Oncofusion

start / graph / fusion-picalm-mllt10

Fusionstrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.Q55-calm-af10

PICALM::MLLT10 (CALM-AF10), t(10;11)(p13;q14) - the DOT1L-dependent HOXA fusion that contains no KMT2A

also called CALM-AF10 · CALM::AF10 · CALM/AF10 · PICALM-MLLT10 · PICALM::MLLT10 · TAF_I-unrelated t(10;11)

A chromosome swap glues the clathrin adaptor PICALM onto the chromatin protein AF10, making a single hybrid protein that switches on the same HOXA growth genes as the KMT2A leukaemias - but without containing any KMT2A at all. It causes an aggressive leukaemia that is usually T-cell, sometimes myeloid, and responds poorly to standard chemotherapy.

The same thing, for a scientist

t(10;11)(p13;q14-21) fuses **PICALM** (11q14, Q13492) to **MLLT10/AF10** (10p12.31, P55197), producing a ~170 kDa chimera in which essentially all of CALM (residues 1-648 of 652) is joined to the C-terminal portion of AF10 [E1][E2]. The fusion retains the AF10 **OM-LZ**, which binds the DOT1L coiled coil, and the CALM **NES** at aa 544-553; it loses the AF10 **PZP** reader module [E2][E3][E4]. Its output programme is **HOXA5/7/9/10 plus MEIS1**, established in transgenic mice [E5] and in patient T-ALL [E6] - though **which HOXA gene is actually required differs from MLL-AF10**: Hoxa5 knockout attenuates CALM-AF10 transformation and does nothing to MLL-AF10, and Hoxa9 was undetectable in that system's CALM-AF10-transformed cells [E17], and it is **absolutely DOT1L-dependent**: conditional Dot1l knockout abolishes initiation and maintenance, and the DOT1L inhibitor EPZ004777 collapses Hoxa and Meis1 expression and selectively impairs CALM-AF10-transformed cells [E7]. It is **T-ALL-skewed**: 65% T-ALL, 27% AML, 8% ambiguous-lineage in a 156-patient aggregate [E8], and within T-ALL it is essentially restricted to the **TCR-gamma-delta and immature delta/gamma lineage** [E6][E9]. **This entry is written as the sharpest available natural test of the menin-interface hypothesis** - same HOXA/MEIS1 output, same DOT1L dependence, no KMT2A. The answer to whether it is menin-dependent is, as of 2026-09-20, **that nobody has tested it directly**; the indirect evidence points to menin-independence and is set out in `hyp-menin-axis-separable-from-hoxa-programme` and `ana-calm-af10-menin-bounded-absence`.

Go deeper
What the fusion contains, and what it does not
elementsourcefatewhat it does
ANTH domainCALMretainedmembrane/PtdIns(4,5)P2 and R-SNARE binding; no assigned role in transformation
clathrin-binding C-terminus (~400-648)CALMretainedsufficient, with OM-LZ, to induce AML in mice 3
NES 544-553CALMretainedCRM1-dependent export; necessary and sufficient for transformation 2
PZP reader moduleAF10lostreads unmodified H3K27 and licenses DOT1L; its restoration in cis abolishes transformation 4
NLSAF10retained
OM-LZAF10retainedbinds the DOT1L coiled coil; necessary and sufficient with the CALM C-terminus 3
Q-rich regionAF10retained
menin-binding motif-absentthere is no KMT2A in this protein at all
CXXC / post-CXXC-absentno CpG-island chromatin address

The last two rows are the whole point of this entry for this project.

Is it menin-dependent? The honest state of the evidence

No primary study has tested it. On 2026-09-20, CALM-AF10 AND menin in PubMed returns 0 records, with clean operand controls (CALM-AF10 alone = 81; menin alone = 2,237), and PICALM AND menin = 0, AF10 AND menin = 0. MLLT10 AND menin returns 3, all about KMT2A::MLLT10. Full-text search in Europe PMC returns 34 co-occurrences, every one of which is a review or a conference abstract book 10. The 2026 comprehensive review of this disease does not contain the word 'menin' at all 11.

What we have instead is three indirect lines, all pointing the same way:

  1. The fusion has no KMT2A. A menin-KMT2A inhibitor has no fusion to displace. For it to work, CALM-AF10 would have to depend on the wild-type menin-KMT2A1 complex to maintain HOXA - which is the mechanism by which NPM1c and NUP98-fusion leukaemias are menin-dependent, and which nobody has tested here.
  2. DepMap. In the CALM-AF10 line U-937, MEN1 gene effect is -0.229, against a KMT2A-rearranged mean of -1.263 and a non-KMT2A leukaemia mean of -0.401 (SD 0.406); DOT1L in the same line is -0.542, at the KMT2A-rearranged level (-0.604); PSIP1/LEDGF is +0.135 against a KMT2A-rearranged -0.553 12. One line, so descriptive only - but the pattern is DOT1L-dependent, menin- and LEDGF-independent, which is exactly the separation this beat was sent to look for.
  3. CRISPR-ChIP partition. H3K79 methylation splits into a fusion-directed unit that is menin-dependent and a separate MLLT10-directed endogenous unit 13. CALM-AF10 works through MLLT10's own DOT1L interface, i.e. through the half of the partition that the screen assigns to the menin-independent unit.

Counterweight, stated because it is real: the 2024 AUGMENT-101 protocol enrols 'acute leukaemia with a genetic mutation expected to lead to HOX/MEIS upregulation' as cohort 2D 14, so the sponsor's own operating hypothesis is that HOXA/MEIS dysregulation predicts menin sensitivity regardless of mechanism. If that is right, CALM-AF10 should respond. Nobody has reported a CALM-AF10 patient treated on that cohort.

Clinical face

Aggregated across 156 patients (12 institutional + 144 from the literature): median age 24 years, T-ALL 65%, AML 27%, ambiguous-lineage 8%; about half of the T-ALL cases are ETP-ALL; 5-year OS 71% in children versus 33% in adults, and 25% in AML versus 54% in T-ALL 8. In a 98-child international cohort, PICALM::MLLT10 ALL had 5-year EFS 67% / OS 76%, while PICALM::MLLT10 AML had 5-year EFS 22% and OS 26%, and transplant did not rescue it (5-year OS 37% with HSCT versus 36% without) 15. Extramedullary and CNS disease are over-represented, and co-mutations differ by lineage: TP53 and NF1 mark PM-AML and associate with progression, EZH2 alterations are enriched in PM-T-ALL, and PHF6 disruption is a shared cooperating event 16.

Facts and figures
gene5PICALM
gene3MLLT10
cytobandt(10;11)(p13;q14-21)
uniprot5Q13492
uniprot3P55197
fusion protein length AA reported1595
fusion protein mw kda reported170
prognosisAdverse, and lineage- and age-dependent: paediatric ALL 5-year OS 76%, paediatric AML 5-year OS 26% with no benefit from transplant; adult 5-year OS 33%.
menin binding motif presentfalse
cxxc domain presentfalse
dot1l routeAF10 OM-LZ binding the DOT1L coiled coil - the same interface used by KMT2A::MLLT10
why this fusion matters hereSame HOXA/MEIS1 output and same DOT1L dependence as a KMT2A fusion, with no KMT2A, no menin-binding motif and no CXXC chromatin address. It is the natural experiment that separates the menin axis from the HOXA programme - if anyone runs it.
lineage distribution 156 patients
ambiguous_lineage_pct8.000ambiguous_lineage_pct: 8median_age_years24.0median_age_years: 24aml_pct27.0aml_pct: 27t_all_pct65.0t_all_pct: 65
diseases
[
  "T-ALL (including ETP-ALL)",
  "AML",
  "acute leukaemia of ambiguous lineage / MPAL",
  "rarely B-ALL"
]
breakpoints
[
  {
    "id": "bp-picalm-mllt10-calm-1-648-af10-omlz",
    "gene5_last_residue": 648,
    "domains_retained_5": [
      "ANTH",
      "clathrin-binding C-terminal region (~400-648)",
      "NES (544-553)"
    ],
    "domains_lost_5": [
      "the last four residues of CALM"
    ],
    "domains_retained_3": [
      "OM-LZ (octapeptide motif-leucine zipper)"
    ],
    "domains_lost_3": [
      "PZP (PHD-Zn knuckle-PHD reader module)"
    ],
    "phenotype_notes": "The canonical construct used across the murine literature. At least eight alternative PICALM::MLLT10 transcripts are reported; see breakpoint-picalm-mllt10-transcript-variants.",
    "domains_3_not_in_the_graph_domain_map": "The fusion also retains an AF10 nuclear localisation signal and a C-terminal glutamine-rich region, and retains PHD2 while losing PHD1. Neither the NLS nor the Q-rich region is annotated in `protein-mllt10-domains`, so they are recorded here in prose rather than as domain entries - naming them in the domain list produces unresolvable lanes in the 3D viewer. This is a real gap in protein-mllt10-domains, not in this entry."
  }
]
frequency
{
  "t_all_pct_unselected": 9,
  "t_all_pct_tcr_gamma_delta": 19,
  "t_all_pct_immature_delta_gamma": 33,
  "paediatric_aml_pct": "<1",
  "adult_t_all_aggregate_pct": 6.5
}
lineage distribution 156 patients — the numbers behind the chart
{
  "t_all_pct": 65,
  "aml_pct": 27,
  "ambiguous_lineage_pct": 8,
  "median_age_years": 24
}
How this could bite us back
  • A DOT1L inhibitor is the mechanistically obvious drug here and it has already failed clinically in KMT2A-rearranged disease; the pinometostat paediatric and adult phase 1 trials enrolled on '11q23 rearrangement', which by definition EXCLUDES t(10;11)(p13;q14) PICALM::MLLT10 because the 11 breakpoint is q14, not q23. So the clinical failure of DOT1L inhibition was never tested in this fusion.
  • If CALM-AF10 turns out menin-independent, that is not only a mechanistic result - it means a growing number of HOXA-high patients will be given a menin inhibitor on a HOX/MEIS eligibility criterion that their disease does not satisfy.
  • CALM-AF10 leukaemias carry global H3K79 hypomethylation and increased chromosomal instability, so a DOT1L inhibitor would deepen a genome-destabilising lesion the disease already has.
  • Half of PICALM::MLLT10 T-ALL is ETP-ALL, a subtype with its own poor outcome and its own biology; a targeted agent tested in this fusion will be tested in a population where the confounding is severe.
Still open
  1. Is CALM-AF10 leukaemia sensitive to a menin inhibitor? As of 2026-09-20 no published experiment addresses this, in any system. It is a two-week in vitro experiment with cell lines that already exist.
  2. Which HOXA gene is actually REQUIRED in human CALM-AF10 cells? The only requirement experiment is murine and says Hoxa5, not Hoxa9, and says the requirement differs from MLL-AF10's. Nobody has repeated it in twenty years.
  3. Does CALM-AF10 require wild-type KMT2A1 and menin to sustain HOXA? Wild-type MLL1 is required for several non-KMT2A HOXA-driven leukaemias; nobody has asked it here.
  4. Does the fusion's HOXA activation happen at the same loci and in the same 3D compartment as a KMT2A fusion's, or does it merely converge on the same output genes? CTCF is required for CALM-AF10 leukaemia, which hints at an architectural rather than an addressing mechanism.
  5. Why is this fusion T-lineage-skewed when KMT2A::MLLT10 is myeloid-skewed, given that both fusions carry the same AF10 module? The difference must lie in the 5' partner or its expression pattern, and no experiment in this beat's reading isolates it.
Every source behind this entry (17)
  1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    The t(10;11)(p13;q14) in the U937 cell line fuses the AF10 gene to CALM, a new member of the AP-3 clathrin assembly protein family.

    PMID 8643484DOI
    What this evidence cannot show

    The founding report, in a cell line rather than a patient. 'AP-3 clathrin assembly protein family' is the 1996 nomenclature; the protein is now classified in the AP180/CALM family and the gene symbol is PICALM. U-937 remains the reference CALM-AF10 human line.

    Evidence that argues against this (1)
    • THE DISCOVERY SYSTEM IS A CELL LINE, AND THE WRONG LINEAGE FOR THIS GRAPH'S ARGUMENT. t(10;11) was characterised in U937, a histiocytic/myeloid line, whereas PICALM::MLLT10 is commonest in T-ALL. That matters beyond provenance: U937 is the single line every DepMap argument about CALM-AF10 in this graph rests on, so the lineage mismatch is inherited by the dependency claims downstream rather than confined to the original report. Oncofusion knowledge graph (internal entry) 2026
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In a murine bone-marrow retroviral system, the CALM-AF10 chimera fuses all but the last four residues of CALM to AF10, and CALM residues 544-553 form a CRM1-dependent NES that is necessary for transformation, with a bare NES sufficient to confer transforming potential on AF10.

    PMID 23487024DOI
    What this evidence cannot show

    Murine bone marrow and HEK293. Numbering frame is stated by the authors (NP_009097.2), which is why it is quotable.

    Evidence that argues against this (1)
    • A RETROVIRAL CHIMERA IN MURINE BONE MARROW IS NOT THE ENDOGENOUS LESION. Expression is driven from a viral promoter at non-physiological level, the cell of origin is chosen by the transduction protocol rather than by the disease, and rule 73 applies throughout. This graph already records that no transduce-and-transplant experiment can establish a cell of origin, because HSCs and GMPs give indistinguishable disease. Oncofusion knowledge graph (internal entry) 2026
  3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In mice, the C-terminal 248 residues of CALM fused to the OM-LZ domain of AF10 are sufficient to induce AML with aberrant Hoxa cluster upregulation.

    PMID 21681188DOI
    What this evidence cannot show

    Murine bone-marrow transplantation with a retroviral minimal construct. The minimal fusion transforms more efficiently than the full-length fusion, so it is a tool, not a model of the patient protein.

    Evidence that argues against this (1)
    • SUFFICIENCY IN A MOUSE IS NOT NECESSITY IN A PATIENT, and a minimal construct sufficient to transform says nothing about which domains the full-length fusion actually uses in human disease. The same objection this graph applies to oligomerisation constructs for KMT2A fusions applies here: a fragment that transforms a murine progenitor is a statement about the assay's permissiveness as much as about the biology. Oncofusion knowledge graph (internal entry) 2026
  4. E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    The PZP domain of AF10 is consistently impaired or deleted in leukemogenic AF10 translocations, and incorporating a functional AF10 PZP into CALM-AF10 in cis prevents transforming activity in murine HSPCs in vitro and in vivo, abrogates leukemogenesis, downregulates Hoxa genes and reverses transformation.

    PMID 34226546DOI
    What this evidence cannot show

    Murine HSPCs and crystallography. The construct is an engineered cis restoration, not a patient allele - it establishes that PZP loss is permissive for transformation, not that PZP loss is the initiating event.

    Evidence that argues against this (1)
    • 'CONSISTENTLY IMPAIRED OR DELETED ACROSS TRANSLOCATIONS' IS A CORRELATION OVER BREAKPOINTS, NOT A DEMONSTRATION THAT PZP LOSS IS REQUIRED. Breakpoint position is constrained by genomic architecture - where the breaks physically occur - so a domain lost in every reported fusion may be lost because of where the gene breaks rather than because its loss is selected. Distinguishing the two needs a PZP-restored fusion, which is not reported. Oncofusion knowledge graph (internal entry) 2026
  5. E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    A CALM-AF10 transgene drives Hoxa cluster overexpression and acute leukaemia in transgenic mice.

    PMID 17804713DOI
    What this evidence cannot show

    Transgenic mouse, long latency (median ~12 months), which the authors and others take as evidence that cooperating events are required.

    Evidence that argues against this (1)
    • The HOXA-activation account sits beside an opposite global phenotype from the same fusion: CALM-AF10 REDUCES H3K79 methylation genome-wide by sequestering DOT1L away from chromatin, with increased chromosomal instability and more secondary aberrations in patients. 'CALM-AF10 drives H3K79 methylation' is therefore false as a global statement and true only focally. Lin YH et al. 2009
  6. E6strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In patient T-ALL, CALM-AF10-positive cases overexpress HOXA and BMI1.

    PMID 16107895DOI
    What this evidence cannot show

    Expression profiling of patient samples; associative, not mechanistic. Small numbers of CALM-AF10 cases, as is unavoidable in a fusion at this frequency.

    Evidence that argues against this (1)
    • AN EXPRESSION ASSOCIATION IN PATIENT MATERIAL IS NOT A DEPENDENCY. HOXA and BMI1 overexpression in CALM-AF10-positive T-ALL establishes that the programme is ON, not that the leukaemia requires it, and this graph separates expressed / required / maintains-a-state precisely because that conflation has burned it before. BMI1 in particular is high across many stem-like leukaemias. Oncofusion knowledge graph (internal entry) 2026
  7. E7strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In mouse models, CALM-AF10-mediated transformation is Dot1l-dependent: conditional Dot1l disruption abolishes in vitro transformation of murine bone-marrow cells and in vivo initiation and maintenance of CALM-AF10 leukaemia, and the DOT1L inhibitor EPZ004777 suppresses Hoxa and Meis1 and selectively impairs CALM-AF10-transformed cells.

    PMID 23138183DOI
    What this evidence cannot show

    Murine. The same paper covers MLL-AF10, so the CALM-AF10 result sits in a comparison that shares reagents with a KMT2A fusion. EPZ004777 is a tool compound, not the clinical DOT1L inhibitor.

    Evidence that argues against this (1)
    • DOT1L inhibition worked in murine CALM-AF10 models and has not been tested clinically in this fusion at all: the pinometostat paediatric and adult phase-1 trials enrolled patients with 'rearrangement involving the MLL gene, including reciprocal chromosomal translocations involving 11q23', and PICALM::MLLT10 breaks at 11q14. The clinical disappointment with DOT1L inhibition is therefore not evidence about this disease. ClinicalTrials.gov 2026
  8. E8moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    Across 156 patients with PICALM::MLLT10 (12 institutional plus 144 from the literature), median age was 24 years, T-ALL 65%, AML 27%, ambiguous-lineage 8%, about half of T-ALL were ETP-ALL; 5-year OS was 71% in children versus 33% in adults, and 25% in AML versus 54% in T-ALL.

    PMID 38971327DOI
    What this evidence cannot show

    144 of the 156 patients are pooled from the published literature, so the cohort inherits every reporting bias of case-level publication and cases may be double-counted against other series. Treatment is heterogeneous across decades.

    Evidence that argues against this (1)
    • PUBLISHED CONFLICTING RESULT on the prognostic axis. The same survival split appears in a prospectively treated cohort with a different explanatory variable: in 22 adults on LALA94/GRAALL03-05 the poor EFS and OS were restricted to the 15 T-cell-receptor-negative cases, ETP status had no prognostic impact per se, and the age skew in immunophenotype is large - 72% of adults TCR-negative against 67% of children TCR-positive. A pooled series that records age and lineage but not surface TCR will attribute a maturation effect to age. The 71%-versus-33% figure is therefore not safe to quote as an age effect. Ben Abdelali R et al. 2013
  9. E9strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    CALM-AF10 was found in 12 of 131 unselected T-ALLs (9%), was absent from TCR-alpha-beta lineage cases, and was present in 19% of TCR-gamma-delta and 33% of immature delta/gamma T-ALLs.

    PMID 12676784DOI
    What this evidence cannot show

    Single large European cohort, adults and children pooled. Detection was by RT-PCR and/or FISH; only about half the positive cases showed t(10;11) by classical cytogenetics, so karyotype-only series undercount this fusion.

    Evidence that argues against this (1)
    • THE LINEAGE IS WRONG FOR THE COMPARISON THIS GRAPH USES IT IN. 9% of unselected T-ALL is a T-lineage frequency, while the index disease here is infant B-lineage KMT2A::AFF1. Whenever CALM-AF10 is invoked as the menin-independent control, the contrast carries a lineage difference as well as a fusion difference, and no published experiment separates them. Oncofusion knowledge graph (internal entry) 2026
  10. E10moderate confidence 3 of 5: One strong study, or several smaller consistent ones.database

    As of 2026-09-20, PubMed returns 0 records for `CALM-AF10 AND menin`, `PICALM AND menin` and `AF10 AND menin`, with operand controls of 81 and 2,237; Europe PMC full-text co-occurrence returns 34 records, all reviews or conference abstract books.

    What this evidence cannot show

    Bounded absence. Name lists tried: menin, MEN1, menin inhibitor / inhibitors / inhibition, revumenib, SNDX-5613, ziftomenib, VTP-50469, VTP50469, MI-503, MI-2-2, bleximenib, JNJ-75276617, against CALM-AF10, CALM AF10, PICALM, PICALM-MLLT10, MLLT10, AF10. An absence of publication is not an absence of experiment, and negative in-house results in menin-inhibitor programmes would be exactly the kind of data that never reaches PubMed.

    Evidence that argues against this (1)
    • A COUNTED ABSENCE IS BOUNDED BY ABSTRACTS, NOT BY PAPERS (rule 80), AND THE STRONGER OBJECTION IS THAT ABSENCE OF MENIN WORK IS NOT EVIDENCE OF MENIN-INDEPENDENCE. This graph's own finding is that chromatin-bound CRM1 recruits SET-NUP214 AND NPM1c onto HOXA, and NPM1c AML is clinically menin-inhibitor sensitive - so a KMT2A-free, CRM1-addressed, HOXA-driving lesion is already known to be menin-DEPENDENT. Nobody having searched CALM-AF10 for menin dependence is a statement about the literature, not about the biology. Oncofusion knowledge graph (internal entry) 2026
  11. E11moderate confidence 3 of 5: One strong study, or several smaller consistent ones.review

    A 2026 comprehensive review of PICALM::MLLT10 leukaemia covering structure, mechanism, prognosis and treatment does not mention menin anywhere.

    PICALM::MLLT10 translocated leukemia. · Cullen JM, et al. · FEBS Lett 2026
    PMID 41532548DOI
    What this evidence cannot show

    This is a REVIEW and is cited here as a map of what the field currently considers relevant, not as evidence about biology. Full text fetched and searched on 2026-09-20 (Europe PMC fullTextXML, 69,891 characters). Silence in one review is weak evidence; it is recorded because the review is recent, comprehensive and written by the group that did much of the primary CALM-AF10 work.

    Evidence that argues against this (1)
    • PUBLISHED CONFLICTING RESULT about what this silence maps. A review's omission tracks what has been tested, and menin dependence has been demonstrated twice outside KMT2A rearrangement since the experiments this review summarises: NUP98-rearranged AML is a menin-MLL1 dependency in vitro and in patient-derived xenografts, and NPM1-mutant AML responds to menin inhibition clinically. The same review does discuss DOT1L, CRM1 and BMI1 - the mechanisms that HAVE been tested in this fusion. So the omission is most economically read as 'nobody has run the experiment in PICALM::MLLT10', which is the opposite of the field having considered and dismissed it, and is far weaker support for separability than an entry quoting a comprehensive recent review implies. Heikamp EB et al. 2022
  12. E12weak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.dataset

    In DepMap CRISPR leukaemia screens, the CALM-AF10 line U-937 shows MEN1 gene effect -0.229 and PSIP1 +0.135 against KMT2A-rearranged means of -1.263 and -0.553, while DOT1L is -0.542 against a KMT2A-rearranged mean of -0.604.

    What this evidence cannot show

    ONE cell line. The second Cellosaurus-annotated PICALM::MLLT10 DepMap line (TUR, ACH-002314) has no screen data. U-937 carries many other lesions (it is a long-passaged histiocytic lymphoma line) and a single line cannot support a selectivity claim. ACTB in the same line is -0.874, so the screen is working. Treat as a hypothesis generator, not a result.

    Evidence that argues against this (1)
    • ONE CELL LINE CANNOT SUPPORT THIS COMPARISON, AND THIS PROJECT HAS ALREADY BOUNDED IT. The CALM-AF10 arm is n=1 (U-937), and the graph's own power analysis put the minimum detectable difference above the observed difference - so the value is consistent with no difference and with a substantial one. It should never be cited as showing menin-independence, only as showing that the experiment able to show it has not been done. The dose-response panel designed independently by two earlier beats is the fix. Oncofusion knowledge graph (internal entry) 2026
  13. E13strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    CRISPR-ChIP partitions H3K79 methylation into an oncogenic DOT1L complex directed by the MLL fusion protein in a menin-dependent manner and a separate endogenous DOT1L complex whose catalytic activity is directed by MLLT10.

    PMID 37679565DOI
    What this evidence cannot show

    ABSTRACT ONLY - the paper is paywalled and route 5 (DOI resolution) returned a 303-character stub. The H3K79me2 CRISPR-ChIP arm and the menin-inhibitor ChIP-seq were performed in **murine MLL-AF9 (KMT2A::MLLT3) AML cells**, resolved from the study's GEO depositions GSE192562/192561/192555/192544 via elink (organism Mus musculus, cell line MLLAF9, samples 'MLLT10 G1/G2', 'menin inhibitor treatment 48hrs'); the human arms are K562 CML and cover H3K4me3 and Pol II, not H3K79me2. So the partition is established in murine MLL-AF9 and has not been shown in CALM-AF10 or in KMT2A::AFF1.

    Evidence that argues against this (1)
    • CRISPR-ChIP PARTITIONS WHAT ITS PERTURBATION CAN REACH. A method that assigns H3K79 methylation to an oncogenic DOT1L complex is reporting the fraction its guides and its timepoint resolve, and this graph separately records that CALM-AF10 STRIPS H3K79 methylation genome-wide while adding it at HOXA - two opposite effects that a partitioning analysis can average if its readout is not locus-resolved. Oncofusion knowledge graph (internal entry) 2026
  14. E14established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.trial_registry

    AUGMENT-101 (NCT04065399) cohort 2D enrols documented relapsed/refractory acute leukaemia with 'a genetic mutation expected to lead to HOX/MEIS upregulation (for example, KMT2Ar, NPM1m, and NUP98r)'; PICALM, MLLT10 and CALM-AF10 are named nowhere in the eligibility criteria.

    NCT04065399
    What this evidence cannot show

    Read from the raw ClinicalTrials.gov v2 API on 2026-09-20, not from the MCP wrapper. Eligibility language is not an efficacy claim; cohort 2D's wording makes a CALM-AF10 patient formally eligible without the sponsor having said the fusion is menin-dependent.

    Evidence that argues against this (1)
    • PUBLISHED CONFLICTING RESULT on what this eligibility text can support. The registration-enabling phase 2 of AUGMENT-101 that reported out restricted its efficacy analysis to patients with CENTRALLY CONFIRMED KMT2A rearrangement (n = 57 evaluable of 94 treated), with the NPM1 cohort reported as separate and still ongoing. A HOX/MEIS clause that makes a CALM-AF10 patient formally eligible has therefore never produced a reported, centrally confirmed, non-KMT2Ar, non-NPM1m patient in this trial. Eligibility language is a recruitment instrument written by a sponsor, and reading it as a statement that the sponsor expects HOX/MEIS-driven leukaemias to respond attributes a biological claim to a document that carefully avoids making one. Issa GC et al. 2025
  15. E15strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In 98 children with PICALM::MLLT10 across ten IBFM-affiliated groups and the COG, PICALM::MLLT10 ALL had 5-year EFS 67% and OS 76%, while PICALM::MLLT10 AML had 5-year EFS 22% and OS 26%, and HSCT did not improve AML outcome (5-year OS 37% with versus 36% without).

    PMID 37743097DOI
    What this evidence cannot show

    Retrospective and multicentre, so treatment is heterogeneous and transplant allocation is not randomised - the null HSCT effect is subject to allocation by response. 55 T-ALL and 39 AML cases.

    Evidence that argues against this (1)
    • A MULTI-GROUP COHORT INHERITS ITS GROUPS' REFERRAL AND TESTING PATTERNS. Ten IBFM affiliates plus the COG is a strong denominator for treated children, but PICALM::MLLT10 is detected by assays that differ between groups and eras, so a frequency or outcome estimate carries ascertainment as well as biology. This graph applies the same caution to the KMT2A recombinome. Oncofusion knowledge graph (internal entry) 2026
  16. E16moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    Genomic profiling of 20 PICALM::MLLT10 paediatric and young-adult cases confirms HOXA activation, identifies PHF6 disruption as a key cooperating event, and shows lineage-divergent co-mutation spectra - TP53 and NF1 in PM-AML, associated with progression and relapse; EZH2 alterations enriched in PM-T-ALL.

    PMID 38429501DOI
    What this evidence cannot show

    n=20, so the co-mutation frequencies are estimates from small strata. The comparison for differential expression is hematopoietic stem cells, which is a strong choice for stemness questions and a weak one for lineage questions.

    Evidence that argues against this (1)
    • TWENTY CASES IS A DESCRIPTION, NOT A TEST. Confirming HOXA activation in 20 profiled cases establishes consistency of the programme, not that the programme is the dependency, and with no comparison arm it cannot separate what PICALM::MLLT10 does from what any HOXA-high leukaemia looks like. The HOXA-high KMT2A-wild-type comparison is the missing control. Oncofusion knowledge graph (internal entry) 2026
  17. E17strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In murine bone marrow, Hoxa5 knockout attenuated the transformation capability of CALM-AF10 but had no effect on MLL-AF10, and Hoxa9 expression was undetectable in the CALM-AF10-transformed cells while being detectable with the same primers in MLL-ENL-transduced cells.

    PMID 16921363DOI
    What this evidence cannot show

    Murine retroviral bone-marrow transduction; read in full text 2026-09-20 (PMC4425349), not in the abstract. The Hoxa5-knockout arm uses a 129Sv/Ev-MF1 background because Hoxa5 knockout is embryonic-lethal in C57BL/6. Three other systems do report Hoxa9 elevation in CALM-AF10, so this is a requirement result in one system rather than a general statement about the disease.

    Evidence that argues against this (1)
    • CALM-AF10 transgenic mice overexpress Hoxa5, Hoxa7, Hoxa9 and Hoxa10, and patient T-ALL shows HOXA9 and HOXA10 elevation, so the cluster-level statement survives even where the gene-level requirement result does not generalise. Caudell D et al. 2007
Connected to