Oncofusion

start / graph / mechanism-kmt2a-mpal-mixed-phenotype

Mechanismstrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.Q4-cells

Mixed-phenotype acute leukemia and KMT2A rearrangement — plasticity caught in the act

also called MPAL · mixed phenotype acute leukemia · B/myeloid MPAL · acute leukemia of ambiguous lineage · ALAL · biphenotypic leukemia · MPAL with KMT2A rearrangement

Some leukemias refuse to pick a side: the same blasts carry both B-cell and myeloid markers at once, and doctors call this mixed-phenotype acute leukemia. KMT2A-rearranged leukemia is one of the classic causes, and it is probably the same underlying flexibility that later lets these cells switch lineage entirely to escape treatment.

The same thing, for a scientist

Mixed-phenotype acute leukemia (MPAL) sits inside the broader family of acute leukemias of ambiguous lineage (ALAL) and is diagnosed when a single blast population, or two coexisting populations, satisfy lineage-defining immunophenotypic criteria for more than one lineage. In WHO 5th edition and the 2022 International Consensus Classification, MPAL is subdivided first by genetics — MPAL with BCR::ABL1 and MPAL with KMT2A rearrangement being the defining-genetic subtypes — and only then by immunophenotype, with AML entities taking diagnostic precedence over ALAL. ALAL with KMT2A rearrangement accounts for roughly 10% of cases, occurs mostly in infants and children, is almost always B/myeloid, and carries the worst outcome of any genetic MPAL subgroup. Single-cell and genomic work converges on a single answer to the 'one plastic cell or two clones' question: the immunophenotypic heterogeneity of MPAL is largely independent of somatic genetic variation, founding lesions arise in primitive haematopoietic progenitors, individual phenotypic subpopulations can reconstitute the full immunophenotypic diversity in vivo, and MPAL blasts share a stem-cell-like transcriptional program. The therapeutic corollary is settled by three independent datasets: ALL-type induction outperforms AML-type and hybrid regimens.

Go deeper
How common, and how common is KMT2A within it1 deeper

MPAL is rare. In a population-based Polish study covering all paediatric oncology centres from 2007 to 2018, 39 of 2,893 (1.35%) children and adolescents diagnosed with acute leukemia fulfilled WHO criteria for MPAL 7.

Within MPAL, the genetic subgroups have been mapped. BCR::ABL1 rearrangement accounts for 15-20% of MPAL and is associated with older age and a B/myeloid phenotype. ALAL with KMT2A rearrangement is found in approximately 10% of cases, mostly in infants and children, with expected 5' partners AFF1 (AF4), MLLT3 (AF9) and MLLT1 (ENL); most present as B/myeloid, with rare KMT2A-r acute undifferentiated leukemia and B/T MPAL reported. B/myeloid is the commonest immunophenotypic combination overall (~two-thirds), followed by T/myeloid 2.

Outcome tracks genetics, and KMT2A is the bad end. In a SEER analysis of 241 patients with WHO-2008-defined MPAL, Philadelphia-positive MPAL had a reduced risk of death relative to other MPAL (HR 0.28, P = .002), while KMT2A-rearranged MPAL had the worst outcome of all, with a 10-fold increased risk of death relative to Ph-positive MPAL (HR 10.2, P < .001) 2.

The classification is genetics-first, and this matters for anyone counting cases

In the WHO 5th edition, acute leukemias of ambiguous lineage are arranged into two families: ALAL with defining genetic abnormalities — which contains exactly 'Mixed-phenotype acute leukaemia with BCR::ABL1 fusion' and 'Mixed-phenotype acute leukaemia with KMT2A rearrangement' — and ALAL, immunophenotypically defined, which contains MPAL B/myeloid, MPAL T/myeloid, MPAL rare types, ALAL not otherwise specified, and acute undifferentiated leukemia. Two further subtypes sit under 'other defined genetic alterations': MPAL with ZNF384 rearrangement (found in ~50% of paediatric B/myeloid MPAL) and ALAL with BCL11B rearrangement (~20-30% of T/myeloid MPAL) 1.

A critical diagnostic-hierarchy rule follows: other WHO-defined AML entities take precedence over ALAL, so ALAL can only be diagnosed after excluding well-defined AML subtypes that may themselves show mixed phenotypes 2. Practically, that means the same biological cell can be reported as 'MPAL with KMT2A rearrangement', 'AML with KMT2A rearrangement' or 'KMT2A-r B-ALL with myeloid antigen expression' depending on the pathologist's threshold and the classification edition in force. Any prevalence figure for KMT2A-r MPAL is therefore partly an artefact of nomenclature.

The WHO 5th edition lineage-assignment criteria, verbatim in substance1 deeper

Lineage is assigned on intensity and pattern relative to the most similar normal population, not on binary positivity, and a coordinated pattern of several same-lineage antigens is more specific than any single antigen (the example given is CD19 + CD22 + CD10 together being more strongly B-associated than each alone) 1.

B lineage — either:

  • CD19 strong (intensity in part exceeding 50% of normal B-cell progenitor level by flow cytometry) plus ≥1 of CD10, CD22 or CD79a also strongly expressed; or
  • CD19 weak (not exceeding 50% of normal B-cell progenitor level) plus ≥2 of CD10, CD22 or CD79a strongly expressed.
  • CD79a may not be used if T lineage is under consideration.

T lineage — cytoplasmic or surface CD3 (anti-CD3 epsilon chain antibody) with intensity in part exceeding 50% of the mature T-cell level by flow cytometry, or immunocytochemistry positive with a non-zeta-chain reagent.

Myeloid lineage — either:

  • Myeloperoxidase with intensity in part exceeding 50% of the mature neutrophil level; or
  • monocytic differentiation, defined as ≥2 of non-specific esterase, CD11c, CD14, CD64 or lysozyme 1.

For B/myeloid MPAL, therefore, the requirement is one qualifying B-lineage pattern plus one qualifying myeloid pattern. Note that the MPO criterion is relative (50% of a mature neutrophil), not a percentage-of-blasts cut-off; proposed flow-cytometric thresholds for MPO positivity in the literature range from 3% to 28% of blasts, the historical 3% cytochemistry cut-off has high sensitivity but poor specificity by flow, a >10% threshold appears to improve specificity, and no consensus cut-off has been established 1. That unresolved threshold is the single largest source of inter-laboratory disagreement about whether a KMT2A-r case is 'B-ALL with myeloid markers' or 'B/myeloid MPAL'.

Biphenotypic versus bilineal

MPAL can present as a single blast population co-expressing markers of more than one lineage (biphenotypic) or, less often, as two discrete blast populations of different lineage assignment (bilineal); for the bilineal form a total of ≥20% blasts must be demonstrated in aggregate across the populations 2. The distinction is exactly the phenotypic shadow of the 'one plastic cell versus two clones' question, and it is why the single-cell data below matter so much.

One plastic cell or two clones? The single-cell and genomic answer1 deeper

Four independent lines of evidence say the ambiguity is a property of one cell, not a coincidence of two.

1. The heterogeneity is not genetic. Genomic profiling of MPAL showed that T/myeloid and B/myeloid MPAL are genetically distinct from each other (ZNF384 rearrangement common in B/myeloid; biallelic WT1 alterations in T/myeloid), but that the intratumoral immunophenotypic heterogeneity characteristic of MPAL is independent of somatic genetic variation. Founding lesions arise in primitive haematopoietic progenitors, and — the decisive experiment — individual phenotypic subpopulations sorted from a patient could reconstitute the full immunophenotypic diversity in vivo. The authors' conclusion is that the cell of origin and the founding lesion, rather than an accumulation of distinct genomic alterations, prime tumour cells for lineage promiscuity 3.

2. Genotype and transcriptome do not predict phenotype. Multiomic single-cell profiling of 14 newly diagnosed adult MPAL patients found that neither genetic profile nor transcriptome reliably correlates with specific MPAL immunophenotypes. Despite that, MPAL blasts express a shared stem-cell-like transcriptional profile indicative of high differentiation potential, patients with the highest differentiation potential had inferior survival, and a derived gene-set score (MPAL95) was predictive of survival in an independent cohort 4.

3. Regulatory programs are shared across a patient's phenotypic compartments. Integrated single-cell protein/transcriptome/chromatin-accessibility profiling of MPAL patients, deconvolved against a normal blood-development epigenetic baseline, identified patient-specific regulatory features that are shared across the phenotypic compartments of individual patients alongside common malignant signatures across patients 5. Shared regulatory wiring across compartments is what you expect from one cell in different states, not from two unrelated clones.

4. In KMT2A-r MPAL specifically, the epigenome picks a side even when the surface does not. Epigenetic profiling reportedly shows that in KMT2A-r MPAL, epigenetic patterns often align with a single dominant lineage identity, typically myeloid or lymphoid, despite immunophenotypic ambiguity 2. If true, the mixed phenotype in KMT2A-r cases is a surface-level failure of commitment sitting on top of a chromatin state that has largely committed — which is precisely the configuration from which a lineage switch is a short step.

The link to lineage switch is direct, not analogical

MPAL is one of the destinations of lineage switch, and myeloid co-expression is the precursor state. In the international Project EVOLVE cohort of post-immunotherapy lineage switch, 17 of 70 cases (24.3%) switched from B-ALL to B/myeloid MPAL or acute undifferentiated leukemia rather than to frank AML. At baseline, of 62 patients with reviewable flow cytometry or immunohistochemistry, 39 (62.9%) already expressed at least one myeloid or monocytic defining or associated antigen, and 29 of those 39 (74.4%) harboured a KMT2A rearrangement; CD15 was expressed in 22 (35.5%), of whom 19 (86.4%) were KMT2A-r. Crucially, no patient met criteria for MPAL at baseline 6.

That is the cleanest available statement of the relationship. The same KMT2A-r cells that carry sub-diagnostic myeloid antigens at presentation are the ones that cross the MPAL threshold, or the AML threshold, under antigen pressure. MPAL and lineage switch are two readings of one continuum of differentiation plasticity, separated only by where a diagnostic line is drawn and when the sample was taken.

Treatment implications: ALL-type induction, and the KMT2A exception nobody has tested

Three independent datasets agree, and they are the strongest evidence in the whole MPAL literature.

Largest paediatric cohort (n = 233). The international iBFM-AMBI2012 retrospective study of 233 children with ALAL found 5-year EFS of 80% ± 4% with ALL-type primary therapy versus 36% ± 7.2% with AML-type and 50% ± 12% with combined-type treatment. Excluding ALL- or AML-specific gene fusions, CD19-positive leukemia had a 5-year EFS of 83% ± 5.3% on ALL-type therapy versus 0% ± 0% on AML-type and 28% ± 14% on combined-type. Superiority of ALL-type therapy held for single-population mixed-phenotype ALAL and for bilineal ALAL. No overall benefit of transplantation was documented. AML-type treatment was preferred only for the minority who were CD19-negative with no other lymphoid features, and a change of treatment type was recommended only for ≥5% blasts after induction 8.

Meta-analysis (1,499 patients; 1,351 analysable). Across 97 reports from 33 countries, AML induction was less likely to achieve complete remission than ALL regimens (WHO2008 OR 0.33, 95% CI 0.18-0.58; EGIL OR 0.18, 95% CI 0.08-0.40), confirmed on patient-level multivariable analysis (OR 0.45, 95% CI 0.27-0.77). Overall survival favoured starting with ALL therapy, and on patient-level multivariable analysis it was specifically the hybrid regimens that fared worst (HR 2.11, 95% CI 1.30-3.43) 9.

Centrally reviewed COG cohort (54 WHO2016 MPAL cases of 70 submitted). ALL induction achieved remission in 72% (28/39) versus 69% (9/13) for AML regimens; 5-year EFS and OS for the whole cohort were 72% ± 8% and 77% ± 7%; and for those receiving ALL chemotherapy alone without HSCT (n = 21), EFS was 75% ± 13% and OS 84% ± 11%. The COG task force concluded that ALL chemotherapy without HSCT may be the preferred initial therapy, reserving AML chemotherapy and/or HSCT for MRD-defined treatment failure 10.

The honest gap: none of these analyses is powered to tell you what to do with a KMT2A-rearranged MPAL, which is the subgroup with a 10-fold increased mortality hazard 2. The general recommendation is ALL-type induction; the subgroup with the worst outcome is also the subgroup with the most myeloid biology and the highest lineage-switch risk, and it has never been separately randomised. Menin inhibition is the obvious addition and is essentially untested here: only one patient with MPAL was enrolled in the AUGMENT-101 trial that led to revumenib's approval (that patient achieved a morphological remission) 2.

Facts and figures
mechanism classdifferentiation plasticity manifesting as simultaneous multi-lineage antigen expression
entity names who5
[
  "Mixed-phenotype acute leukaemia with BCR::ABL1 fusion",
  "Mixed-phenotype acute leukaemia with KMT2A rearrangement",
  "Mixed-phenotype acute leukaemia with ZNF384 rearrangement",
  "Acute leukaemia of ambiguous lineage with BCL11B rearrangement",
  "Mixed-phenotype acute leukaemia, B/myeloid",
  "Mixed-phenotype acute leukaemia, T/myeloid",
  "Mixed-phenotype acute leukaemia, rare types",
  "Acute leukaemia of ambiguous lineage, not otherwise specified",
  "Acute undifferentiated leukaemia"
]
diagnostic criteria who5
{
  "b_lineage": "CD19 strong (intensity in part >50% of normal B-cell progenitor by flow) plus >=1 of CD10, CD22, CD79a strongly expressed; OR CD19 weak (<=50% of normal B-cell progenitor) plus >=2 of CD10, CD22, CD79a strongly expressed. CD79a unusable if T lineage under consideration.",
  "t_lineage": "Cytoplasmic or surface CD3 (anti-CD3 epsilon) with intensity in part >50% of mature T-cell level by flow; OR immunocytochemistry positive with a non-zeta-chain reagent.",
  "myeloid_lineage": "Myeloperoxidase with intensity in part >50% of mature neutrophil level; OR monocytic differentiation defined as >=2 of non-specific esterase, CD11c, CD14, CD64, lysozyme.",
  "mpo_flow_threshold_debate": "Proposed cut-offs range 3-28% of blasts; historical 3% cytochemistry cut-off is sensitive but not specific by flow; >10% appears to improve specificity; no consensus cut-off established.",
  "bilineal_blast_requirement": ">=20% blasts in aggregate across the discrete populations",
  "hierarchy_rule": "WHO-defined AML entities take precedence over ALAL; ALAL is diagnosed only after excluding defined AML subtypes that may show mixed phenotypes."
}
frequencies
[
  {
    "metric": "MPAL among paediatric acute leukemias",
    "value": 1.35,
    "unit": "%",
    "n": 2893,
    "population": "population-based, Poland, 2007-2018",
    "source_evidence": "E7"
  },
  {
    "metric": "KMT2A rearrangement among ALAL/MPAL cases",
    "value": 10,
    "unit": "% (approximate)",
    "population": "mostly infants and children",
    "source_evidence": "E2"
  },
  {
    "metric": "BCR::ABL1 rearrangement among MPAL cases",
    "value": "15-20",
    "unit": "%",
    "source_evidence": "E2"
  },
  {
    "metric": "B/myeloid share of immunophenotypically defined MPAL",
    "value": "~67",
    "unit": "%",
    "source_evidence": "E2"
  },
  {
    "metric": "ZNF384 rearrangement among paediatric B/myeloid MPAL",
    "value": 50,
    "unit": "% (approximate)",
    "source_evidence": "E1"
  },
  {
    "metric": "post-immunotherapy lineage switches whose destination was MPAL/ALAL rather than AML",
    "value": 24.3,
    "unit": "%",
    "n": 70,
    "source_evidence": "E6"
  }
]
kmt2a partners in mpal
[
  "AFF1 (AF4)",
  "MLLT3 (AF9)",
  "MLLT1 (ENL)"
]
prognosis
{
  "kmt2a_r_mpal": "Worst of any genetic MPAL subgroup: 10-fold increased risk of death versus Ph-positive MPAL (HR 10.2, P<0.001) in a 241-patient SEER analysis",
  "ph_positive_mpal": "HR 0.28 for death versus other MPAL (P=0.002); comparable to Ph-positive ALL in matched analysis",
  "overall_paediatric": "5-year EFS 72%+/-8% and OS 77%+/-7% in the centrally reviewed COG cohort; 5-year OS 51.8% and EFS 44.2% in the Polish population-based cohort"
}
treatment implication
{
  "preferred_induction": "ALL-type",
  "evidence_strength": "three independent datasets (233-patient iBFM cohort, 1,351-patient meta-analysis, 54-patient centrally reviewed COG cohort) all favour ALL-type over AML-type induction",
  "hybrid_regimens": "worst performing on patient-level multivariable analysis (HR 2.11, 95% CI 1.30-3.43)",
  "transplant": "no overall benefit demonstrated in the largest paediatric cohort; reserve for MRD-defined treatment failure",
  "kmt2a_r_specific": "untested; not separately randomised in any dataset; only one MPAL patient was enrolled in the AUGMENT-101 revumenib registration trial",
  "mrd": "end-of-induction MRD positivity predicts EFS (HR 6.00) and OS (HR 9.57) in ALL-treated paediatric MPAL"
}
one cell vs two clones
{
  "verdict": "predominantly one plastic cell",
  "supporting_observations": [
    "intratumoral immunophenotypic heterogeneity is independent of somatic genetic variation",
    "founding lesions arise in primitive haematopoietic progenitors",
    "sorted individual phenotypic subpopulations reconstitute the full immunophenotypic diversity in vivo",
    "neither genotype nor transcriptome predicts immunophenotype; blasts share a stem-cell-like transcriptional program",
    "patient-specific regulatory features are shared across a patient's phenotypic compartments"
  ],
  "residual_uncertainty": "the bilineal presentation (two discrete blast populations) still permits a two-clone reading in individual cases, and shared regulatory wiring does not formally exclude two clones from a common pre-leukemic ancestor"
}
How this could bite us back
  • Diagnostic categories can be gamed by biology: a KMT2A-rearranged cell that adds myeloid antigens crosses from B-ALL into MPAL and then into AML without any new mutation, so the patient's protocol eligibility changes while the disease does not.
  • The one intervention with the strongest evidence in MPAL is ALL-type induction, i.e. lymphoid-directed pressure — which is exactly the pressure that drives lineage switch in KMT2A-rearranged disease. The optimal therapy for the category may be the worst therapy for this genotype, and no trial has tested that.
  • Adding CD19-directed immunotherapy to an ALL backbone in a leukemia that already carries myeloid antigens on most blasts is selecting on a phenotype the clone can discard.
  • MRD monitoring in MPAL inherits the lineage-gating problem: a panel built for the dominant phenotype at diagnosis becomes progressively less sensitive as the clone shifts compartments, and in MPAL both compartments exist from day one.
  • Menin inhibitors relieve a differentiation block. In a cell whose defining feature is that it has never committed to a lineage, forcing differentiation could plausibly resolve the ambiguity in either direction, and the myeloid direction is the one associated with dismal outcome.
Still open
  1. Should KMT2A-rearranged MPAL be treated as MPAL or as KMT2A-rearranged leukemia? Every dataset supporting ALL-type induction pools genetic subgroups, and the one subgroup with a reported 10-fold mortality hazard has never been separately analysed. A pooled analysis of KMT2A-r cases extracted from the iBFM, COG and SEER MPAL cohorts is a zero-bench-cost study that has not been done.
  2. Does a menin inhibitor work in KMT2A-rearranged MPAL? Exactly one MPAL patient entered the AUGMENT-101 registration trial. Since revumenib is approved for KMT2A-rearranged acute leukemia irrespective of lineage, and MPAL with KMT2A rearrangement is a WHO5 entity, the label arguably already covers these patients and nobody has reported a series.
  3. Is the reported alignment of KMT2A-r MPAL epigenomes with a single dominant lineage real, and does it predict which lineage the disease will relapse as? If chromatin already knows the answer that the surface markers do not, an ATAC or methylation assay at diagnosis could pre-assign induction type in a disease where induction type is the single biggest prognostic lever.
  4. Where exactly is the boundary between 'KMT2A-r B-ALL with myeloid antigen expression' and 'B/myeloid MPAL'? 62.9% of patients who later underwent lineage switch already expressed a myeloid or monocytic antigen at baseline yet none met MPAL criteria. With no consensus myeloperoxidase cut-off, this boundary is currently set by local laboratory practice, which makes every prevalence and outcome figure in this node partly a measurement artefact.
  5. Does the stem-cell-like differentiation-potential score derived in adult MPAL transfer to KMT2A-rearranged infant disease, where it would be a candidate biomarker for lineage-switch risk before any immunotherapy is given?
Every source behind this entry (10)
  1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.review

    The WHO 5th edition groups acute leukaemias of ambiguous lineage into 'ALAL with defining genetic abnormalities' — comprising MPAL with BCR::ABL1 fusion and MPAL with KMT2A rearrangement — plus subtypes with other defined genetic alterations (MPAL with ZNF384 rearrangement, found in ~50% of paediatric B/myeloid MPAL; ALAL with BCL11B rearrangement, ~20-30% of T/myeloid MPAL) and an immunophenotypically defined family; lineage assignment requires CD19 strong plus one or CD19 weak plus two of CD10/CD22/CD79a for B lineage, cytoplasmic or surface CD3 exceeding 50% of mature T-cell intensity for T lineage, and myeloperoxidase exceeding 50% of mature neutrophil intensity or two or more of non-specific esterase/CD11c/CD14/CD64/lysozyme for myeloid lineage; proposed flow-cytometric myeloperoxidase thresholds range from 3% to 28% of blasts with >10% improving specificity and no consensus cut-off established.

    PMID 35732831DOI
    What this evidence cannot show

    A consensus classification document, not primary data: the criteria are expert-derived and the classification itself states that no consensus myeloperoxidase cut-off exists, so two laboratories applying WHO5 faithfully can disagree on whether the same KMT2A-r case is MPAL. The 2022 International Consensus Classification is a parallel and partly divergent framework published the same year; I verified the WHO5 criteria from the open-access full text but could not obtain the ICC full text, so no ICC-specific criterion is asserted in this node beyond what a verified secondary source states. Criteria have also changed between editions, which makes historical MPAL cohorts non-comparable.

    Evidence that argues against this (1)
    • A WHO classification is a consensus decision about how to group cases, and the groupings change between editions. Recording that the 5th edition creates 'ALAL with defining genetic abnormalities' documents the current agreement; it is not evidence that the categories carve the biology, and the 4th edition drew the lines differently. Blood Cells, Molecules and Diseases 2010
  2. E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.review

    ALAL with KMT2A rearrangement accounts for approximately 10% of ALAL cases, mostly in infants and children, usually B/myeloid, with AFF1, MLLT3 and MLLT1 the commonest partners; BCR::ABL1 rearrangement accounts for 15-20% of MPAL; B/myeloid is the commonest phenotype at roughly two-thirds of cases; in a SEER analysis of 241 WHO-2008-defined MPAL patients, Ph-positive MPAL had a reduced risk of death (HR 0.28, P = .002) whereas KMT2A-rearranged MPAL had the worst outcome with a 10-fold increased risk of death relative to Ph-positive MPAL (HR 10.2, P < .001); AML-defined entities take diagnostic precedence over ALAL; in KMT2A-r MPAL epigenetic patterns often align with a single dominant lineage identity despite immunophenotypic ambiguity; and only one MPAL patient was enrolled in the AUGMENT-101 revumenib registration trial.

    Acute leukemia of ambiguous lineage: the known and the uncertain · Sherban A, Wolach O · Haematologica 2026
    PMID 41063702DOI
    What this evidence cannot show

    A narrative review, cited here as a map because the underlying primary sources for the ~10% KMT2A-r frequency and the SEER hazard ratios are secondary citations within it that I did not independently retrieve. The '~10%' figure is given without a denominator or cohort, and SEER-based MPAL series depend on registry coding of a diagnosis whose criteria changed twice in the study period. The statement that KMT2A-r MPAL epigenomes align with a dominant lineage is the review's summary of a single cited study (Milan et al., Haematologica 2020) that I did not read; treat it as a lead, not a result.

    Evidence that argues against this (1)
    • Percentages from a review inherit the case definitions of whatever series it aggregated, and MPAL frequency depends critically on the immunophenotypic criteria applied — which the WHO has revised repeatedly. "Approximately 10% of ALAL" is therefore a figure about classification practice as much as about disease. Blood Cells, Molecules and Diseases 2010
  3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Genomic profiling shows T/myeloid and B/myeloid MPAL are genetically distinct (ZNF384 rearrangement common in B/myeloid, biallelic WT1 alterations in T/myeloid), that the intratumoral immunophenotypic heterogeneity characteristic of MPAL is independent of somatic genetic variation, that founding lesions arise in primitive haematopoietic progenitors, and that individual phenotypic subpopulations can reconstitute the full immunophenotypic diversity in vivo — implicating cell of origin and founding lesion rather than accumulated genomic alterations in priming lineage promiscuity.

    The genetic basis and cell of origin of mixed phenotype acute leukaemia · Alexander TB, Gu Z, Iacobucci I, Dickerson K, Choi JK, et al. · Nature 2018
    PMID 30209392DOI
    What this evidence cannot show

    The in vivo reconstitution experiments were necessarily performed on a small number of patient samples in immunodeficient mice, which lack the T-cell compartment that shapes lineage selection in patients. The cohort is dominated by ZNF384-rearranged and WT1-altered cases; KMT2A-rearranged MPAL is not the focus and its representation is not stated in the abstract, so the conclusion about lineage promiscuity is a general MPAL finding extrapolated to the KMT2A-r subgroup here. I verified the abstract; the KMT2A-specific numbers should be read from the supplementary tables before being quoted.

    Evidence that argues against this (1)
    • The genetic distinction between T/myeloid and B/myeloid MPAL is established in a cohort where the KMT2A subset is small, and ZNF384 and WT1 are the discriminating lesions — neither of which is the beat's subject. Immunophenotypic heterogeneity being independent of genetics is the finding most relevant here and it argues that the phenotype the classification rests on is not genetically anchored. Oncofusion knowledge graph (internal analysis) 2026
  4. E4moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    Multiomic single-cell profiling of 14 newly diagnosed adult MPAL patients showed that neither genetic profile nor transcriptome reliably correlates with specific MPAL immunophenotypes, that MPAL blasts nonetheless express a shared stem-cell-like transcriptional profile indicating high differentiation potential, that patients with the highest differentiation potential had inferior survival, and that a derived gene-set score (MPAL95) predicted survival in an independent cohort.

    Multiomic single cell sequencing identifies stemlike nature of mixed phenotype acute leukemia · Peretz CAC, Kennedy VE, Walia A, Delley CL, Koh A, et al. · Nature Communications 2024
    PMID 39294124DOI
    What this evidence cannot show

    Fourteen adult patients. MPAL in adults is genetically different from paediatric and infant MPAL (more BCR::ABL1, less ZNF384 and KMT2A), so the direct relevance to KMT2A-rearranged infant disease is an extrapolation. The survival association of the differentiation-potential score is observational in a small discovery set with one independent validation cohort, and the MPAL95 score has not been applied prospectively.

    Evidence that argues against this (1)
    • Fourteen ADULT patients. MPAL with KMT2A rearrangement is overwhelmingly an infant and childhood disease, so a multiomic study in adults describes a differently constituted population, and a shared stem-cell-like transcriptional programme is what single-cell studies recover from most immature leukaemias. Oncofusion knowledge graph (internal analysis) 2026
  5. E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Single-cell integration of multiplexed protein quantification, transcriptome profiling and chromatin-accessibility analysis, deconvolved against a normal blood-development epigenetic baseline, identified common malignant signatures across MPAL patients as well as patient-specific regulatory features shared across the phenotypic compartments of individual patients, together with 91,601 putative peak-to-gene linkages and leukemia-specific regulatory transcription factors.

    Single-cell multiomic analysis identifies regulatory programs in mixed-phenotype acute leukemia · Granja JM, Klemm S, McGinnis LM, Kathiria AS, Mezger A, et al. · Nature Biotechnology 2019
    PMID 31792411DOI
    What this evidence cannot show

    Primarily a methods paper; the MPAL cohort is small and the number of patients is not stated in the abstract, and the authors emphasise widespread epigenetic heterogeneity across patients. 'Regulatory features shared across phenotypic compartments' is evidence consistent with a common cell of origin but does not by itself exclude two clones descended from a shared pre-leukemic ancestor. No KMT2A-rearranged subgroup analysis.

    Evidence that argues against this (1)
    • Deconvolution against a normal blood-development epigenetic baseline is the step that generates the signatures, and the baseline is adult. Projecting infant or fetal-origin blasts onto an adult reference will assign them to whichever adult stage they least mismatch, which manufactures apparent stage assignments. Oncofusion knowledge graph (internal analysis) 2026
  6. E6strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In the Project EVOLVE post-immunotherapy lineage-switch cohort, 17 of 70 cases (24.3%) switched from B-ALL to B/myeloid MPAL or acute undifferentiated leukemia rather than to AML; at baseline, 39 of 62 patients (62.9%) with reviewable immunophenotyping already expressed at least one myeloid or monocytic defining or associated antigen, of whom 29 (74.4%) harboured a KMT2A rearrangement, CD15 was expressed in 22 (35.5%) of whom 19 (86.4%) were KMT2A-rearranged, and no patient met criteria for MPAL at baseline.

    Project EVOLVE: an international analysis of postimmunotherapy lineage switch, an emergent form of relapse in leukemia · Silbert SK, Rankin AW, Hoang CN, Semchenkova A, Myers RM, et al. · Blood 2025
    PMID 40193715DOI
    What this evidence cannot show

    A solicited international case series with no denominator, so the 62.9% baseline myeloid-antigen figure describes patients who went on to switch and cannot be read as a risk estimate for KMT2A-r patients generally — the comparison group of KMT2A-r patients who did not switch was not assembled. Baseline immunophenotyping was reviewed centrally only where material was available (62 of 70). CD15 expression in KMT2A-r B-ALL may be aberrant rather than genuinely myeloid, as the authors note.

    Evidence that argues against this (1)
    • These are POST-IMMUNOTHERAPY switches, so the 24.3% figure describes what CD19-directed pressure produces rather than the natural history of KMT2A-rearranged MPAL. The 62.9% baseline immunophenotypic finding also depends on retrospective review of material collected for other purposes. Journal of Clinical Oncology 2025
  7. E7strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In a population-based study of all Polish paediatric oncology centres from 2007 to 2018, 39 of 2,893 (1.35%) children and adolescents diagnosed with acute leukemia fulfilled WHO criteria for MPAL, T/myeloid was the most prevalent phenotype, 32 patients achieved first complete remission including 23 (92.0%) of those treated with ALL-directed chemotherapy versus 9 (64.3%) of those treated with AML-type induction regimens, and 5-year overall and event-free survival were 51.8% and 44.2%.

    Mixed phenotype acute leukemia: Biological profile, clinical characteristic and treatment outcomes: Report of the population-based study · Zajac-Spychala O, Irga-Jaworska N, Drozynska E, Muszynska-Roslan K, Krawczuk-Rybak M, et al. · European Journal of Haematology 2020
    PMID 32223036DOI
    What this evidence cannot show

    Thirty-nine patients over eleven years; the treatment comparison is non-randomised, retrospective and confounded by why a given patient was assigned AML-type therapy. T/myeloid predominance here contrasts with the B/myeloid predominance reported elsewhere, likely reflecting local diagnostic practice and the small n. Genetic characterisation is limited (cytogenetic findings in only 2 of 39), so the KMT2A-rearranged fraction of this cohort is unknown.

    Evidence that argues against this (1)
    • Thirty-nine cases over eleven years across a whole country, and T/myeloid predominating, which is the opposite of the B/myeloid predominance the beat reports for KMT2A-rearranged ALAL. A population-based denominator is the right design and 39 cases cannot resolve subtype proportions. Statistics in Medicine 2006
  8. E8strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In the international iBFM-AMBI2012 retrospective study of 233 children with acute leukemia of ambiguous lineage, 5-year event-free survival was 80% +/- 4% with ALL-type primary therapy versus 36% +/- 7.2% with AML-type and 50% +/- 12% with combined-type treatment; excluding ALL- or AML-specific gene fusions, CD19-positive leukemia had a 5-year EFS of 83% +/- 5.3% on ALL-type therapy versus 0% +/- 0% on AML-type and 28% +/- 14% on combined-type; superiority of ALL-type treatment held for single-population mixed-phenotype and bilineal ALAL; and no overall benefit of transplantation was documented.

    International cooperative study identifies treatment strategy in childhood ambiguous lineage leukemia · Hrusak O, de Haas V, Stancikova J, Vakrmanova B, Janotova I, et al. · Blood 2018
    PMID 29720486DOI
    What this evidence cannot show

    Retrospective and multinational, so treatment assignment was not randomised: children given AML-type therapy may have been selected for more myeloid-appearing or sicker disease, which would inflate the apparent superiority of ALL-type therapy. The '0% +/- 0%' EFS for CD19-positive ALAL on AML-type therapy is a very small subgroup and should not be read as a precise estimate. The KMT2A-rearranged fraction and its subgroup outcome are not reported in the abstract, which is exactly the gap this node flags.

    Evidence that argues against this (1)
    • The 80% versus 36% comparison is retrospective and treatment assignment was not random: clinicians chose ALL-type therapy for patients who looked lymphoid, and looking lymphoid is itself prognostic. Confounding by indication is the whole difficulty of this literature and it is not addressed by the size of the effect. Statistics in Medicine 2006
  9. E9strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.review

    A systematic review and meta-analysis identifying 97 reports from 33 countries with 1,499 unique MPAL patients (1,351 analysable) found that AML induction was less likely to achieve complete remission than ALL regimens (WHO2008 OR 0.33, 95% CI 0.18-0.58; EGIL OR 0.18, 95% CI 0.08-0.40; patient-level multivariable OR 0.45, 95% CI 0.27-0.77), that overall survival favoured starting with ALL rather than AML therapy, and that on patient-level multivariable analysis it was hybrid therapy that fared worst (HR 2.11, 95% CI 1.30-3.43).

    PMID 29550836DOI
    What this evidence cannot show

    A meta-analysis of case series, so it inherits every selection bias of its inputs and cannot adjust for the clinical reasoning that led to a given induction choice. It pools children and adults and spans decades of supportive care. MPAL definitions used were EGIL and WHO2008, both superseded, so the population is not the WHO5/ICC-defined MPAL of today. No KMT2A-rearranged subgroup analysis.

    Evidence that argues against this (1)
    • A meta-analysis of 97 reports from 33 countries pools case series with different diagnostic criteria across two WHO editions, which the odds ratios cannot adjust away. Published case series also over-represent treated and reported patients, so the induction comparison carries the same confounding by indication in aggregated form. Statistics in Medicine 2006
  10. E10strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In a centrally reviewed Children's Oncology Group cohort, 54 of 70 submitted cases fulfilled WHO2016 MPAL criteria; ALL induction achieved remission in 72% (28/39) versus 69% (9/13) for AML regimens; 5-year event-free and overall survival for the whole cohort were 72% +/- 8% and 77% +/- 7%; and for the 21 patients receiving ALL chemotherapy alone without transplantation event-free survival was 75% +/- 13% and overall survival 84% +/- 11%, leading the ALAL Task Force to propose ALL chemotherapy without HSCT as preferred initial therapy with AML chemotherapy or HSCT reserved for MRD-defined treatment failure.

    PMID 31661160DOI
    What this evidence cannot show

    Fifty-four patients assembled from ALL and AML banking studies rather than treated on a common protocol, with treatment and outcome data captured separately — so the comparison of ALL versus AML induction rests on 39 versus 13 patients and is not a trial. The fact that only 54 of 70 submitted cases survived central review is itself a warning about how often MPAL is over-diagnosed locally. The proposed prospective COG trial had not reported at the time of writing.

    Evidence that argues against this (1)
    • MRD kinetics, not induction type alone, may be what determines outcome in paediatric MPAL: end-of-induction MRD positivity predicted 5-year EFS (HR 6.00, P < 0.001) and OS (HR 9.57, P = 0.003) in a centrally reviewed multicentre cohort uniformly treated with ALL regimens, and patients who cleared MRD only by end of consolidation did worse than those negative at end of induction. Oberley MJ et al. 2020
Connected to