Oncofusion

start / graph / comut-flt3-kmt2a-r

Genestrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.Q10-genome

FLT3 in KMT2A-r leukemia — near-universal overexpression, but activating mutation in only ~3% of infants

also called FLT3 in MLL-rearranged ALL · FLT3 overexpression MLL signature gene · FLT3 activation loop mutation infant ALL · CD135

FLT3 is a growth receptor that KMT2A-rearranged leukemia cells switch on very strongly — so strongly that the receptor fires even without its signal. That is different from FLT3 being mutated, which is what happens in adult AML. In infant KMT2A-rearranged leukemia the gene is almost never mutated: about 1 case in 36, with no internal tandem duplications at all.

The same thing, for a scientist

FLT3 in KMT2A-rearranged leukemia is a case study in confusing two distinct lesion classes. **Overexpression** is near-universal and is a defining feature of the MLL expression signature: FLT3 expression in 41 infant KMT2A-r cases was significantly higher than in germline-KMT2A infants (P < 0.001) and non-infant ALL (P = 0.001), and high-level expression of wild-type receptor produces **ligand-independent FLT3 phosphorylation**. **Activating mutation** is a different and much rarer matter: activation-loop mutations occurred in only ~3% (1 of 36) of infant KMT2A-r cases, with **no FLT3/ITD observed at all**. The commonly cited higher figure — 5 of 30 (17%) — comes from an earlier, differently composed MLL cohort, and the discrepancy between 17% and 3% has never been formally resolved. In non-infant KMT2A-r leukemia FLT3 is genuinely recurrent (n = 3 of 20), including two cases with the activating N676K allele. The therapeutic implication is that FLT3 in this disease is a **target by expression, not a target by genotype** — which is why FLT3-directed CAR T cells are a more coherent strategy here than FLT3 kinase inhibitors selected on mutation status.

Go deeper
Overexpression: near-universal, and the original reason anyone looked at FLT3

FLT3 entered this field through expression, not sequencing. KMT2A-rearranged ALL has a characteristic, highly distinct gene-expression profile — consistent with an early haematopoietic progenitor expressing select multilineage markers and individual HOXA genes — separable by clustering from both conventional ALL and AML, and the authors' explicit conclusion was that establishing MLL as a unique entity mandates the examination of selectively expressed genes for urgently needed molecular targets 1. FLT3 was the first gene pulled out of that signature.

The quantification came next 3:

  • FLT3 expression in 41 infants with KMT2A-r was significantly higher than in 8 infants with germline KMT2A (P < 0.001) and than in 23 non-infant ALL patients with germline KMT2A (P = 0.001).
  • Leukemic cells from infants with KMT2A-r were significantly more sensitive to the FLT3 inhibitor PKC412 (midostaurin) than non-infant ALL cells, and at least as sensitive as FLT3/ITD-positive AML cells — despite not carrying ITDs.
  • Measuring FLT3 phosphorylation across infants expressing varying levels of wild-type receptor showed that high-level FLT3 expression is associated with ligand-independent FLT3 activation 3.

That last mechanism is the crux. In KMT2A-r leukemia the receptor is constitutively active because there is so much of it, not because it is mutated. Autoactivation by density is a real and sufficient route to signalling, and it means genotype carries almost no information about whether the pathway is on.

Expression also carries prognostic weight independent of mutation. In 97 infant ALL cases from COG trial P9407, statistical modelling identified three genes predictive of event-free survival beyond age and KMT2A status — FLT3, IRX2 and TACC2 — and low FLT3 expression identified a group of 11 infants with a 5-year EFS of 100% 5. Nobody has reported that FLT3 mutation status stratifies outcome in this disease at all.

Mutation: the real number, and the discrepancy behind it1 deeper

This is where the literature is routinely misquoted. Two primary studies give different answers, both are real, and the field has not reconciled them.

studycohortactivation-loop mutationsITDsource
Armstrong 2003, Cancer Cell30 MLL-rearranged leukemias5/30 (17%) — three novel I836 deletions, two D835 mutationsnot reported2
Stam 2005, Blood36 infant KMT2A-r ALL1/36 (~3%)none observed3

The honest statement is the low one, for two reasons. First, Stam's cohort is explicitly infant KMT2A-r ALL, which is the population this beat is about; Armstrong's 30 'MLLs' were not restricted to infants and the composition is not specified in the abstract. Second, the low figure is independently corroborated. In the St. Jude PCGP cohort, FLT3 appears among the non-recurrent kinase/PI3K/RAS lesions in infant KMT2A-r ALL — i.e. one case of 22 whole-genome cases — while being genuinely recurrent at n = 3 of 20 in the non-infant KMT2A-r cohort 4. Two independent datasets therefore agree that FLT3 mutation is rare in infants and commoner in older children with the same fusion, which is the simplest explanation for the 17% versus 3% gap: cohort age composition. This node offers that as the likely explanation but does not assert it, because neither paper reports the age breakdown needed to test it.

Allele spectrum. The variants actually observed differ from the AML canon. The public curation of the PCGP study returns 5 FLT3 mutations in 5 of 93 sequenced samples: D839G, N676K, P934L and R595_E596insNI 6. Note what is absent: no D835, no I836del, and no juxtamembrane ITD. Two of these were functionally characterised. N676K — previously reported in an AML patient who developed resistance to PKC412 and as a recurrent mutation in core-binding-factor leukemia — was found in two PCGP cases, and both N676K and the novel PIK3R1 alleles were shown to be activating by conferring factor-independent growth on IL-3-dependent BaF3 cells 4. So N676K is the one FLT3 allele in this disease with both recurrence (2 cases) and functional validation.

The N676K identity is uncomfortable for a therapeutic strategy: it was originally described as a PKC412 resistance allele 4, and PKC412/midostaurin is precisely the FLT3 inhibitor to which KMT2A-r blasts were shown to be sensitive 3.

For contrast: where FLT3 mutation really is common in childhood ALL

FLT3 mutation in childhood ALL is not rare everywhere — it is concentrated in a different subtype. Roughly 25% (6 of 25) of hyperdiploid ALL samples carry FLT3 mutations, versus 1 of 29 ETV6::RUNX1-rearranged samples (P = 0.04, Fisher exact) 7. Three of those were novel in-frame deletions within a seven-amino-acid region of the receptor juxtamembrane domain.

That comparison sharpens the point: high FLT3 expression occurs in both hyperdiploid and KMT2A-r ALL, but FLT3 mutation is a hyperdiploid phenomenon, not a KMT2A-r one. Anyone quoting a double-digit FLT3 mutation rate for KMT2A-r infant ALL is, most likely, carrying over a number from an adjacent subtype.

Therapeutic consequence: target the protein, not the genotype

If the lesion is overexpression rather than mutation, three things follow.

1. Mutation-based patient selection is wrong here. Every FLT3 inhibitor programme in AML selects on ITD or TKD mutation. Applied to infant KMT2A-r ALL that would enrol roughly 3% of patients and exclude the ~97% whose receptor is constitutively active by density 3. Selection should be on expression.

2. Kinase inhibition may be insufficient on its own. The most important caveat in the founding paper is usually dropped: at concentrations of PKC412 minimally required to fully inhibit FLT3 phosphorylation, the cytotoxic effects were only fractional. The authors concluded that PKC412-induced apoptosis in infant KMT2A-r cells is unlikely to be a consequence of FLT3 inhibition alone and may involve inhibition of multiple other kinases 3. That is a direct, primary-source argument that FLT3 is not a sufficient dependency in this disease — the apparent single-agent activity of a multikinase inhibitor was doing something else.

3. Antigen-directed approaches fit the biology better, and address a specific failure mode. FLT3 is highly expressed in KMT2A-r B-ALL, and in relapses after CD19-directed therapy — especially in KMT2A-r B-ALL — most B-cell antigens are lost through lineage conversion to a myeloid phenotype, which defeats multi-B-antigen CAR T strategies. FLT3-directed CAR T cells showed antitumour efficacy against CD19-knockout KMT2A-r B-ALL both in vitro and in vivo, and dual CD19/FLT3 CAR T cells killed both wild-type and CD19-negative cells 8. That is a coherent use of FLT3 in this disease: as a surface antigen retained across lineage switch, not as a kinase selected by mutation. The lineage-switch failure mode it targets is documented clinically, with KMT2A-r infant ALL showing a higher proportion of myeloid-lineage-switch relapses after CD19 CAR T and a 3.6-fold increased risk of all-cause death 9.

Note for graph consistency: the canonical id gene-flt3 belongs to another beat and describes the gene and protein. This node covers only FLT3's status as a co-lesion in KMT2A-rearranged leukemia.

Facts and figures
gene symbolFLT3
proteinFLT3 / CD135 (fms-like tyrosine kinase 3)
hgnc IDHGNC:3765
entrez gene ID2322
uniprotP36888
cytoband13q12.2
canonical graph ID noteThe canonical node id gene-flt3 belongs to another beat. This node covers only FLT3's status as a co-lesion in KMT2A-rearranged leukemia.
genome build noteNo coordinates asserted. Protein changes in HGVS p. notation against the canonical FLT3 transcript as curated by cBioPortal; source analysis aligned to GRCh37-lite (NCBI build 37).
two distinct lesion classes
{
  "overexpression": {
    "prevalence": "near-universal in KMT2A-r; a defining MLL-signature gene",
    "evidence": "FLT3 expression in 41 infant KMT2A-r cases significantly higher than 8 germline-KMT2A infants (P<0.001) and 23 non-infant ALL (P=0.001)",
    "mechanism": "high-level wild-type receptor expression drives ligand-independent FLT3 phosphorylation",
    "prognostic": "low FLT3 expression identified 11 of 97 COG P9407 infants with 5-year EFS 100%",
    "pmids": [
      "15956279",
      "22210879",
      "11731795"
    ]
  },
  "activating_mutation": {
    "infant_kmt2a_r_activation_loop": "1/36 (~3%)",
    "infant_kmt2a_r_itd": "0/36 — none observed",
    "pcgp_infant_kmt2a_r": "non-recurrent (1 of 22 whole-genome cases)",
    "pcgp_non_infant_kmt2a_r": "recurrent, n = 3 of 20",
    "commonly_overquoted_figure": "5/30 (17%) in a non-age-restricted MLL cohort (Armstrong 2003)",
    "likely_reconciliation": "cohort age composition — FLT3 mutation is rare in infants and commoner in older children with the same fusion; neither paper reports the age breakdown needed to test this",
    "pmids": [
      "15956279",
      "12620411",
      "25730765"
    ]
  }
}
observed allele spectrum cbioportal all stjude 2015
{
  "n_mutations": 5,
  "n_samples_with_mutation": 5,
  "n_samples_sequenced": 93,
  "denominator_warning": "mixed cohorts; not a KMT2A-r frequency",
  "alleles": [
    "D839G",
    "N676K",
    "P934L",
    "R595_E596insNI"
  ],
  "absent_aml_canonical_alleles": [
    "D835",
    "I836del",
    "juxtamembrane ITD"
  ],
  "retrieved": "2026-09-18"
}
functionally validated allele
{
  "allele": "N676K",
  "n_cases_in_pcgp": 2,
  "assay": "factor-independent growth of IL-3-dependent BaF3 cells",
  "prior_context": "reported in an AML patient after PKC412 (midostaurin) resistance and as recurrent in core-binding-factor leukemia",
  "therapeutic_irony": "the one recurrent, functionally validated FLT3 allele in this disease was first described as a resistance allele to the FLT3 inhibitor KMT2A-r blasts are reported to be sensitive to"
}
contrast subtype
{
  "hyperdiploid_all_flt3_mutated": "6/25 (25%)",
  "etv6_runx1_all_flt3_mutated": "1/29",
  "p_value": 0.04,
  "pmid": "14670924",
  "point": "FLT3 mutation in childhood ALL is a hyperdiploid phenomenon, not a KMT2A-r one",
  "title": "FLT3 mutations in childhood acute lymphoblastic leukemia.",
  "authors": "Armstrong SA, Mabon ME, Silverman LB, et al.",
  "venue": "Blood",
  "year": 2004,
  "title_provenance": "Title/authors/venue backfilled from PubMed esummary on 2026-09-20 because this citation stored a bare PMID with no title, which made it unscoreable by verify-citations.py. BACKFILLING IS NOT VERIFICATION: it confirms the PMID resolves to a real paper, not that this is the right paper for the claim. That still requires reading."
}
therapeutic positioning
{
  "kinase_inhibition_caveat": "at PKC412 concentrations fully inhibiting FLT3 phosphorylation, cytotoxicity was only fractional; the authors attribute apoptosis to multikinase inhibition rather than FLT3 blockade alone",
  "correct_selection_biomarker": "expression, not genotype — mutation-based selection would enrol ~3% of infant patients",
  "antigen_directed": "FLT3 CAR T (single or dual CD19/FLT3) active against CD19-knockout KMT2A-r B-ALL preclinically; rationale is antigen retention across myeloid lineage switch",
  "clinical_status": "no FLT3 CAR T clinical data in KMT2A-r leukemia; no FLT3-inhibitor trial selecting on expression in infant KMT2A-r ALL located"
}
How this could bite us back
  • Resistance: the one recurrent FLT3 allele in this disease, N676K, was first characterised as a midostaurin-resistance mutation, so FLT3-mutant KMT2A-r cases may be precisely the ones a FLT3 inhibitor cannot help.
  • Insufficiency: the primary source shows that fully inhibiting FLT3 phosphorylation produces only fractional cytotoxicity in infant KMT2A-r cells, so FLT3 is not a sufficient dependency and single-agent FLT3 inhibition should be expected to fail.
  • Essential-function toxicity: FLT3 is expressed on normal haematopoietic stem and progenitor cells, so a FLT3-directed CAR T or antibody risks durable myeloablation — a far heavier price in an infant than CD19-directed B-cell aplasia.
  • Compensation: FLT3 signals into RAS-MAPK and PI3K, the same pathway activated by KRAS/NRAS mutation in a third to a half of these cases. A downstream RAS mutation bypasses FLT3 blockade entirely, and those mutations are already present subclonally at diagnosis.
  • Biomarker error: importing the AML practice of selecting patients on FLT3-ITD would exclude essentially the entire infant KMT2A-r population, since no ITD was found in a dedicated 36-case screen.
  • Antigen escape: targeting FLT3 to survive CD19 loss assumes FLT3 cannot itself be lost. In a disease whose relapse mechanism appears to be lineage and transcriptional plasticity, downregulating one more surface antigen is exactly the escape route the biology already uses.
Still open
  1. Is the 17% versus 3% FLT3 mutation discrepancy entirely explained by cohort age composition? Reanalysing the Armstrong 2003 cohort by age at diagnosis would settle a number that is still miscited in reviews, and the data exist.
  2. Are both PCGP N676K cases infant or non-infant? N676K is the only recurrent, functionally validated FLT3 allele in KMT2A-r leukemia, and whether it belongs to the infant or the older-child population changes its clinical relevance entirely.
  3. Does N676K, a described PKC412-resistance allele, confer resistance to current-generation FLT3 inhibitors in a KMT2A-r background? This is directly testable and would determine whether FLT3 mutation in this disease is a target or a contraindication.
  4. Would a FLT3 inhibitor selected on FLT3 expression rather than mutation show activity in infant KMT2A-r ALL? Given ligand-independent activation from overexpression, this is the trial design the biology implies, and it has never been run.
  5. Is FLT3 expression retained on the myeloid population after lineage switch in actual patients, not CRISPR-engineered lines? The entire rationale for FLT3 CAR T in this disease depends on this, and it has not been shown in lineage-switched clinical material.
  6. Is FLT3 expression prognostic independently of KMT2A fusion partner? The COG classifier adjusted for KMT2A status as a binary, not by partner, and HOXA-high versus HOXA-low biology could confound it.
Every source behind this entry (9)
  1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    KMT2A(MLL)-rearranged acute lymphoblastic leukemias have a characteristic, highly distinct gene-expression profile consistent with an early haematopoietic progenitor expressing select multilineage markers and individual HOXA genes, separable by clustering from conventional ALL and AML, and constitute a distinct disease entity whose selectively expressed genes should be examined for molecular targets.

    MLL translocations specify a distinct gene expression profile that distinguishes a unique leukemia · Armstrong SA, Staunton JE, Silverman LB, Pieters R, den Boer ML, Minden MD, Sallan SE, Lander ES, Golub TR, Korsmeyer SJ · Nature Genetics 30(1):41-7 2002
    PMID 11731795DOI
    What this evidence cannot show

    Early oligonucleotide-microarray study; absolute expression values are not comparable to modern RNA-seq. The abstract does not name FLT3 specifically — FLT3's identification as the leading selectively expressed target comes from the companion and follow-up work cited as E2 and E3. Cited here for the signature framework, not for a FLT3 number.

    Evidence that argues against this (1)
    • A distinct expression profile is a statement about transcriptional identity, not about what the cells need. This project's own per-line data breaks the inference: RS4;11, SEM and SEMK2 are all KMT2A::AFF1 B-ALL and would cluster together on any such profile, yet SEMK2 and SEM are among the most FLT3-dependent lines in an 8,248-gene sweep while RS4;11 is not FLT3-dependent at all. Whatever separates them is invisible to the profile that defines the entity. Oncofusion knowledge graph (internal analysis) 2026
  2. E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    In 30 MLL-rearranged leukemias, 5 (17%) contained activation-loop FLT3 mutations resulting in constitutive activation — three a newly described deletion of I836 and two D835 mutations; PKC412 was cytotoxic to Ba/F3 cells dependent on either activated FLT3 mutant and was differentially cytotoxic to MLL-translocated cells with FLT3 activated by either wild-type overexpression or mutation, and was active against an MLL mouse model in vivo.

    Inhibition of FLT3 in MLL. Validation of a therapeutic target identified by gene expression based classification · Armstrong SA, Kung AL, Mabon ME, Silverman LB, Stam RW, Den Boer ML, Pieters R, Kersey JH, Sallan SE, Fletcher JA, Golub TR, Griffin JD, Korsmeyer SJ · Cancer Cell 3(2):173-83 2003
    PMID 12620411DOI
    What this evidence cannot show

    This 17% figure is the most frequently over-generalised number in the FLT3/KMT2A literature. The cohort is described as 30 'MLLs' with no age breakdown in the abstract, so it is NOT established to be infant-restricted; the infant-specific figure from the same research network two years later was ~3% (E3). The I836del and D835 alleles reported here do not appear in the later whole-genome cohort's FLT3 spectrum (E6), which is a further reason to treat the two cohorts as non-interchangeable. **Regraded moderate -> established on 2026-09-20** under `concept-the-confidence-grading-rule`: 5 of 30 is a count within the paper's own reported cohort.

    Evidence that argues against this (1)
    • In 36 infant KMT2A-r ALL cases, activation-loop mutations occurred in only ~3% (1 of 36) and no FLT3/ITD was observed — a roughly five-fold lower rate in the infant-restricted population. Stam RW et al. 2005
  3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    FLT3 expression in 41 infants with KMT2A-r ALL was significantly higher than in 8 germline-KMT2A infants (P < 0.001) and 23 non-infant germline-KMT2A ALL patients (P = 0.001); activation-loop mutations occurred in only ~3% (1 of 36) with no FLT3/ITD observed; high-level wild-type FLT3 expression associates with ligand-independent FLT3 activation; and at PKC412 concentrations minimally required to fully inhibit FLT3 phosphorylation the cytotoxic effects were only fractional, so apoptosis is unlikely to result from FLT3 inhibition alone.

    Targeting FLT3 in primary MLL-gene-rearranged infant acute lymphoblastic leukemia · Stam RW, den Boer ML, Schneider P, Nollau P, Horstmann M, Beverloo HB, van der Voort E, Valsecchi MG, de Lorenzo P, Sallan SE, Armstrong SA, Pieters R · Blood 106(7):2484-90 2005
    PMID 15956279DOI
    What this evidence cannot show

    Mutation screening covered the activation loop and ITD region by Sanger sequencing in 36 cases — it would miss non-activation-loop alleles such as N676K, which the later whole-genome work found (E4). Expression comparison groups are small (n = 8 and n = 23). PKC412/midostaurin is a multikinase inhibitor, which the authors themselves invoke to explain their cytotoxicity data, so the ex vivo sensitivity result is not FLT3-specific.

    Evidence that argues against this (1)
    • High wild-type FLT3 expression was taken into a randomised trial and the trial did not support it: adding the FLT3 inhibitor lestaurtinib to chemotherapy in newly diagnosed KMT2A-rearranged infant ALL (COG AALL0631) did not improve outcome. Expression as a drug rationale has had its clinical test in exactly this population. Leukemia 2021
  4. E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In the St. Jude PCGP study FLT3 was a non-recurrent kinase/PI3K/RAS lesion in the 22-case infant KMT2A-r whole-genome cohort but was recurrent at n = 3 of 20 in the non-infant KMT2A-r cohort; two cases carried the FLT3 N676K allele — previously reported in an AML patient after PKC412 resistance and as recurrent in core-binding-factor leukemia — and N676K was shown to be activating by conferring factor-independent growth on IL-3-dependent BaF3 cells.

    The landscape of somatic mutations in infant MLL-rearranged acute lymphoblastic leukemias · Andersson AK, Ma J, Wang J, Chen X, Gedman AL, Dang J, et al. · Nature Genetics 47(4):330-7 2015
    PMID 25730765DOI
    What this evidence cannot show

    n = 3 of 20 in a non-infant cohort that mixes 9 ALL, 10 AML and 1 acute undifferentiated leukemia, so the recurrence may be driven by the myeloid cases where FLT3 mutation is expected. The two N676K cases are not stated to be both infant or both non-infant in the retrieved text. BaF3 factor-independence is a standard but permissive activation assay and does not establish that the allele is a driver in a KMT2A-r blast.

    Evidence that argues against this (1)
    • The cohort is too small to support the word it is being given. "Non-recurrent at n = 22" and "recurrent at 3 of 20" are the same statement inside their confidence intervals: 0 of 22 is compatible with a true frequency up to roughly 13%, and 3 of 20 with anything from 3% to 38%. The infant/non-infant contrast the entry draws from them is not distinguishable from sampling. Leukemia 2021
  5. E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In 97 infant ALL cases from Children's Oncology Group trial P9407, three genes were predictive of event-free survival beyond age and KMT2A status — FLT3, IRX2 and TACC2 — and low FLT3 expression identified a group of 11 infants with a 5-year event-free survival of 100%, while differential IRX2/TACC2 expression split the remainder into groups with 5-year EFS of 16% versus 64% (P < 0.001).

    Gene expression profiles predictive of outcome and age in infant acute lymphoblastic leukemia: a Children's Oncology Group study · Kang H, Wilson CS, Harvey RC, Chen IM, Murphy MH, Atlas SR, et al. · Blood 119(8):1872-81 2012
    PMID 22210879DOI
    What this evidence cannot show

    The 100% EFS group is n = 11, so the point estimate is fragile even though the classifier validated in an independent Interfant-99 cohort. This is an expression classifier, not a FLT3-specific causal claim, and the cohort is all infant ALL rather than KMT2A-r only. It establishes that FLT3 expression carries prognostic information; no comparable evidence exists that FLT3 mutation status does.

    Evidence that argues against this (1)
    • A 100% five-year event-free survival in 11 infants is compatible with a true rate near 75%, and the expression cut-point that defined the group was derived in the same 97 patients it was evaluated in. Neither the threshold nor the survival estimate has an independent validation cohort, which is what would be needed before low FLT3 expression could de-escalate anyone's treatment. Leukemia 2021
  6. E6moderate confidence 3 of 5: One strong study, or several smaller consistent ones.database

    The public curation of the PCGP study contains 5 FLT3 mutations across 5 of 93 sequenced samples, with protein changes D839G, N676K, P934L and R595_E596insNI — and notably no D835, no I836 deletion and no juxtamembrane internal tandem duplication.

    What this evidence cannot show

    5/93 is NOT a KMT2A-r frequency — the 93 samples mix infant KMT2A-r, infant non-KMT2A-r, non-infant KMT2A-r and relapse specimens, and the FLT3-recurrent cases are expected to be the non-infant ones. Absence of ITD in a curated short-read mutation table is weak evidence, since ITDs are systematically under-called by standard SNV/indel pipelines; the stronger evidence for ITD absence is the dedicated ITD screen in E3. Variant lists are deduplicated protein changes, not per-sample genotypes.

    Evidence that argues against this (2)
    • "No ITD observed" is partly a statement about the assay. FLT3 internal tandem duplications are insertions of tens to hundreds of bases and are systematically missed by standard short-read alignment and variant calling, which is why dedicated ITD callers exist and why published methods report gains in sensitivity over the general-purpose pipelines. A curation built on ordinary SNV calling cannot be read as excluding ITD. Journal of Molecular Diagnostics 2013
    • The same point from the other direction: purpose-built ITD detection from NGS data is still being published as a methodological advance more than a decade later, which is not what the field looks like when a class of variant is reliably caught by default pipelines. BMC Bioinformatics 2023
  7. E7strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Approximately 25% (6 of 25) of hyperdiploid childhood ALL samples carry FLT3 mutations versus 1 of 29 ETV6::RUNX1-rearranged samples (P = 0.04, Fisher exact), including three novel in-frame deletions within a seven-amino-acid region of the receptor juxtamembrane domain — establishing that FLT3 mutation in childhood ALL is concentrated in hyperdiploidy rather than KMT2A rearrangement.

    FLT3 mutations in childhood acute lymphoblastic leukemia · Armstrong SA, Mabon ME, Silverman LB, Li A, Gribben JG, Fox EA, Sallan SE, Korsmeyer SJ · Blood 103(9):3544-6 2004
    PMID 14670924DOI
    What this evidence cannot show

    71 ALL patients total, so each subgroup is small. This study did not re-screen a KMT2A-r cohort; its statement that FLT3 is 'often mutated' in MLL-rearranged ALL refers back to the 5/30 figure in E2 and inherits that figure's cohort ambiguity. Cited here for the hyperdiploid contrast only.

    Evidence that argues against this (1)
    • This is a different disease subtype. Hyperdiploid childhood ALL is not KMT2A-rearranged, so a FLT3-mutation frequency measured there constrains nothing about KMT2A-r; the entry imports it as context and the surrounding claims read it as support. The internal statistics are also fragile: 6 of 25 against 1 of 29 at P = 0.04 turns on one or two patients. Leukemia 2021
  8. E8moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    FLT3 is highly expressed in KMT2A-r B-ALL; FLT3-directed CAR T cells showed antitumour efficacy against CRISPR-generated CD19-knockout KMT2A-r B-ALL cells in vitro and in vivo, and dual CD19/FLT3-targeted CAR T cells were cytotoxic against both wild-type and CD19-negative cells, addressing the lineage-conversion antigen-loss failure mode characteristic of KMT2A-r B-ALL relapse.

    Targeting FLT3-specific chimeric antigen receptor T cells for acute lymphoblastic leukemia with KMT2A rearrangement · Suematsu M, Yagyu S, Nagao N, Kubota S, Shimizu Y, Tanaka M, Nakazawa Y, Imamura T · Cancer Immunology, Immunotherapy 72(4):957-968 2023
    PMID 36214866DOI
    What this evidence cannot show

    Preclinical only, using CRISPR-engineered CD19-knockout cell-line models rather than patient lineage-switched material, which is a simplification of the clinical failure mode. No clinical FLT3 CAR T data in KMT2A-r leukemia exists. FLT3 is expressed on normal haematopoietic stem and progenitor cells, so on-target myelotoxicity is a serious and unaddressed concern in this design.

    Evidence that argues against this (1)
    • FLT3 is expressed on normal haematopoietic stem and progenitor cells, so a FLT3-directed CAR is on-target against the compartment that has to reconstitute the patient. For CD19 the analogous toxicity is B-cell aplasia, which is survivable with immunoglobulin replacement; the FLT3 equivalent is not. This is the predictable limiting problem and the in vitro and xenograft work quoted here does not address it. Leukemia 2021
  9. E9moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    Among 231 children and young adults receiving CD19-directed CAR T-cell therapy, KMT2A-r infant ALL (n = 13) showed no increased hazard of relapse overall but a higher proportion of relapses associated with myeloid lineage switch and a 3.6-fold increased risk of all-cause death (95% CI 1.04-12.75, P = 0.0434).

    Impact of high-risk cytogenetics on outcomes for children and young adults receiving CD19-directed CAR T-cell therapy · Leahy AB, Devine KJ, Li Y, Liu H, Myers R, DiNofia A, et al. · Blood 139(14):2173-2185 2022
    PMID 34871373DOI
    What this evidence cannot show

    n = 13 for the KMT2A-r infant subgroup; the 95% CI on the death hazard barely excludes 1, and the lineage-switch observation is a proportion rather than a tested rate. Retrospective across five trials plus commercial product in a relapsed/refractory population. Cited to establish that the failure mode E8 targets is clinically real, not to quantify it.

    Evidence that argues against this (1)
    • Thirteen patients, a confidence interval running from 1.04 to 12.75 and P = 0.0434 is a result sitting on its threshold, and the same study reports NO increased hazard of relapse in the same 13. A mortality signal without a relapse signal in a group this size is more consistent with competing risks and with how sick infants are than with a biological effect of the rearrangement on CAR-T durability. Leukemia 2021
Connected to