Oncofusion

start / graph / cell-type-csf1r-positive-fetal-lmpp

Cell typemoderate confidence 3 of 5: One strong study, or several smaller consistent ones.Q139-single-cell

The CSF1R-positive fetal lymphoid-primed multipotent progenitor — a fetal-restricted subset of LMPPs that makes AML rather than ALL when KMT2A::MLLT3 is switched on, and exists in human embryos

also called CSF1R+ LMPP · fetal myeloid-biased LMPP · foetal-restricted CSF1R+ multipotent progenitor

Among the multipotent blood progenitors in a fetal liver there is a subset carrying a receptor called CSF1R, and these are the ones that turn into myeloid leukaemia rather than lymphoid leukaemia when the KMT2A fusion is switched on in a mouse. The same population exists in early human embryos, and blocking CSF1R kills a KMT2A-rearranged paediatric leukaemia cell line, which makes it a possible target.

The same thing, for a scientist

The cell-of-origin question in KMT2A-rearranged infant leukaemia has a lineage half that has never had a clean answer: the **same fusion, KMT2A::MLLT3, produces myeloid or lymphoid disease in different infants**. Camiolo et al. resolve part of it by subsetting the progenitor rather than the fusion: **a subset of fetal lymphoid-primed multipotent progenitors (LMPPs) expressing CSF1R gives rise to AML upon KMT2A::MLLT3 induction in a mouse model, with the myeloid phenotype at least in part dependent on CSF1R signalling** [E1]. The CSF1R+ LMPPs carry a **stem-cell-like and myeloid-biased expression signature**, **require autophagy** to expand and form blast-like colonies in methylcellulose, and — the translational hinge — **interrogation of public datasets confirms a human fetal-restricted CSF1R+ LMPP population at early stages of embryonic development**, with CSF1R inhibition causing significant cell death in a KMT2A::MLLT3+ paediatric leukaemia line [E1]. **This is a cell type, not a cell state, and the qualifier that makes it matter to this project is 'fetal-restricted.'** It joins the LIN28B and CD10-negative pre-pro-B threads: a series of compartments that exist during fetal life, are permissive for transformation, and are gone by the time an adult progenitor could be tested. For the index fusion the entry is a signpost rather than a result — **KMT2A::AFF1 was not tested**, and the lineage asymmetry it would have to explain runs the other way, since t(4;11) is overwhelmingly lymphoid.

Go deeper
What is actually established, separated from what is extrapolated
claimsystemstrength
CSF1R+ fetal LMPPs give rise to AML on KMT2A::MLLT3 inductionmouse, inducible modeldirect experiment
the myeloid phenotype depends in part on CSF1R signallingmousedirect experiment
CSF1R+ KMT2A::MLLT3 LMPPs have a stem-like, myeloid-biased signature and need autophagy to form blast coloniesmouse, methylcellulosedirect experiment
a fetal-restricted CSF1R+ LMPP population exists in human early embryoshuman, public dataset interrogationobservational, reanalysis
CSF1R inhibition kills a KMT2A::MLLT3+ paediatric leukaemia linehuman linesingle line, in vitro
CSF1R-CAR-T might benefit these patientsnoneauthors' proposal

The chain is mouse mechanism + human existence + one human cell line. That is a normal and respectable structure for a 2026 cell-of-origin paper, and it is not the same as a demonstration in human disease. The graph should hold it at moderate.

Why a fetal-restricted compartment keeps being the answer

This corpus now holds several independent instances of the same shape:

  • LIN28B is essential for human fetal B-lymphopoiesis and for initiation of KMT2A::AFF1 infant leukaemia (cell-state-lin28b-dependent-fetal-b-and-nk-lymphopoiesis, gene-lin28b).
  • The CD10-negative fetal pre-pro-B progenitor is an ontogeny-restricted B-lineage compartment (cell-type-fetal-preprob-cd10-negative).
  • The neonatal window of transformability and the CD24+ pre-leukaemic state are ontogeny-specific responses to the same fusion (cell-state-neonatal-window-of-transformability, cell-state-cd24-positive-preleukaemic-kmt2a-aff1).
  • And now a fetal-restricted CSF1R+ LMPP that steers the same fusion to a myeloid outcome.

The pattern is that ontogeny, not the fusion, is doing the lineage work — which is the strongest available form of this project's own hypothesis that the cell state is permissive and the fusion is opportunistic. It also predicts something uncomfortable: an adult-derived model cannot reproduce the disease's defining biology, and most of this corpus's mechanistic evidence comes from adult-derived lines.

The counter-pattern worth stating: 'fetal-restricted' is a claim about a marker combination's presence, and markers are not compartments. A CSF1R+ LMPP in a human embryonic dataset is a cluster in somebody's reanalysis, and whether it is the same cell as the mouse population that was functionally tested is an inference across species, across assay and across developmental time.

The therapeutic hook, and why it is riskier than it looks

CSF1R inhibitors are clinically available (the class includes approved agents in tenosynovial giant cell tumour and a large development history in macrophage-directed oncology), and the paper proposes anti-CSF1R CAR-T 1. Two things follow, in opposite directions.

In favour: this is a target with existing pharmacology and a stated lineage-steering role, and if the myeloid phenotype is CSF1R-dependent then CSF1R blockade is a differentiation-steering intervention rather than a cytotoxic one — the same conceptual class as a menin inhibitor.

Against, and it is serious: CSF1R is the receptor for CSF1 and IL-34 and is the defining dependency of the entire tissue-macrophage compartment, including microglia. A CAR-T against it in an infant is not a targeted therapy in any comfortable sense. The paper's own framing — CSF1R marks a fetal progenitor — is a reason for concern rather than reassurance in a patient population whose macrophage compartment is still being established.

Facts and figures
locationfetal liver / early embryonic haematopoiesis; described as fetal-restricted, i.e. not present in adult bone marrow
shown for fusionKMT2A::MLLT3 (mouse induction, and one KMT2A::MLLT3+ paediatric line for the drug effect). NOT tested for KMT2A::AFF1.
functional dependencyautophagy required for expansion and blast-like colony formation
druggable axisCSF1R inhibition; anti-CSF1R CAR-T proposed by the authors, untested
source pmid41545699
cited in this corpus0
markers
[
  "CSF1R+",
  "LMPP surface phenotype",
  "stem-cell-like and myeloid-biased transcriptional signature"
]
atlas refs
[
  "public human embryonic haematopoiesis single-cell datasets, interrogated by the authors (specific accessions not stated in the abstract)"
]
How this could bite us back
  • CSF1R is the master dependency of tissue macrophages including microglia. An anti-CSF1R CAR-T — a persistent, non-titratable depletion — in an infant is a very different proposition from a reversible small-molecule inhibitor, and the neurodevelopmental exposure in a population this graph already tracks for neurocognitive late effects would be essentially unbounded.
  • If CSF1R marks a fetal-restricted compartment, then a CSF1R-directed therapy works best precisely where the normal counterpart is most abundant and most needed. The therapeutic index of a marker defined by its developmental restriction is worst in the patients the restriction identifies.
  • Targeting the lineage-steering receptor rather than the fusion risks converting a myeloid leukaemia into a lymphoid one rather than curing it — which in a KMT2A-rearranged disease that already switches lineage under pressure is not a hypothetical failure mode.
Still open
  1. What does KMT2A::AFF1 do in a CSF1R+ fetal LMPP? If it also makes myeloid disease, the compartment rather than the partner sets lineage; if it makes B-ALL, the partner matters and the index fusion's lymphoid bias has a cellular address.
  2. Which public human embryonic datasets were interrogated? The accession would let this project re-derive the CSF1R+ LMPP cluster and check its relationship to the LIN28B-high and CD10-negative pre-pro-B compartments it already holds.
  3. Is CSF1R expressed on the HSPC-like blast fraction in the 25-patient KMT2A-r single-cell multiome? That is a one-gene lookup in deposited data and would connect this fetal compartment to the human disease directly.
  4. The autophagy dependency is a second, cleaner drug axis than CSF1R and has no obvious macrophage-compartment toxicity. Has any autophagy inhibitor been tested in KMT2A-rearranged infant leukaemia?
  5. Does the CSF1R+ LMPP persist into the neonatal window that cell-state-neonatal-window-of-transformability describes, or does it close earlier? The two windows may or may not be the same window.
Every source behind this entry (2)
  1. E1moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    A subset of fetal lymphoid-primed multipotent progenitors expressing CSF1R gives rise to AML upon KMT2A::MLLT3 induction in a mouse model, with the myeloid phenotype partly dependent on CSF1R signalling; these cells have a stem-cell-like myeloid-biased signature and require autophagy to expand and form blast-like colonies; a human fetal-restricted CSF1R+ LMPP population exists at early embryonic stages in public datasets; and CSF1R inhibition causes significant cell death in a KMT2A::MLLT3+ paediatric leukaemia cell line.

    CSF1R marks a subset of foetal haematopoietic multipotent progenitor cells with acute myeloid leukaemia propagation properties · Camiolo G, González Silvera D, Leah T, et al. (senior author Ottersbach K) · Leukemia 2026
    PMID 41545699DOI
    What this evidence cannot show

    The functional leukaemia-propagation result is in a MOUSE inducible KMT2A::MLLT3 model. The human arm is (a) reanalysis of public embryonic datasets to show the population exists and (b) one KMT2A::MLLT3+ paediatric cell line for the drug effect — no human primary leukaemia and no xenograft. Partner is KMT2A::MLLT3, NOT the index fusion KMT2A::AFF1, and the lineage question runs the other way for t(4;11), which is overwhelmingly lymphoid; nothing here says what a CSF1R+ LMPP does with KMT2A::AFF1. The anti-CSF1R-CAR-T suggestion is the authors' proposal with no supporting experiment. Read at abstract level; figures not opened. `gene-csf1r` does not exist in this graph, so this entry does not link to it.

    Evidence that argues against this (1)
    • Cell-of-origin claims in KMT2A leukaemia have repeatedly failed to transfer between mouse and human and between partners. The same corpus records that only cord-blood-derived HSPCs, not adult ones, are susceptible to MLL-AF4 transformation, and that MLL-AF9 immortalises different lineages at different points in ontogeny — a body of work whose lesson is that the permissive compartment is narrow and system-specific. A mouse LMPP subset is therefore a hypothesis about the human compartment, not a description of it. Secker KA et al. 2020
  2. E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.dataset

    This paper is not cited anywhere in this corpus, and no entry is named for CSF1R, although CSF1R is mentioned in prose in six entries.

    What this evidence cannot show

    A PMID scan plus an id-level check that `gene-csf1r` is absent from the 2,603 node ids, plus a case-insensitive grep for the token CSF1R over all node files EXCLUDING this beat's own folder. The grep returns six entries - exp-bivalency-cuttag-kmt2a-aff1, fusion-cbfb-myh11, fusion-runx1-runx1t1, gene-sfpq, collab-syndax-pharmaceuticals and gene-foxp1 - so the correct statement is 'no entry OF ITS OWN', not 'unmentioned'. The first version of this evidence item said 'no entry on CSF1R at all', which the grep refuted; recorded because an id-level check is not a mention-level check and the two were conflated.

    Evidence that argues against this (1)
    • The paper is from January 2026 and this corpus's cycle cadence means a nine-month-old paper being uncited is unremarkable rather than a coverage failure. The more useful observation is not that it is uncited but that the graph has no CSF1R entry at all despite CSF1R being a clinically drugged receptor in the myeloid compartment. Camiolo G et al. 2026
Connected to