Oncofusion

start / graph / fusion-kmt2a-mllt3

Fusionestablished confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.Q11-comparative

KMT2A::MLLT3 (MLL-AF9), t(9;11)(p21.3;q23.3)

also called MLL-AF9 · MLL::AF9 · KMT2A-AF9 · t(9;11)(p22;q23) · MLL/MLLT3

This is the same broken KMT2A gene as in infant leukemia, but joined to a different partner called AF9, and the combination produces myeloid leukemia rather than lymphoid. AF9 carries a small reading head that recognises chemical tags on chromatin and brings in an enzyme called DOT1L, so this version of the fusion has an extra, partly druggable step that the AFF1 version does not.

The same thing, for a scientist

KMT2A::MLLT3 is the second commonest KMT2A rearrangement overall (612/3401, 18.0%) and the commonest in myeloid disease, accounting for about 30.4% of KMT2A-rearranged AML and 36.6% of paediatric KMT2A-r AML, versus only ~11.8% of KMT2A-r ALL [E1]. The KMT2A contribution is the canonical retained module set (AT-hooks, MBM1/MBM2, LEDGF-binding region, CXXC; SET and all PHD fingers lost). MLLT3/AF9 contributes an N-terminal **YEATS** domain that reads acetylated and crotonylated histone H3 lysines and a C-terminal **ANC1-homology domain (AHD)** that binds the AFF1/AFF4 CHD, DOT1L, BCoR and CBX8 [E2]. AF9-dependent recruitment of DOT1L deposits H3K79me2/3 across the fusion target genes, coupling acetylation reading to an active elongation mark [E2]. In the mouse, MLL-AF9 transforms a **committed granulocyte-macrophage progenitor** into a leukemia stem cell by reactivating a self-renewal program, which is why the disease presents later than KMT2A::AFF1 and is myeloid [E3]. Prognostically t(9;11) AML sits in the middle of the KMT2A-r AML spectrum: the historically claimed favourable outcome was not confirmed in an 11-group retrospective study of 756 children in which KMT2A-r AML overall had a 5-year EFS of 44% +/- 5% with subgroup EFS spanning 11% to 92% [E4]. Menin inhibition is partner-agnostic and active here [E5]; DOT1L inhibition, the partner-specific idea, produced only modest single-agent clinical activity [E6].

What it looks like
loading 2LM0…
PDB 2LM0 ↗
shape

This is real experimental data — atoms placed by crystallography or cryo-EM, deposited by the people who did the experiment.

I have never read one of these pictures before

A protein is one long chain of amino acids that folds up. The ribbon traces that chain from one end to the other; it is not a wire, it is a simplification of where the backbone runs. Drag to rotate, scroll to zoom.

Domains are the stretches of chain that do a particular job, and they are what the coloured blocks above the picture mark out. Click one and the view flies to it. In this project the important question about any domain is usually whether the fusion still has it — the labels say lost where it does not.

Anything in yellow sticks is a bound small molecule, which is usually the drug.

Go deeper
What is kept and what is lost on each side1 deeper

Kept from KMT2A: AT-hooks 1-3, SNL/repression region, MBM1 and MBM2, the LEDGF/PSIP1-binding region, CXXC. Lost from KMT2A: PHD1-3, bromodomain, ePHD4, FYRN/FYRC, TAD, WIN motif and the catalytic SET domain 1. Identical to KMT2A::AFF1 - this is the partner-agnostic half.

Kept from MLLT3/AF9: the C-terminal portion including the AHD (ANC1 homology domain), an intrinsically disordered ~70-residue module that folds only on binding and is the docking site for AFF1/AFF4 (the SEC scaffold), DOT1L, BCoR and CBX8 2. In most fusion transcripts the break lies such that AF9 contributes the C-terminal half.

Lost from MLLT3/AF9 (usually): the N-terminal YEATS domain, the acetyl-/crotonyl-lysine reader. This is an important and frequently muddled point: the YEATS domain is what makes wild-type AF9 a chromatin reader, and in the commonest KMT2A::MLLT3 transcripts it is not in the fusion - the fusion instead inherits AF9's protein-protein interaction surface. AF9's YEATS domain still matters to the disease, because endogenous wild-type AF9 and its paralogue ENL remain present and YEATS-dependent, but it is a separate node from the fusion protein itself.

So the fusion's mechanistic gain is interaction, not reading: the AHD hands the retargeted KMT2A a SEC scaffold and DOT1L.

Breakpoint distribution

KMT2A breaks in KMT2A::MLLT3 fall across the major BCR (introns 9-11), with therapy-induced cases shifting toward intron 11; the AML preference for intron 9 seen at cohort level applies here 1. CORRECTED 2026-09-19: this previously read "Roughly half of MLLT3 fusions in the DCAL series are 'spliced' fusions...". That figure belongs to MLLT1, not MLLT3, and was mis-generalised. Spliced fusions are RARE in MLLT3: 2 of 612 cases (0.3%). The ~50% figure is MLLT1-specific (241 of 460). MLLT3 breakpoints do occur at more than one site, so a single canonical exon pair still does not describe every case, but spliced fusions are not why 1.

Mechanism: the DOT1L arm of the elongation hijack

Class: chromatin-elongation-hijack. As with AFF1, menin/LEDGF docking plus CXXC targeting delivers the chimera to CpG-island promoters. The partner-specific step is what happens next.

The AF9 AHD binds DOT1L, the sole H3K79 methyltransferase. DOT1L deposits H3K79me2/3 across the bodies of fusion target genes, a mark tightly coupled to productive elongation, and in wild-type AF9 the YEATS domain physically links histone acetylation to this DOT1L-mediated H3K79 methylation 2. Genetic or pharmacological inactivation of Dot1l abrogates MLL-fusion-driven transformation in models, which is why DOT1L looked like the ideal partner-specific target 2.

The AHD is also a competitive hub: AFF1/AFF4, DOT1L, BCoR and CBX8 bind overlapping surfaces, so the fusion's output depends on which of them wins. That competition is a real and underexploited pharmacological handle - a peptidomimetic or small molecule occupying the AHD would disable the fusion without touching menin.

Cell of origin. In mouse models MLL-AF9 converts a committed granulocyte-macrophage progenitor into a leukemia stem cell, reactivating a self-renewal signature that partly overlaps but is not identical to the embryonic/HSC program 3. That is a later, more mature origin than KMT2A::AFF1's fetal lympho-myeloid progenitor, and it maps onto the clinical picture: myeloid lineage, peak in infancy-to-early-childhood AML rather than in the first months of life.

Clinical phenotype, age and how the prognosis differs from AFF1

Lineage skew. MLLT3 is the myeloid partner par excellence: ~30.4% of KMT2A-r AML versus ~11.8% of KMT2A-r ALL, and 36.6% of paediatric KMT2A-r AML 1. Where AFF1 is essentially absent from AML (0.8%), MLLT3 dominates it - the clearest single demonstration in this family that partner identity, not breakpoint position, sets lineage 1.

Age skew. Infant AML 28.4%, paediatric AML 36.6%, adult AML 24.5% of KMT2A-r cases in that age band 1. So MLLT3 spans childhood with a paediatric peak, unlike AFF1's sharp under-one-year peak.

Morphology. Frequently monocytic/myelomonocytic (FAB M4/M5) with extramedullary disease, gum and skin infiltration.

Prognosis - the corrected story. t(9;11) was long taught as the 'good' KMT2A-r AML. In the 756-patient international retrospective series this could not be confirmed: the favourable-outcome subgroup was t(1;11)(q21;q23) (HR 0.1), while t(6;11)(q27;q23) (AFDN, HR 2.2) and both t(10;11) subgroups (HR 1.5 and 2.5) were independently adverse; KMT2A-r AML overall had a 5-year EFS of 44% +/- 5% with subgroup EFS ranging from 11% +/- 5% to 92% +/- 5% 4. The practical reading: MLLT3 is intermediate, not favourable, and it is favourable only relative to AFDN and MLLT10.

Contrast with AFF1. In ALL, AFF1 marks the worst-prognosis infant disease. In AML, MLLT3 is the reference partner against which the bad ones (AFDN, MLLT10) are measured. Same 5' half, opposite clinical role - because the cell it lands in is different.

Druggability and therapeutic status

Menin inhibition (partner-agnostic): clinically validated for the class. Revumenib's registration cohort enrolled KMT2A-translocated acute leukemia regardless of partner, and the FDA label is partner-agnostic; the retained MBM1/MBM2 module is identical here, so KMT2A::MLLT3 AML is squarely within the treated population 5. Responses in AML are the bulk of the menin-inhibitor experience.

DOT1L inhibition (partner-specific): the cautionary tale. Pinometostat (EPZ-5676) is a potent, selective DOT1L inhibitor that reduced H3K79 methylation in patients but produced only modest clinical activity as a single agent in adult acute leukemia 6. The mechanism was confirmed; the therapeutic index was not. This is the single most instructive negative result in the KMT2A field: a validated, enzymatic, partner-specific target with on-target pharmacodynamics still failed to deliver monotherapy responses, because H3K79me2 loss is slow and the leukemic program is buffered.

Undrugged and promising: the AF9 AHD interaction hub (AFF1/AFF4, DOT1L, CBX8, BCoR compete for it) has no clinical molecule. YEATS-domain ligands exist as chemical probes against wild-type AF9/ENL and are a distinct, non-fusion-directed strategy.

Emerging overall - clinically validated only through the partner-agnostic menin node, with no approved partner-specific agent.

Facts and figures
gene5KMT2A
gene3MLLT3
cytobandt(9;11)(p21.3;q23.3)
cytoband noteWritten t(9;11)(p22;q23) in most of the clinical literature; p21.3 is the current ISCN band assignment for MLLT3.
prognosisIntermediate within KMT2A-r AML, not favourable: 11q23/KMT2A-r paediatric AML overall 5-year EFS 44% +/- 5% (n=756), and the previously claimed favourable prognosis of t(9;11) was not confirmed; t(6;11) (HR 2.2) and t(10;11) (HR 1.5 and 2.5) were the independently adverse subgroups.
uniprot5Q03164
uniprot3P42568
uniprot5 verifiedQ03164 = KMT2A_HUMAN, 3969 aa (UniProt REST, checked 2026-09-18)
uniprot3 verifiedP42568 = AF9_HUMAN, gene MLLT3, protein AF-9, 568 aa, Homo sapiens (UniProt REST, checked 2026-09-18)
exon numbering noteKMT2A exon numbers follow the Meyer/Marschalek DCAL consensus. MLLT3 exon numbers are approximate and vary between transcript annotations; treat them as indicative of which half of AF9 is retained rather than as exact.
PDB noteNo structure of the KMT2A::MLLT3 fusion protein exists. 2LM0 = solution structure of the AF4(AFF1)-AF9 complex, i.e. the AF9 AHD bound to an AFF1 CHD peptide - the one interface that is genuinely inside the fusion's mechanism; 4TMP = AF9 YEATS domain bound to an H3K9ac peptide, a domain of wild-type AF9 that is generally NOT retained in the fusion and is shown here for contrast; 7UJ4 and 4GQ6 = menin with revumenib and with an MLL peptide, the partner-agnostic drugged node. Verified against the RCSB entry API on 2026-09-18.
druggability classemerging
druggability noteCovered by the partner-agnostic menin inhibitor class (clinically validated for KMT2A-r as a whole), but the partner-specific nodes - DOT1L and the AF9 AHD - have respectively underperformed in the clinic and have no clinical molecule at all.
stage of hematopoiesiscommitted granulocyte-macrophage progenitor (GMP) reprogrammed into a leukemia stem cell; a later and more lineage-restricted origin than KMT2A::AFF1, and correspondingly myeloid
age peak1-10 y (paediatric AML peak), with substantial infant and adult AML representation
developmental order rank5
developmental order noteRanked 5 of 10 - later than AFF1 (fetal HSC/LMPP), MLLT1 and MLLT10 because the best-characterised transforming cell is a committed GMP rather than a stem or lympho-myeloid-primed cell; the rank reflects a murine model, not a demonstrated human cell of origin.
domain name notedomains_retained_5 / domains_lost_5 use the exact domain names from data/graph/nodes/king/gene-kmt2a.json so the dashboard's 3D domain ruler can match them. The LEDGF/PSIP1-binding region is retained biologically but has no lane in the parent node, so it is recorded in each breakpoint's phenotype_notes instead.
frequency
kmt2a_r_all_pct11.8kmt2a_r_all_pct: 11.8all_kmt2a_r_acute_leukemi…18.0all_kmt2a_r_acute_leukemia_pct: 18kmt2a_r_adult_aml_pct24.5kmt2a_r_adult_aml_pct: 24.5kmt2a_r_infant_aml_pct28.4kmt2a_r_infant_aml_pct: 28.4kmt2a_r_aml_pct30.4kmt2a_r_aml_pct: 30.4kmt2a_r_paediatric_aml_pct36.6kmt2a_r_paediatric_aml_pct: 36.6n_cases_dcal_cohort612.0n_cases_dcal_cohort: 612
diseases
[
  "paediatric AML",
  "infant AML",
  "adult AML",
  "therapy-related AML",
  "KMT2A-r ALL (minority)",
  "MPAL"
]
frequency — the numbers behind the chart
{
  "all_kmt2a_r_acute_leukemia_pct": 18,
  "kmt2a_r_aml_pct": 30.4,
  "kmt2a_r_all_pct": 11.8,
  "kmt2a_r_infant_aml_pct": 28.4,
  "kmt2a_r_paediatric_aml_pct": 36.6,
  "kmt2a_r_adult_aml_pct": 24.5,
  "n_cases_dcal_cohort": 612
}
breakpoints
[
  {
    "id": "bp-kmt2a-intron-9-mllt3",
    "gene5_exon_last": 9,
    "gene3_exon_first": 5,
    "frequency_pct": 37.2,
    "population": "KMT2A-r AML across ages; frequency_pct is the share of all KMT2A breaks in BCR subregion A (exon 9 + intron 9), the subregion AML prefers, not the share of MLLT3 fusions",
    "domains_retained_5": [
      "MBM1",
      "MBM2",
      "AT-hook 1",
      "AT-hook 2",
      "AT-hook 3",
      "SNL1/2",
      "CXXC"
    ],
    "domains_lost_5": [
      "PHD1",
      "PHD2",
      "PHD3",
      "Bromodomain",
      "PHD4",
      "FYRN",
      "TAD",
      "FYRC",
      "WIN",
      "SET"
    ],
    "domains_retained_3": [
      "AHD (ANC1 homology domain; binds AFF1/AFF4, DOT1L, CBX8, BCoR)",
      "AF9 central intrinsically disordered region"
    ],
    "domains_lost_3": [
      "YEATS domain (acetyl-/crotonyl-lysine reader)"
    ],
    "phenotype_notes": "Myelomonocytic AML (FAB M4/M5) with extramedullary disease. gene3_exon_first is approximate: MLLT3 breakpoints occur at more than one site, so no single exon pair describes all cases. CORRECTED 2026-09-19: this note previously added \"and about half of MLLT3 fusions in the DCAL series are spliced fusions\", which is wrong - that is the MLLT1 figure. In MLLT3 spliced fusions are 2 of 612 (0.3%). The retained KMT2A half also contains the LEDGF/PSIP1-binding region (no counterpart lane in the parent gene-kmt2a domain map, so it is recorded here rather than in the domain arrays)."
  },
  {
    "id": "bp-kmt2a-intron-11-mllt3-therapy",
    "gene5_exon_last": 11,
    "gene3_exon_first": 5,
    "frequency_pct": 39.5,
    "population": "therapy-induced KMT2A::MLLT3 cases shift toward intron 11; frequency_pct is the share of all KMT2A breaks in BCR subregion C",
    "domains_retained_5": [
      "MBM1",
      "MBM2",
      "AT-hook 1",
      "AT-hook 2",
      "AT-hook 3",
      "SNL1/2",
      "CXXC"
    ],
    "domains_lost_5": [
      "PHD1",
      "PHD2",
      "PHD3",
      "Bromodomain",
      "PHD4",
      "FYRN",
      "TAD",
      "FYRC",
      "WIN",
      "SET"
    ],
    "domains_retained_3": [
      "AHD",
      "AF9 central intrinsically disordered region"
    ],
    "domains_lost_3": [
      "YEATS domain"
    ],
    "phenotype_notes": "Therapy-related AML after topoisomerase-II-directed chemotherapy; the intron-11 shift in therapy-induced disease is a cohort-level finding across partners. The retained KMT2A half also contains the LEDGF/PSIP1-binding region (no counterpart lane in the parent gene-kmt2a domain map, so it is recorded here rather than in the domain arrays)."
  }
]
PDB IDS
[
  "2LM0",
  "4TMP",
  "7UJ4",
  "4GQ6"
]
mechanism class
[
  "chromatin-elongation-hijack"
]
drugs
[
  "drug-revumenib",
  "drug-ziftomenib",
  "drug-pinometostat"
]
key interactors
[
  "gene-dot1l",
  "gene-aff1",
  "gene-aff4",
  "gene-cbx8",
  "gene-men1",
  "gene-psip1",
  "gene-hoxa9",
  "gene-meis1"
]
trials
[
  "NCT04065399"
]
How this could bite us back
  • Resistance: menin-inhibitor resistance via MEN1 binding-site mutations applies to this fusion identically, since the drugged interface is on the shared KMT2A half.
  • Compensation: DOT1L is the only H3K79 methyltransferase, yet its inhibition is well tolerated and poorly effective - evidence that the elongation program is buffered by redundant routes (SEC-driven P-TEFb release, BRD4, acetylation readers) which will also buffer AHD-directed drugs.
  • Essential-function toxicity: DOT1L and H3K79me2 are required in normal haematopoiesis and in cardiac and neural tissue; the AF9 AHD also serves wild-type AF9 in normal transcription, so an AHD ligand will not be fusion-selective.
  • Lineage/phenotype plasticity: t(9;11) disease includes myelomonocytic AML, KMT2A-r ALL and MPAL, so any antigen-directed immunotherapy faces a genuine possibility of lineage switch, and menin-inhibitor-induced differentiation in a monocytic blast population carries a specific differentiation-syndrome hazard.
  • Delivery: the pinometostat experience shows that a validated epigenetic target can be defeated by pharmacokinetics alone - continuous intravenous infusion is not a viable long-term paediatric modality.
Still open
  1. Pinometostat proved DOT1L engagement in patients and still barely worked. Was the failure kinetic (H3K79me2 decays too slowly relative to leukemic proliferation), pharmacological (continuous infusion, adults only), or biological (the fusion program does not actually require ongoing H3K79 methylation once established)? Distinguishing these decides whether DOT1L belongs in combinations or should be retired.
  2. The AF9 AHD is a single disordered surface competed for by AFF1/AFF4, DOT1L, CBX8 and BCoR. Does the relative occupancy of that hub differ between t(9;11) AML and t(9;11) ALL, and would a ligand that biases occupancy toward the repressive partners (CBX8/BCoR) be therapeutic rather than merely inhibitory?
  3. Why is MLLT3 the dominant AML partner while AFF1 is essentially absent from AML? If this is cell-of-origin driven rather than protein-property driven, the same AF9 fusion introduced into a fetal lympho-myeloid progenitor should produce lymphoid disease - a directly testable prediction.
  4. Is the intermediate prognosis of t(9;11) an average over two biologically distinct groups (for example, HOXA-high intron-9 versus HOXA-low intron-11 cases), and does that split survive modern MRD-directed therapy?
Every source behind this entry (6)
  1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    KMT2A::MLLT3 accounts for 612/3401 (18.0%) of genomically characterised KMT2A rearrangements and about 30.4% of KMT2A-rearranged AML versus 11.8% of KMT2A-rearranged ALL, with 28.4% of infant AML, 36.6% of paediatric AML and 24.5% of adult AML KMT2A-r cases.

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

    Referral-based single-centre cohort; AML ascertainment in particular depends on which groups send samples for breakpoint typing, and roughly a third of AML cases in older series never had a partner identified. Percentages are of cases with an identified partner.

    Evidence that argues against this (2)
    • The AML-versus-ALL split is the least robust number here, because lineage assignment in a referral series is made by the referring centre's immunophenotyping panel rather than centrally. KMT2A-rearranged disease is exactly where that matters: it accounts for about 10% of acute leukaemias of ambiguous lineage, mostly infants, so cases at the boundary are assigned inconsistently and a 30.4%-versus-11.8% contrast inherits that inconsistency. SOURCING CORRECTION 2026-09-20 (Q42 citation-integrity repair). SOURCING NOT FOUND for the figure 'about 10% of acute leukaemias of ambiguous lineage': it is not in the recombinome this note cites, not in the iBFM-AMBI2012 report (PMID 29720486, abstract re-read 2026-09-20), and a PubMed search on 2026-09-20 did not resolve it. It is recorded here as unsourced rather than removed, because the argument it supports does not depend on the exact number. What the cited paper does support is given in the note below, and it points the same way. Leukemia 2023
    • Supporting the ambiguous-lineage argument above from the cited paper itself, read from PMC10169636 on 2026-09-20: 61 of the 3,401 KMT2A-rearranged patients were diagnosed as MPAL, and 35 of those 61 were infants. So lineage-ambiguous disease is concentrated in exactly the age group that dominates the AFF1 and MLLT3 counts, and a 30.4%-versus-11.8% AML/ALL contrast drawn from centre-assigned phenotypes inherits whatever those 61 assignments got wrong. This is the direction the paper measures (what fraction of KMT2A-r is MPAL); the converse figure quoted above is the one that could not be sourced. Meyer C et al. 2023
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    The AF9 YEATS domain reads acetylated histone H3 and physically links that acetylation to DOT1L-mediated H3K79 methylation, and AF9 is the docking module through which DOT1L is recruited to MLL-fusion target genes.

    PMID 25417107DOI
    What this evidence cannot show

    This paper characterises wild-type AF9 and its YEATS domain, which is generally NOT retained in the KMT2A::MLLT3 fusion protein; the fusion inherits the C-terminal AHD instead. The DOT1L-recruitment claim therefore transfers to the fusion, but the acetylation-reading step does not. Biochemistry plus cell lines, no patient material.

    Evidence that argues against this (1)
    • The AF9 YEATS-to-DOT1L axis is the best-characterised effector arm of this fusion and the clinical test of its downstream node bounds what it is worth. Pinometostat, a first-in-class DOT1L inhibitor, was given to 51 adults with advanced acute leukaemia; H3K79 methylation fell pharmacodynamically, the maximum tolerated dose was never reached, and two patients achieved complete remission - both with t(11;19), neither with t(9;11). The authors describe single-agent efficacy as modest. Reading the AF9-YEATS-DOT1L chain as the therapeutic handle for KMT2A::MLLT3 requires explaining why the partner most identified with DOT1L recruitment produced no reported single-agent response. Stein EM et al. 2018
  3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    MLL-AF9 transforms a committed granulocyte-macrophage progenitor into a leukemia stem cell by reactivating a self-renewal program, placing the cell of origin later in haematopoiesis than the fetal progenitor implicated in KMT2A::AFF1 infant ALL.

    Transformation from committed progenitor to leukaemia stem cell initiated by MLL-AF9 · Krivtsov AV, Twomey D, Feng Z, et al. · Nature 2006
    PMID 16862118DOI
    What this evidence cannot show

    A murine knock-in model; the cell of origin in human t(9;11) AML is not directly demonstrated, and later work has shown that HSCs can also be transformed by MLL-AF9 with different latency and phenotype, so 'GMP origin' describes a permissive cell rather than the only one.

    Evidence that argues against this (1)
    • This is a direct disagreement in the literature and the entry states only one side of it. Under ENDOGENOUS regulatory control, Mll-AF9 efficiently transformed LSK stem cells while committed granulocyte-monocyte progenitors were transformation-resistant and did not cause leukaemia; Mll-AF9 was expressed at higher levels in stem cells than in GMPs, and GMPs were efficiently transformed only at the higher dosage produced by retroviral transduction. On that reading the cell of origin is the stem cell and the GMP result is a dosage artefact. The two papers cannot both be describing the same experiment, and which is right decides whether the 'cell of origin is later in haematopoiesis' claim survives. Chen W et al. 2008
  4. E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In 756 children with 11q23/KMT2A-rearranged AML the 5-year event-free survival was 44% +/- 5% with subgroup EFS ranging from 11% to 92%, and the previously reported favourable prognosis of t(9;11)(p22;q23) could not be confirmed, whereas t(6;11) and both t(10;11) subgroups were independently adverse.

    PMID 19528532DOI
    What this evidence cannot show

    Retrospective pooling of 11 study groups treated on different protocols across many years, with karyotype-based rather than molecular subgroup assignment; some subgroups are very small, which is why the EFS range is so wide. Pre-dates menin inhibitors, gemtuzumab-era intensification and modern MRD-directed therapy, so absolute numbers are historical.

    Evidence that argues against this (1)
    • The subgroup boundaries in that analysis are cytogenetic and the molecular denominators have since been resolved differently. Genomic breakpoint characterisation across 3401 acute leukaemia patients catalogued 107 in-frame KMT2A fusions, 37 recurrent and 63 seen only once, with the seven commonest partners plus partial tandem duplications covering more than 90% of recombinations. An EFS range of 11% to 92% across karyotypic subgroups in a retrospective international cohort is computed over groups of very unequal size, and the inability to confirm a favourable prognosis for t(9;11) in that analysis is as much a statement about subgroup sizes and era as about the partner. Meyer C et al. 2023
  5. E5established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    Revumenib was evaluated and approved in KMT2A-translocated relapsed/refractory acute leukemia without restriction by fusion partner, achieving CR+CRh 22.8% and ORR 63.2% in the registration cohort.

    Menin Inhibition With Revumenib for KMT2A-Rearranged Relapsed or Refractory Acute Leukemia (AUGMENT-101) · Issa GC, Aldoss I, Thirman MJ, et al. · Journal of Clinical Oncology 2025
    PMID 39121437DOINCT04065399
    What this evidence cannot show

    Partner-specific response rates were not the primary analysis and the MLLT3 subgroup is not separately powered, so 'active in KMT2A::MLLT3' is an inference from the partner-agnostic design and the shared retained MBM module rather than a partner-stratified result.

    Evidence that argues against this (1)
    • The registration figure does not tell you what happens next, and the escape route most likely to matter is not the one that gets sequenced. Menin-inhibitor-resistant AML cells generated by repeated drug exposure carry an altered epigenome, transcriptome and proteome with NO menin mutation, and the resistant state is defined by druggable dependencies on BRD4, SMARCA4 and CREBBP; patient-derived MLL1-rearranged cells with ex vivo resistance behaved the same way, and combinations with a BET inhibitor or a dual BRG1/BRM inhibitor restored lethality. A CR+CRh of 22.8% quoted without a resistance model implies that non-responders are pharmacologically identical to responders, and they are not. Fiskus W et al. 2026
  6. E6established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    The DOT1L inhibitor pinometostat reduced H3K79 methylation in patients but produced only modest clinical activity as a single agent in adult acute leukemia.

    PMID 29724899DOI
    What this evidence cannot show

    Adults only, heavily pretreated, continuous-infusion administration with a difficult schedule, and no partner stratification; the trial does not exclude activity in a paediatric MLLT3 population or in combination. A negative monotherapy result is not a refutation of the DOT1L dependency itself.

    Evidence that argues against this (1)
    • Preclinically, genetic or pharmacological Dot1l inactivation abrogates MLL-AF10 and CALM-AF10 transformation, so the target dependency is real even though monotherapy underperformed. Chen L et al. 2013
Connected to