Oncofusion

start / graph / panel-infant-all-pdx-kmt2a-aff1-hopkins

Explainerstrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.Q21-index-fusion-models

The 'infant ALL' PDX panel with four KMT2A::AFF1 models — one of which the full text calls adult

also called infant ALL PDX panel · entospletinib PDX panel · ALL142MR · ALL3113 · ALL135MR · ALL3103

A published set of mouse models grown from leukaemia patients contains four with the exact fusion this project studies. The abstract calls them infant models; reading the paper shows one of them came from an adult, and two others carry an extra RAS mutation that made a targeted drug fail.

The same thing, for a scientist

A SYK-inhibitor study gives the composition of its patient-derived xenograft panel by fusion: **KMT2A-AFF1 n=4**, KMT2A-MLLT3 n=1, KMT2A-MLLT1 n=4, non-KMT2A-rearranged n=3. The abstract frames the whole panel as *infant ALL* PDX models. **The full text does not.** It names `ALL3113` explicitly as *'an adult RAS wild-type KMT2A-AFF1 ALL PDX model'* and, in the selumetinib experiment, contrasts *'RAS-mutant (ALL142MR; infant) and RAS wild-type (ALL3113MR; adult) KMT2A-AFF1 ALL PDX models'*. So the index-fusion arm of an 'infant ALL PDX' panel contains at least one adult model, and the headline experiment of the paper compares an infant model against an adult one while varying RAS status. The scientific result stands and is important: entospletinib inhibited leukaemia proliferation in RAS-wild-type KMT2A-AFF1, KMT2A-MLLT3 and KMT2A-MLLT1 models with enhanced activity plus vincristine, and **failed in two KMT2A-AFF1 models carrying NRAS or KRAS mutations** (ALL142MR carries NRAS G12D); combining entospletinib with the MEK inhibitor selumetinib was superior in KMT2A-AFF1 models.

Go deeper
The correction this entry carries, and how it was found

This entry was first written from the abstract and said 'four infant PDX models carrying the index fusion'. The paper's PMC full text (PMC8018117, <body> present, 47,682 characters) says otherwise for at least one of the four 1.

That is the same failure mode this project's provenance beat measured across the graph — an experimental system that is not what the summary says — occurring inside a single paper between its abstract and its results. It is not misconduct and it is not hidden; the full text states the ages plainly. It is invisible to anyone working from abstracts, which is how this graph builds most of its entries.

Operational lesson: a panel described as 'infant' in an abstract is a claim about the panel's centre of gravity, not about every member. Where a specific model carries a specific conclusion, the model's own description has to be read.

What is genuinely established about the index-fusion models
  • ALL142MR — KMT2A-AFF1, NRAS G12D, infant, multiply relapsed. Insensitive to entospletinib in vitro (no inhibition of pERK1/2, pAKT S473, pSTAT5, cMYC) and in vivo; responds to vincristine; responds to selumetinib alone and in combination 1.
  • ALL3113 / ALL3113MR — KMT2A-AFF1, RAS wild-type, adult. Significant effect of entospletinib alone and with vincristine 1.
  • Two further KMT2A-AFF1 specimens make up the n=4 and are not individually named in the retrieved text.
  • ALL135MR — KMT2A-MLLT1, RAS wild-type, infant, multiply relapsed, entospletinib-sensitive.
  • ALL3103 — KMT2A-MLLT3.
  • A non-KMT2A-rearranged infant control carries TCF3-PBX1 from t(1;19) with a concomitant KRAS mutation and was also entospletinib-insensitive — which is the cleanest evidence in the paper that the insensitivity tracks RAS and not the fusion 1.
The epidemiological frame the paper states

KMT2A rearrangements occur in approximately 10% of childhood and adult B-ALL, with highest frequency — 75% — in infants diagnosed at under 365 days old 1. That is the reason a panel assembled for infant ALL is overwhelmingly KMT2A-rearranged, and it is also the reason an adult KMT2A-rearranged model can slip into such a panel without seeming out of place.

Facts and figures
total specimens12
index fusion models4
index fusion models confirmed infant1
index fusion models confirmed adult1
index fusion models age unstated2
hostNOD-scid background (full strain designation truncated in retrieved text)
readoutflow-cytometric human CD45+CD19+ ALL in murine peripheral blood and end-study spleen
epidemiology statedKMT2A rearrangements in ~10% of childhood and adult B-ALL; 75% of infants diagnosed at <365 days
full text retrievalPMC8018117 via Europe PMC fullTextXML; body present, 47,682 characters; Table 1 rows not extractable from the XML
obtainablenot established; no public repository accession given
panel composition by fusion
KMT2A-MLLT31.000KMT2A-MLLT3: 1non-KMT2A-rearranged3.000non-KMT2A-rearranged: 3KMT2A-AFF14.000KMT2A-AFF1: 4KMT2A-MLLT14.000KMT2A-MLLT1: 4
panel composition by fusion — the numbers behind the chart
{
  "KMT2A-AFF1": 4,
  "KMT2A-MLLT3": 1,
  "KMT2A-MLLT1": 4,
  "non-KMT2A-rearranged": 3
}
named models
[
  {
    "model": "ALL142MR",
    "fusion": "KMT2A-AFF1",
    "ras": "NRAS G12D",
    "age_class": "infant",
    "entospletinib": "insensitive",
    "selumetinib": "sensitive"
  },
  {
    "model": "ALL3113 / ALL3113MR",
    "fusion": "KMT2A-AFF1",
    "ras": "wild-type",
    "age_class": "ADULT",
    "entospletinib": "sensitive alone and with vincristine"
  },
  {
    "model": "ALL135MR",
    "fusion": "KMT2A-MLLT1",
    "ras": "wild-type",
    "age_class": "infant, multiply relapsed",
    "entospletinib": "sensitive"
  },
  {
    "model": "ALL3103",
    "fusion": "KMT2A-MLLT3",
    "age_class": "not stated in retrieved text"
  },
  {
    "model": "non-KMT2A-R control",
    "fusion": "TCF3-PBX1 t(1;19)",
    "ras": "KRAS-mutant",
    "age_class": "infant",
    "entospletinib": "insensitive"
  }
]
How this could bite us back
  • A targeted agent failed in the index-fusion models carrying RAS mutations, and the RAS-mutant models here are relapse-derived — exactly the setting a new drug would first be used in.
  • The insensitivity is not fusion-specific: a TCF3-PBX1 infant model with KRAS behaved the same way. A resistance mechanism attributed to KMT2A::AFF1 biology here would be misattributed.
  • Peripheral-blood readouts under-report disease persisting in spleen and marrow, so xenograft efficacy in this field may be systematically overstated by the standard endpoint.
  • PDX passage selects for engraftability, and the models that fail to engraft leave no record. If infant blasts engraft less readily than adult ones, an 'infant' panel will drift adult without anyone deciding that it should.
Still open
  1. What are the donor ages of the two unnamed KMT2A-AFF1 models? Table 1 holds them and was not extractable from the PMC XML; the PDF or the supplementary appendix would settle it.
  2. How many other 'infant' preclinical panels in this literature contain adult models? This is the first case found by reading rather than by assuming, and it took one paper.
  3. What fraction of primary infant KMT2A::AFF1 leukaemias carry an activating RAS mutation at diagnosis rather than at relapse? Both RAS-mutant models here are relapse-derived (MR suffix), which is a different population.
Every source behind this entry (3)
  1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    In this PDX panel, basal kinase signalling and HOXA9/MEIS1 expression differed among KMT2A-AFF1 (n=4), KMT2A-MLLT3 (n=1), KMT2A-MLLT1 (n=4) and non-KMT2A-rearranged (n=3) specimens; entospletinib inhibited leukaemia proliferation in RAS-wild-type models with enhanced activity combined with vincristine but did not decrease burden in two KMT2A-AFF1 models carrying NRAS or KRAS mutations, and entospletinib plus selumetinib was superior in KMT2A-AFF1 models. The full text names ALL3113 as an ADULT RAS wild-type KMT2A-AFF1 model and ALL142MR as an infant NRAS G12D KMT2A-AFF1 model, and states that KMT2A rearrangements occur in about 10% of childhood and adult B-ALL with highest frequency, 75%, in infants diagnosed at under 365 days old.

    PMID 32414848DOI
    What this evidence cannot show

    Full text read via Europe PMC (PMC8018117), body element confirmed present, 47,682 characters. Table 1 (molecular and cytogenetic characteristics of the PDX models) has no extractable rows in the XML, and the per-model age and RAS table is therefore not available here; the adult attribution for ALL3113 comes from two statements in the results and figure legends, not from a table. Only two of the four KMT2A-AFF1 models are named in the body text, so whether the other two are infant-derived is not established. Donor ages in months or days are not given anywhere in the retrieved text.

    Evidence that argues against this (2)
    • The RAS-insensitivity result is not fusion-specific and the paper says so: a non-KMT2A-rearranged INFANT control model with TCF3-PBX1 and a concomitant KRAS mutation was also entospletinib-insensitive. So the panel's most transferable finding is about RAS, not about KMT2A::AFF1, and reading it as an index-fusion result would overstate it. Loftus JP et al. 2021
    • An independent mouse model driven by Mll-Af4 plus activated Ras found MEK and ATR inhibition effective, which supports the RAS-pathway reading — but it does so in a murine model with an engineered RAS lesion, so the two lines of evidence are not independent of the assumption that RAS status is the operative variable. Chu SH et al. 2018
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Independent infant KMT2A-rearranged ALL xenograft programmes exist and report in vivo survival endpoints, including azacitidine and decitabine monotherapy significantly prolonging survival in infant KMT2A-rearranged ALL xenografts and azacitidine plus venetoclax producing a significant in vivo survival advantage.

    PMID 36380143DOI
    What this evidence cannot show

    Abstract read; full text is available in PMC (PMC9883157, body present, 46,126 characters) and was NOT read for this item, which is a known gap given that reading the Haematologica full text changed this entry's headline. The abstract says 'KMT2A-rearranged infant ALL xenografts' without stating how many carry AFF1 or the donors' ages.

    Evidence that argues against this (1)
    • The same paper reports antagonism between hypomethylating agents and methotrexate in a subset of samples, with ABC-transporter upregulation as a candidate mechanism. A xenograft programme finding both synergy and antagonism within one panel is showing that infant KMT2A-rearranged disease is heterogeneous enough that a small panel can give either answer. Cheung LC et al. 2023
  3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Infant KMT2A-rearranged ALL patient-derived xenografts have been used to monitor menin-inhibitor response: immune-deficient mice engrafted with three MLL-rearranged ALL PDXs were treated with the menin inhibitor SNDX-50469 and circulating tumour DNA tracked leukaemic burden, outperforming peripheral-blood flow cytometry at low disease burden and revealing persistence of leukaemia in spleen and bone marrow after treatment.

    PMID 39682177DOI
    What this evidence cannot show

    Abstract read; full text not read. Three PDXs, partner not stated in the abstract, so it is not established that any of the three carries KMT2A::AFF1 — and, given what full-text reading did to E1, that qualification should be taken seriously rather than assumed away. SNDX-50469 is a tool compound related to revumenib, not revumenib itself.

    Evidence that argues against this (1)
    • The finding that leukaemia persists in spleen and bone marrow while peripheral blood clears is a caution about every xenograft efficacy readout in this beat that used peripheral-blood human CD45 as its endpoint — which includes the entospletinib work, where treatment response was quantified as ALL cell counts in peripheral blood during treatment and in end-study spleens. A drug can look active on the standard readout while the disease is intact in the niche. Doculara L et al. 2024
Connected to