T, et al. · Virchows Arch 2023
    PMID 37713130DOI
    What this evidence cannot show

    Two single case reports, both read at abstract level, with follow-up of 32 and 5 months respectively - far too short to support the implied indolent behaviour. The first author explicitly calls the arrangement 'highly likely irreproducible'. These cases widen the entity's molecular definition but cannot revise its natural history.

    Evidence that argues against this (1)
    • Both non-YAP1, non-VIM cases were alive and disease-free at last follow-up after surgery alone, which sits badly with the entity's reputation for aggressive behaviour and 50% metastasis. Either the partner determines behaviour, or the aggressive figures come from series enriched for aggressive cases. With n=1 per partner, neither can be distinguished. Wang Z et al. 2025
Connected to