Oncofusion

start / graph / concept-flow-cytometry-immunophenotype-kmt2a-r

Explainerstrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.Q7-clinical

Step 2: flow cytometry and the CD10-negative pro-B fingerprint

also called immunophenotyping · flow cytometry · pro-B ALL · EGIL BI · CD10-negative ALL · NG2 · CSPG4 · 7.1 antigen · CALLA-negative

Within hours a machine can read which proteins sit on the surface of the leukaemia cells, and KMT2A-rearranged infant leukaemia has a very distinctive answer: it is missing a marker called CD10 that normal maturing B cells switch on, and it often wears myeloid markers it should not have. One extra marker, NG2, flags a broken KMT2A gene in about nine cases out of ten - but flow cytometry still cannot see the broken gene itself.

The same thing, for a scientist

Multiparameter flow cytometry gives the lineage assignment and maturation stage within hours of the diagnostic marrow. KMT2A-rearranged infant ALL characteristically presents as a pro-B / EGIL BI phenotype: CD19-positive, CD22 and CD79a-positive, usually CD34-positive and TdT-positive, and CD10-negative, with frequent aberrant co-expression of the myeloid antigens CD15, CD65 and CD33 [E1][E4][E5]. CD10-negativity reflects a differentiation arrest upstream of the common/pre-B stage at which CD10 (neprilysin/MME, the CALLA antigen) is normally induced, which is the immunophenotypic footprint of the cell-of-origin argument. The chondroitin sulfate proteoglycan NG2/CSPG4, detected by monoclonal antibody 7.1, is the single most informative surrogate marker: it was positive in 58 of 72 (81%) KMT2A-rearranged samples at a 20% cut-off in a 533-sample cohort [E2], and in adult B-precursor ALL reached sensitivity 0.89, specificity 0.89 and positive predictive value 0.93 for KMT2A rearrangement [E3]. It is nonetheless a surrogate: NG2-positive KMT2A-germline cases and NG2-negative KMT2A::AFF1 cases both exist [E2][E3], so flow cytometry raises the hypothesis fast and never closes it.

Go deeper
What flow cytometry measures, and why it comes back in hours

Flow cytometry works on a cell suspension, not a slide. Marrow or blood is incubated with panels of antibodies, each conjugated to a different fluorochrome, then the cells are streamed single-file past lasers; scattered light gives size and granularity, and emitted fluorescence gives the intensity of each antigen on each individual cell. Modern eight-to-twelve-colour panels measure a dozen markers simultaneously on tens of thousands of cells.

The practical consequence is timing. Staining and acquisition take a couple of hours and analysis another hour or two, so a lineage assignment and maturation stage are typically available the same working day as the marrow - before the cytogenetic culture has even been harvested. In a disease where the first days of treatment differ between ALL and AML backbones, and where the Interfant protocols deliberately hybridise the two because KMT2A-rearranged infant leukaemia sits at the lineage boundary 1, that speed is the point.

Flow also delivers the second thing the marrow is needed for: the immunophenotypic baseline against which post-induction minimal residual disease will later be measured.

The pro-B / EGIL BI pattern, and what CD10-negative physically means2 deeper

The EGIL classification divides B-lineage ALL by how far along normal B-cell differentiation the blast got before it froze. The most immature class, BI or pro-B ALL, is defined by expression of B-lineage-committing antigens - CD19 and/or CD79a and/or CD22 - with no expression of the later B differentiation markers, CD10 in particular 4. KMT2A-rearranged infant ALL lands overwhelmingly in this class 15.

The typical panel result reads:

MarkerStatus in KMT2A-r infant ALLWhat it means
CD19positiveB-lineage commitment has happened
CD22, CD79apositiveconfirms B lineage, including cytoplasmic CD79a
CD34usually positivean immature progenitor compartment
TdTusually positiveearly lymphoid
CD10negativethe blast never reached the common-B stage
CD15, CD65, CD33frequently positiveaberrant myeloid co-expression
NG2 / CSPG4 (7.1)frequently positivesurrogate flag for a broken KMT2A

The review literature gives the same panel from the other direction: expression of CD19 and CD34 with co-expression of CD15, CD33, CD65 and CD68, and often a lack of CD10, in a typically immature pro-B immunophenotype 5.

CD10 is an enzyme, not a label

CD10 is not an abstract classification token. It is neprilysin, a zinc-dependent membrane metallo-endopeptidase encoded by MME, historically called CALLA - the common acute lymphoblastic leukaemia antigen, because it was the first surface molecule found on most childhood ALL blasts. In normal B-cell ontogeny CD10 is induced as the cell passes from the earliest CD19-positive pro-B stage into the common-B / pre-B compartment where immunoglobulin heavy-chain rearrangement is completed.

So 'CD10-negative' has a physical meaning: the transforming event caught the cell before that transition, and the clone is locked there. The blast is not a mature B cell that lost a marker; it is a cell that never switched it on. That is the immunophenotypic signature of a very early progenitor origin, and it is the observable that anchors the cell-of-origin argument for KMT2A-rearranged infant leukaemia - a lesion arising in a fetal-stage progenitor that had barely entered the B lineage.

The corollary matters clinically as well as conceptually. In CCG 1953, multivariable analysis of 115 molecularly characterised infants ranked CD10-negativity as the strongest independent adverse prognostic factor, ahead of age under six months and ahead of KMT2A rearrangement itself (see the counter-evidence under 4). A differentiation-stage marker carrying more prognostic weight than the driver lesion is a strong hint that the arrested state, not merely the fusion, is what makes this disease hard to treat.

Aberrant myeloid co-expression, and why it is not a diagnostic error

A meaningful fraction of KMT2A-rearranged pro-B blasts co-express myeloid-associated antigens - CD15, CD65 and CD33 most often 145. In the original NG2 series, antigen-positive cases were significantly more likely to be CD10-negative and to express CD15 and CD65 1; in adult B-precursor ALL, NG2 positivity clustered tightly with a CD10-negative / CD24-negative / CD65s-positive / CD15-positive phenotype 3.

This is not sample contamination and not a scoring artefact. It is the immunophenotypic shadow of a fusion protein that operates in a progenitor with latent access to both lineage programmes. It is also the reason KMT2A-rearranged leukaemia keeps forcing the classification systems to add ambiguous-lineage categories, and the reason the Interfant backbone is a deliberate ALL/AML hybrid rather than a pure lymphoid regimen.

The diagnostic risk is specific: enough myeloid co-expression can push a case towards a mixed-phenotype or ambiguous-lineage label, and the label chosen on day two determines the chemotherapy backbone.

NG2 / CSPG4, the 7.1 antigen: the flow flag for a broken KMT2A

NG2 is the human homologue of the rat neuron-glial antigen 2, a chondroitin sulfate proteoglycan (CSPG4), detected in flow panels by monoclonal antibody 7.1. It is the closest thing flow cytometry has to a KMT2A read-out. The published performance:

CohortnFinding
Consecutive childhood ALL, prospective screen 11049 (8.6%) were NG2-positive; all 9 had t(4;11) or t(11;19) with MLL rearrangement, and none of the 95 NG2-negative cases had either translocation. MLL rearrangement was absent in all 89 NG2-negative samples tested, including 7 with del(11)(q23) or inv(11)(p13q23).
Adult and childhood acute leukaemias 2533At a 20% cut-off, 58 of 72 (81%) MLL-rearranged samples were 7.1-positive: 28/31 t(4;11), 16/17 t(9;11), 3/5 t(11;19), 2/6 del(11)(q23). All 7.1-positive ALL (n=34) and childhood AML (n=17) had MLL rearrangements, but 11 adults with 7.1-positive AML did not. Conclusion: sensitive but not entirely specific.
Adult ALL 3313NG2 in 57% of pro-B, 20% of pre-B and 2% of common ALL, absent in T-ALL and mature B-ALL. Against MLL rearrangement: sensitivity 0.89, specificity 0.89, positive predictive value 0.93. NG2 predicted MLL rearrangement with 100% accuracy in the subset that was CD65s-positive and/or CD15-positive.

Two failure directions are documented in the same papers. NG2 was positive in three adult patients with no detectable MLL rearrangement, and negative in eight patients who had demonstrable MLL::AF4 transcripts 3; eleven adult AMLs were 7.1-positive with germline MLL on Southern blot 2. So the marker moves the posterior probability hard, in both directions, and settles nothing.

A useful secondary property: NG2 is absent from non-malignant haematopoietic cells 3, which is why it was proposed as a minimal residual disease marker and, later, as a candidate immunotherapy target.

Why flow cytometry is fast but not definitive

Flow cytometry measures protein on a membrane. The diagnosis being made is about a chromosomal rearrangement. The gap between those two things cannot be closed by adding antibodies.

Concretely, flow cytometry cannot:

  • see the fusion. There is no antibody that distinguishes a KMT2A::AFF1 fusion protein from wild-type KMT2A in a routine panel. NG2 is a downstream consequence, and an imperfect one 23.
  • name the partner. Whether the 3' partner is AFF1, MLLT1, MLLT3 or one of the long tail of rare genes changes nothing measurable on the surface, yet it appears on trial eligibility forms.
  • exclude the lesion. An NG2-negative, CD10-positive pro-B case can still be KMT2A-rearranged 3, and CD10-positive KMT2A-rearranged infants do exist and behave differently [see counter-evidence under E4].
  • survive the disease changing lineage. The phenotype it measures is not fixed over time (see bite-back).

What flow cytometry does is generate a specific, testable hypothesis - this is a KMT2A-rearranged pro-B leukaemia - inside a few hours, with enough confidence to start the right induction and to order the right confirmatory genetics. The FISH break-apart probe and the molecular assays exist to settle what flow cytometry can only propose.

Correction: 'CD34 usually positive' is contradicted by the NG2-positive phenotype, and the ~90% NG2 figure is a review's

Two corrections to the table and the NG2 section above, both from records the coordinator fetched and read back on 2026-09-20.

1. CD34 is not reliably positive in the NG2-positive subset. The table says CD34 usually positive. PMID 12886247 associates NG2 expression in adult ALL with a CD10(-)/CD24(-)/CD65s(+)/CD15(+) phenotype, and PMID 38468826 independently measures significantly lower CD34 MFI in KMT2A-rearranged B-ALL along with lower CD22 and TdT. The subset this entry uses NG2 to flag is the subset in which CD34 is least dependable, so a reader using the table to build a panel is being pointed at the marker that fails on the cases the panel exists to catch.

2. The ~90% NG2 figure comes from a review and no primary measurement reaches it. Every primary series we fetched is lower, and the four designs are not comparable to one another:

recorddesignfigure
PMID 8562939 (Behm 1996)104 consecutive children, prospective screen9/9 NG2+ cases had t(4;11) or t(11;19); no NG2- case had either - a specificity result on nine cases
PMID 10914547 (Wuchter 2000)533 samples, adults and childrensensitivity 81%
PMID 12886247 (Schwartz 2003)313 adult ALL57% of pro-B ALL, 2% of common ALL, 20% of pre-B ALL
PMID 38468826 (Bisegna 2024)single-centre B-ALL panelPPV 96.6%

The 81% and the 96.6% are not the same quantity and must not be averaged: one is sensitivity, the other a positive predictive value. PPV is a property of the centre, not of the antigen - it moves with how much KMT2A-rearranged disease walks through the door. At a sensitivity and specificity of 0.89 each, PPV is 0.47 at 10% prevalence and 0.92 at 60%, computed by the coordinator. A referral centre's 96.6% does not transfer to a general paediatric service.

And the markers carrying the most diagnostic weight in PMID 12886247 are CD65s and CD15 - carbohydrate epitopes, not gene products. They have no entry here, and our named-but-unwritten instrument is keyed on HGNC symbols, so it structurally cannot see them. That is a blind spot in the instrument, recorded here because this is where it was found.

Facts and figures
assaymultiparameter flow cytometry, 8-12 colour panel
specimenbone marrow aspirate in EDTA or heparin; peripheral blood acceptable when blasts are circulating
egil classBI (pro-B ALL)
ng2 performance
Behm 1996, consecutive ch…104.0Behm 1996, consecutive childhood ALL: 104Schwartz 2003, adult ALL313.0Schwartz 2003, adult ALL: 313Wuchter 2000, adult and c…533.0Wuchter 2000, adult and childhood acute leukaemias: 533
turnaround hours
[
  3,
  8
]
markers
[
  {
    "marker": "CD19",
    "status": "positive",
    "note": "B-lineage commitment; also the target of blinatumomab and CD19 CAR-T, and therefore a marker that can be lost under pressure"
  },
  {
    "marker": "CD79a",
    "status": "positive",
    "note": "cytoplasmic; part of the EGIL BI definition alongside CD19 and CD22"
  },
  {
    "marker": "CD22",
    "status": "positive",
    "note": "B-lineage; part of the EGIL BI definition"
  },
  {
    "marker": "CD34",
    "status": "usually positive",
    "note": "immature progenitor compartment in infants; note that adult NG2-positive ALL was associated with CD34-NEGATIVITY, a genuine age-related discordance"
  },
  {
    "marker": "TdT",
    "status": "usually positive",
    "note": "terminal deoxynucleotidyl transferase; early lymphoid"
  },
  {
    "marker": "CD10",
    "status": "negative",
    "note": "neprilysin/MME, the CALLA antigen. Induced at the common-B/pre-B transition, so its absence places the arrest upstream of that stage. In CCG 1953 CD10-negativity was the strongest independent adverse prognostic factor in infant ALL"
  },
  {
    "marker": "CD24",
    "status": "often negative",
    "note": "reported as part of the NG2-associated phenotype in adult ALL"
  },
  {
    "marker": "CD15",
    "status": "frequently positive",
    "note": "aberrant myeloid co-expression; with CD65s it defined the subset in which NG2 predicted MLL rearrangement with 100% accuracy in adults"
  },
  {
    "marker": "CD65 / CD65s",
    "status": "frequently positive",
    "note": "aberrant myeloid co-expression"
  },
  {
    "marker": "CD33",
    "status": "variably positive",
    "note": "myeloid-associated; reported positive in paediatric KMT2A-r series but negative in the adult NG2-associated profile"
  },
  {
    "marker": "NG2 / CSPG4 (7.1 antigen)",
    "status": "frequently positive",
    "note": "surrogate flag for KMT2A rearrangement. 58/72 (81%) of MLL-rearranged samples positive at a 20% cut-off; sensitivity 0.89 / specificity 0.89 / PPV 0.93 in adult B-precursor ALL. Absent on normal haematopoietic cells"
  }
]
ng2 performance — the numbers behind the chart
[
  {
    "cohort": "Behm 1996, consecutive childhood ALL",
    "n": 104,
    "ng2_positive": 9,
    "concordance": "9/9 NG2-positive had t(4;11) or t(11;19) with MLL rearrangement; 0/95 NG2-negative had either"
  },
  {
    "cohort": "Wuchter 2000, adult and childhood acute leukaemias",
    "n": 533,
    "sensitivity_mll_rearranged_pct": 81,
    "detail": "58/72 MLL-rearranged samples 7.1-positive at 20% cut-off; 11 adult AMLs 7.1-positive without MLL rearrangement"
  },
  {
    "cohort": "Schwartz 2003, adult ALL",
    "n": 313,
    "sensitivity": 0.89,
    "specificity": 0.89,
    "ppv": 0.93,
    "detail": "NG2 in 57% pro-B, 20% pre-B, 2% common ALL; 3 NG2-positive without MLL rearrangement, 8 NG2-negative with MLL::AF4"
  }
]
what flow cannot do
[
  "detect the fusion protein or the chromosomal break",
  "identify which 3' partner gene is involved",
  "exclude a KMT2A rearrangement when NG2 is negative",
  "remain valid if the clone switches lineage"
]
How this could bite us back
  • Lineage switch makes a flow-based diagnosis perishable. KMT2A-rearranged infant B-ALL can convert to AML under CD19-directed pressure while retaining the same clone: a case with t(1;11)(p32;q23) switched from B-ALL to AML after blinatumomab, with identical IGH/IGL clonal sequences across the original B-ALL, the switched AML and a later CD19-positive B-ALL relapse. Any assay, therapy or MRD panel keyed to the day-one immunophenotype can be reading a phenotype the disease has already abandoned.
  • The switch can even reverse spontaneously, which means a single re-phenotyping at relapse may not capture the clone's actual range of states.
  • NG2 has two-sided failure: NG2-negative KMT2A::AFF1 cases and NG2-positive KMT2A-germline cases are both documented. A protocol that uses NG2 as a gate for confirmatory genetics will misroute patients in both directions.
  • CD19 is simultaneously the diagnostic anchor and the therapeutic target. Selecting for CD19-loss or myeloid escape degrades the very marker on which the diagnosis, the MRD panel and the immunotherapy all depend.
  • Aberrant myeloid co-expression can tip the case into a mixed-phenotype or ambiguous-lineage label, and that label chosen on day two determines which chemotherapy backbone the infant receives before any genetic confirmation exists.
  • A CD10-positive result can wrongly reassure against KMT2A rearrangement, and CD10-positive KMT2A-rearranged infants exist; the phenotype is a strong prior, not a rule-out.
  • A panel built from the table above will lean on CD34 in exactly the subset where CD34 is weakest, and a service quoting the ~90% NG2 figure to families or referrers is quoting a review number that no primary series reaches. At 10% prevalence a 0.89/0.89 test is right less than half the time it says positive - flow is a triage signal that must be confirmed by FISH or RT-PCR, and this entry's own final section says so.
Still open
  1. Do modern standardised multicolour panels (EuroFlow-style) applied to infants achieve better sensitivity and specificity for KMT2A rearrangement than monoclonal antibody 7.1 alone, and has anyone published that comparison in an infant-only cohort? The best sensitivity/specificity numbers available are still from an adult ALL series.
  2. Why are 10-20% of KMT2A-rearranged cases NG2-negative? Does NG2 expression track with the 3' partner gene, with the KMT2A intron 9 versus intron 11 breakpoint, or with the differentiation state of the transformed cell?
  3. CD10-negativity out-ranked KMT2A rearrangement itself as a prognostic factor in CCG 1953. Is the arrested pro-B state a causal driver of chemoresistance in its own right, or a proxy for something else (breakpoint position, cell of origin, fetal-stage gene programme)?
  4. NG2 is absent from normal haematopoietic cells, which makes it a theoretically attractive immunotherapy target. Why has no NG2-directed agent reached KMT2A-rearranged leukaemia, and is expression on non-haematopoietic tissue the blocker?
Every source behind this entry (7)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In 104 consecutive children with ALL screened prospectively with monoclonal antibody 7.1, all 9 NG2-positive cases (8.6%) carried t(4;11)(q21;q23) or t(11;19)(q23;p13) with MLL rearrangement, no NG2-negative case carried either translocation, and no MLL rearrangement was found in 89 NG2-negative samples including 7 with del(11)(q23) or inv(11)(p13q23); NG2-positive cases were significantly more likely to be infants, to have hyperleukocytosis and CNS involvement, to be CD10-negative and to express CD15 and CD65.

    PMID 8562939
    What this evidence cannot show

    Only 9 positive cases, so the apparent perfect concordance has very wide confidence intervals; single institution; Southern blot was the molecular reference and would itself miss some rearrangements. Notably the 7 cases with del(11)(q23) or inv(11)(p13q23) were NG2-negative and MLL-germline, so the marker tracks translocation rather than 11q23 involvement per se.

    Evidence that argues against this (1)
    • The apparent 100% specificity did not hold in larger mixed-age cohorts: 11 adults with 7.1-positive AML had germline MLL on Southern blot, so the marker is sensitive but not entirely specific. Wuchter C et al. 2000
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In 533 acute leukaemia samples from adults and children, 58 of 72 (81%) MLL-rearranged samples were monoclonal antibody 7.1-positive at a 20% cut-off (28/31 t(4;11), 16/17 t(9;11), 3/5 t(11;19), 2/6 del(11)(q23)); all 7.1-positive ALL and childhood AML cases had MLL rearrangements, but 11 adult AML cases were 7.1-positive with germline MLL, making 7.1 a sensitive but not entirely specific marker.

    Detection of acute leukemia cells with mixed lineage leukemia (MLL) gene rearrangements by flow cytometry using monoclonal antibody 7.1 · Wuchter C, Harbott J, Schoch C, Schnittger S, Borkhardt A, Karawajew L, Ratei R, Ruppert V, Haferlach T, Creutzig U, Dorken B, Ludwig WD · Leukemia 2000
    PMID 10914547DOI
    What this evidence cannot show

    The 20% positivity cut-off is an analytic choice that directly determines the reported sensitivity and specificity; a different cut-off would move both. MLL status was determined by Southern blot and RT-PCR, so cryptic rearrangements could be misclassified as germline, which would inflate the apparent false-positive rate. The ALL subset with MLL rearrangement is modest (72 samples across four rearrangement classes), so the per-translocation rates rest on small denominators.

    Evidence that argues against this (1)
    • A CD10-negative pro-B phenotype with NG2 expression is not sufficient on its own: in adult ALL, eight patients with demonstrable MLL::AF4 transcripts were NG2-negative, so a negative 7.1 result cannot exclude the lesion. Schwartz S et al. 2003
  3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In 313 adult ALL patients, NG2 was expressed in 57% of pro-B, 20% of pre-B and 2% of common ALL and absent in T-ALL and mature B-ALL; against MLL rearrangement it showed sensitivity 0.89, specificity 0.89 and positive predictive value 0.93, was associated with a CD10-negative/CD34-negative/CD24-negative/CD65s-positive/CD15-positive/CD13-negative/CD33-negative phenotype, predicted MLL rearrangement with 100% accuracy among CD65s-positive and/or CD15-positive cases, and is absent from non-malignant haematopoietic cells.

    PMID 12886247DOI
    What this evidence cannot show

    This is an ADULT ALL cohort; the sensitivity and specificity figures should not be transferred uncritically to infants, where the pro-B fraction and the KMT2A prevalence are both far higher, which raises the positive predictive value and lowers the negative predictive value. MLL status was determined in only 120 of the 313 patients. The paper reports NG2 associated with CD34-NEGATIVITY in adults, whereas infant KMT2A-rearranged pro-B ALL is typically CD34-positive - a real discordance, not a transcription error, and a reason to treat the adult marker profile as a related but distinct phenotype.

    Evidence that argues against this (1)
    • The same paper is its own counter-evidence: three NG2-positive patients had no detectable MLL rearrangement and eight NG2-negative patients had MLL::AF4 transcripts, so neither direction of the test is conclusive. source 2003
  4. E4established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    The EGIL immunological classification defines the most immature B-lineage class (BI, pro-B ALL) by expression of CD19 and/or CD79a and/or CD22 with no expression of later B differentiation markers, and provides a scoring system for distinguishing biphenotypic acute leukaemia from lineage-marker co-expression.

    Proposals for the immunological classification of acute leukemias. European Group for the Immunological Characterization of Leukemias (EGIL) · Bene MC, Castoldi G, Knapp W, Ludwig WD, Matutes E, Orfao A, van't Veer MB · Leukemia 1995
    PMID 7564526
    What this evidence cannot show

    A 1995 consensus proposal, explicitly framed as guidelines whose reproducibility the group intended to test later. It predates modern standardised multicolour panels and has been substantially superseded by WHO and ICC classifications for reporting, though 'pro-B' and 'BI' remain in everyday use and in the KMT2A literature. It is a classification convention, not a measurement.

    Evidence that argues against this (3)
    • CD10 status is not merely a classification label and is not uniformly negative. In CCG 1953, which molecularly characterised all 115 enrolled infants for MLL/11q23 rearrangement, multivariate Cox regression ranked CD10-negativity as the strongest independent adverse prognostic factor - ahead of age under six months and ahead of MLL rearrangement itself. That implies both that CD10-positive KMT2A-rearranged infants exist and that CD10 carries information the rearrangement status does not. Hilden JM et al. 2006
    • The immunophenotype the classification assigns is not stable over time. A 6-month-old with t(1;11)(p32;q23) KMT2A-rearranged B-ALL switched lineage to AML after blinatumomab, with the same three dominant IGH/IGL clonal sequences present in the original B-ALL, the lineage-switched AML and the later CD19-positive post-transplant B-ALL relapse - i.e. one clone reprogramming its lineage, not two diseases. Any EGIL class assigned at diagnosis can therefore be obsolete at relapse. Du J et al. 2021
    • The switch is not necessarily one-way: a KMT2A-rearranged infant ALL that switched to AML under blinatumomab was reported to revert spontaneously, illustrating how plastic this phenotype is and how little a single re-phenotyping at relapse constrains the clone's actual repertoire of states. Wolfl M et al. 2018
  5. E5moderate confidence 3 of 5: One strong study, or several smaller consistent ones.review

    KMT2A-rearranged paediatric ALL typically shows an immature pro-B immunophenotype with expression of CD19 and CD34 and co-expression of CD15, CD33, CD65 and CD68, often lacking CD10, and expresses NG2/CSPG4 in about 90% of cases.

    Updates in KMT2A Gene Rearrangement in Pediatric Acute Lymphoblastic Leukemia · Gorecki M, Koziol I, Kopystecka A, Budzynska J, Zawitkowska J, Lejman M · Biomedicines 2023
    PMID 36979800DOI
    What this evidence cannot show

    A narrative review; the 'about 90%' NG2 frequency is higher than the 81% measured in the largest primary cohort at a 20% cut-off [E2] and the 0.89 sensitivity reported in adults [E3], and the review does not state which primary series it derives from. Treat 81-90% as the plausible range rather than 90% as a point estimate.

    Evidence that argues against this (1)
    • "Typically" is doing heavy work. The phenotype described is the modal one for KMT2A::AFF1 pro-B disease and the beat's own comparative entries record that KMT2A-rearranged AML carries a different partner spectrum and a myeloid phenotype, while lineage switch under CD19-directed therapy moves patients out of this description entirely. An immunophenotype that changes under treatment is a snapshot, not a definition. Oncofusion knowledge graph (internal analysis) 2026
  6. E-cd34-not-reliable-in-ng2-positivemoderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    NG2 expression in ALL is associated with a CD10-negative, CD24-negative, CD65s-positive, CD15-positive B-cell phenotype, and KMT2A-rearranged B-ALL shows significantly lower CD34 mean fluorescence intensity than non-rearranged B-ALL, so CD34 positivity is not a dependable feature of the KMT2A-rearranged subset that NG2 identifies.

    PMID 12886247
    What this evidence cannot show

    Abstracts fetched and read back on 2026-09-20; full texts not read. **PMID 12886247 is an ADULT ALL series of 313 patients** and this entry is about a disease whose index population is infants, so the phenotype association is carried across an age boundary. **Lower MFI is a group-level shift, not a rule about any one patient**, and neither record claims CD34 is usually negative - the correction is that 'usually positive' overstates what the measurements support in this subset.

    Evidence that argues against this (1)
    • CD34 positivity in KMT2A-rearranged infant ALL is reported often enough elsewhere that the table's claim is not invented, and a lower mean does not make the marker useless. The defensible correction is to drop the word 'usually' and record CD34 as variable in this subset, not to invert it. Bisegna ML et al. 2024
  7. E-ng2-figures-not-comparablestrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Published NG2 performance figures in KMT2A-rearranged leukaemia measure different quantities in different populations: 9 of 9 NG2-positive cases carried t(4;11) or t(11;19) in a prospective screen of 104 children, sensitivity was 81% across 533 adult and paediatric samples, 57% of pro-B ALL was NG2-positive in 313 adults, and a single-centre B-ALL panel reported a positive predictive value of 96.6%.

    PMID 10914547
    What this evidence cannot show

    Four abstracts fetched and read back on 2026-09-20. **Strong as a statement that the figures are heterogeneous and non-comparable**, which is what it claims - it is not a pooled estimate and must not be read as one. The antibody (7.1) is shared across the series, so reagent differences are not the explanation; design and population are. **No meta-analysis of these series exists**, and producing one would require case-level data we do not have.

    Evidence that argues against this (1)
    • Heterogeneity across designs does not mean the marker is unreliable - all four series agree on direction and on high specificity, which is the property the clinical use depends on. The correction is about how the numbers are quoted, not about whether NG2 works. Behm FG et al. 1996
Connected to