The myeloid/lymphoid metabolic split in KMT2A-rearranged leukaemia is set by CELL OF ORIGIN (fetal versus adult), not by lineage or by the fusion partner
also called ontogeny hypothesis for metabolic dependency · fetal origin sets the metabolic set-point
Myeloid and lymphoid KMT2A leukaemias want opposite things from their mitochondria, and everyone assumes that is because one is myeloid and one is lymphoid. There is a second explanation: the lymphoid ones start before birth from fetal blood cells and the myeloid ones mostly do not, so the difference may be about when the leukaemia began rather than what kind of cell it became.
The same thing, for a scientist
Every metabolic axis in this beat that has been measured in both lineages gives opposing answers. **OXPHOS:** MLL/AF9 AML depends on oxidative phosphorylation and is repressed by metformin [E1]; in MLL-rearranged B-ALL, OXPHOS inhibition blunts blasts but **promotes leukaemia-initiating cell emergence**, while glycolysis/hypoxia inhibition impairs them [E2]. **Lipids:** the myeloid dependency is ACSL4 on polyunsaturated substrates [E3]; the infant lymphoid requirement is ELOVL1 on saturated very-long-chain substrates, and crucially it was identified **from within a conserved fetal transcriptional signature retained in infant ALL blasts** [E4]. **Nucleotides:** IMPDH-inhibitor synergy was present in a subset of AML cells and explicitly **absent in ALL cell lines** [E5]. The universal reading is *lineage*. This hypothesis proposes the alternative: **ontogeny**. Infant KMT2A::AFF1 ALL arises prenatally from fetal haematopoietic progenitors and retains a fetal programme; the myeloid MLL-fusion models are adult or postnatal. Fetal and adult haematopoietic stem cells differ in metabolic set-point independently of lineage, so the observed split may track *when the cell of origin existed* rather than *what lineage it became*. The lipid evidence is the sharpest hint, because the lymphoid dependency was found by asking a developmental question, not a metabolic one [E4].
Why this is worth testing rather than merely noting
Because the two explanations make different predictions about what to do with a patient.
- If the split is lineage, then a myeloid-derived metabolic strategy is inapplicable to B-ALL and vice versa, permanently, and the correct response is two separate programmes.
- If the split is ontogeny, then the relevant variable is the age and developmental origin of the disease, and a metabolic strategy validated in infant disease might transfer to infant AML (which exists, is KMT2A-rearranged in a large fraction, and is also of prenatal origin) while failing in adult B-ALL with the same fusion. That is a testable, clinically actionable reordering of who gets what.
The second prediction is the discriminating one and nobody has tested it: adult t(4;11) B-ALL is the natural control. It has the index fusion and the index lineage but not the fetal origin. If its metabolic phenotype resembles adult myeloid disease rather than infant t(4;11), ontogeny wins.
Why the existing literature cannot answer it
The confound is total. In published KMT2A-rearranged models, lineage and ontogeny are almost perfectly correlated: the lymphoid models are infant- or fetal-derived and the myeloid models are adult- or postnatal-derived. This graph's own DepMap notes record a related version of the same problem - the KMT2A-rearranged arm in DepMap is myeloid-enriched, so a comparison against a mixed comparator mixes fusion status with lineage, and two ranked hits in this project have already turned out to be lineage artefacts.
Add ontogeny as a third correlated variable and no observational analysis can separate them. Only an isogenic experiment can.
| statement | The opposing metabolic phenotypes of myeloid and lymphoid KMT2A-rearranged leukaemia are determined by the developmental stage of the cell of origin (fetal versus adult/postnatal), not by lineage identity and not by the fusion partner. |
|---|---|
| origin | data-pattern |
| prior probability | 0.25 |
| why it cannot be answered observationally | in published KMT2A models lineage and ontogeny are almost perfectly correlated, so no amount of reanalysis of existing data separates them |
predictions
[ "Adult t(4;11) B-ALL - index fusion, index lineage, NON-fetal origin - will have a metabolic phenotype resembling adult myeloid KMT2A-rearranged disease rather than infant t(4;11).", "Infant KMT2A-rearranged AML, which is also of prenatal origin, will resemble infant KMT2A-rearranged ALL metabolically more than it resembles adult KMT2A-rearranged AML.", "The same fusion introduced into fetal-derived versus adult-derived progenitors of the SAME lineage will produce divergent OXPHOS/glycolysis set-points.", "ELOVL1 dependence will track fetal origin rather than B-lineage identity, and will therefore appear in fetal-derived myeloid models." ]
experiments
[ "exp-isogenic-fusion-fetal-versus-adult-progenitor-metabolic-readout" ]
analyses
[ "ana-metabolic-enzyme-enumeration-kmt2a-bounded-absences" ]
falsifiers
[ "Adult t(4;11) B-ALL shows the infant t(4;11) metabolic phenotype (glycolysis/hypoxia dependence, OXPHOS inhibition expanding initiating cells) despite non-fetal origin. That is straightforward lineage and the hypothesis is dead.", "The same fusion in fetal-derived versus adult-derived progenitors of the same lineage gives indistinguishable Seahorse and tracer profiles.", "ELOVL1 is dispensable in fetal-derived myeloid models carrying a KMT2A fusion, or ACSL4 is required in fetal-derived B-lymphoid models - either result breaks the ontogeny mapping of the lipid split.", "The ALL negative in the IMPDH studies is fully explained by SAMHD1 protein levels, removing one of the three axes; two independent axes removed this way would leave the pattern unsupported." ]
- If this hypothesis is right, a metabolic agent validated in adult or paediatric myeloid KMT2A-rearranged AML should not be given to an infant with t(4;11) on genotype grounds - and genotype is exactly what a precision-medicine protocol would use to assign it.
- If it is wrong and the split is plain lineage, then effort spent building fetal-origin models is effort not spent on the two separate lineage-specific programmes that are actually needed.
- The hypothesis is attractive partly because this graph already believes in fetal origin for infant KMT2A::AFF1, which is a reason to be more suspicious of it, not less: it is the explanation this project was primed to reach for.
- Does anyone hold a cohort of ADULT t(4;11) B-ALL samples with viable material for a Seahorse or tracer experiment? That single comparison is the cheapest discriminating test and requires no new model building.
- Do fetal and adult haematopoietic stem cells differ in OXPHOS/glycolysis set-point by enough to survive transformation? If the normal difference is small, the hypothesis has no room to work.
- Is infant KMT2A-rearranged AML metabolically more like infant ALL or adult AML? Nobody has asked, and infant AML samples exist in the paediatric PDX platforms this beat already cites.
- Could the split be neither lineage nor ontogeny but assay - bulk proliferation in the myeloid studies versus limiting-dilution transplantation in the lymphoid one? Running both readouts in one system would test this before any ontogeny experiment is worth doing.
- E1moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
AML cell lines show high metabolic heterogeneity; cells with MLL/AF9 have upregulated mitochondrial activity and depend mainly on oxidative phosphorylation for energy production, and metformin repressed their proliferation by inhibiting mitochondrial respiration.
High Metabolic Dependence on Oxidative Phosphorylation Drives Sensitivity to Metformin Treatment in MLL/AF9 Acute Myeloid Leukemia. · Liu L, Patnana PK, Xie X, ..., Khandanpour C · Cancers (Basel) 2022What this evidence cannot show
Abstract read this session; identifiers re-fetched from PubMed rather than copied from sibling entries. Cell lines only, small-n genotype comparison within a heterogeneous panel. Metformin is a weak, pleiotropic complex I inhibitor requiring OCT1-mediated uptake and is a poor mechanistic probe. Cell lines have no ontogeny in any meaningful sense, which is itself a limitation on using them to argue about developmental origin.
Evidence that argues against this (1)
- The lymphoid study reaches the opposite operational conclusion about the same axis, and it uses quantitative xenotransplantation rather than proliferation, which is a readout the cell-line study cannot provide. The disagreement may be an assay difference (bulk proliferation versus initiating capacity) rather than a biological one, which would defeat both the lineage and the ontogeny readings. Morris V et al. 2022
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
In MLL-rearranged B-ALL, single-cell transcriptomics and quantitative xenotransplantation showed that while inhibiting oxidative phosphorylation blunts blast proliferation, this intervention promotes leukaemia-initiating cell emergence, whereas inhibiting hypoxia and glycolysis impairs leukaemia-initiating cells and provides therapeutic benefit in xenotransplantation systems; engraftable leukaemia-initiating cells in MLL-rearranged B-ALL are abundant compared with reported frequencies in AML.
Hypoxic, glycolytic metabolism is a vulnerability of B-acute lymphoblastic leukemia-initiating cells. · Morris V, Wang D, Li Z, ..., Rowe RG · Cell Rep 2022What this evidence cannot show
Abstract read this session; identifiers re-fetched from PubMed. Xenotransplantation engraftment is a proxy for leukaemia-initiating capacity. The OXPHOS inhibitor is not named in the abstract. MLL-rearranged B-ALL as a class - infants are not separately reported, which is precisely the stratification this hypothesis needs and cannot get from the published work.
Evidence that argues against this (1)
- A paediatric AML PDX platform reports that complex I inhibition combined with venetoclax reduces KMT2A-rearranged AML progression at both onset and relapse. Both studies are well conducted and in different diseases, so the contradiction is not resolvable by weighing quality. Da Ros A et al. 2026
- E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
ACSL4, acting on polyunsaturated lipid species, is a selective and functionally validated vulnerability in KMT2A-rearranged AML, with exogenous PUFA supplementation rescuing the growth defect of ACSL4 loss.
ACSL4-associated lipid metabolism is a distinct therapeutic vulnerability in KMT2A-rearranged acute myeloid leukemia. · Schafer S, Rahimian E, Schmitz-Hubsch L, ..., Wurm AA · Cell Rep 2026What this evidence cannot show
Abstract read this session. Myeloid only. Derived from CRISPR screening data whose KMT2A-rearranged arm is myeloid-enriched, then functionally validated - the validation carries the claim, not the screen. ACSL4 has never been tested in lymphoid KMT2A-rearranged cells, so the lineage contrast in this hypothesis is a juxtaposition of two studies, not a measured interaction.
Evidence that argues against this (1)
- Because neither lipid enzyme has been tested in the other lineage's system, the apparent split could be an artefact of which discovery route was applied where: a CRISPR dependency screen and a fetal-signature analysis are capable of finding different things, and each found the one its method could see. Symeonidou V et al. 2021
- E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
A conserved fetal transcriptional signature, defined across human and mouse fetal, neonatal and adult haematopoietic stem and progenitor cells, persists in infant ALL blasts; functional validation from within that signature identified genes critical for survival of MLL-AF4-positive ALL cells, of particular interest PLK1 and ELOVL1, the latter highlighting altered fatty acid metabolism as a feature of infant ALL.
Defining the fetal origin of MLL-AF4 infant leukemia highlights specific fatty acid requirements. · Symeonidou V, Jakobczyk H, Bashanfer S, ..., Ottersbach K · Cell Rep 2021What this evidence cannot show
Abstract read this session. **This is the load-bearing evidence for the hypothesis and it is suggestive rather than decisive:** it shows that a metabolic requirement of infant MLL-AF4 ALL sits inside a retained FETAL programme, which is consistent with ontogeny setting the metabolic set-point, but the study did not compare fetal-origin against adult-origin leukaemia of the same lineage and fusion. The comparator is developmental stage in normal HSPCs, not in leukaemia.
Evidence that argues against this (1)
- A gene being inside a retained fetal signature does not establish that its requirement is caused by fetal origin rather than by B-lymphoid identity; the fetal signature is defined in HSPCs and the dependency is measured in B-lineage blasts, so lineage remains an equally good explanation of the same data. Schafer S et al. 2026
- E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
IMPDH-inhibitor synergy with cytarabine and fludarabine was SAMHD1-dependent and present in a subset of AML cells but absent in acute lymphoblastic leukaemia cell lines, and sensitivity to IMPDH inhibitors was independent of KMT2A status.
Targeting IMPDH to inhibit SAMHD1 in KMT2A-rearranged leukaemia. · Klootsema Y, Tsesmetzis N, Sharma S, ..., Herold N · Cell Cycle 2026What this evidence cannot show
Abstract read this session. Cell lines, which have no ontogeny, so this row supports a lineage split more naturally than an ontogeny split and is included partly because it is the weakest for the hypothesis. The simplest explanation of the ALL negative is SAMHD1 expression, which the abstract does not report for the ALL lines - so it may not be a lineage finding at all.
Evidence that argues against this (1)
- If the ALL negative is explained by low SAMHD1 expression rather than by lineage or ontogeny, this row supports neither reading and the hypothesis loses one of its three axes. That is the cheapest way to weaken it and it requires only a western blot. Yague-Capilla M et al. 2026
a fetal-programme metabolic requirement in the infant lymphoid disease
states the sharper testable version of the lineage confound
the crossed ontogeny-by-lineage design that discriminates the two explanations
rests on the retained fetal programme of infant KMT2A::AFF1 ALL
proposes ontogeny rather than lineage as the variable behind the OXPHOS split
predicts adult t(4;11) B-ALL will not share the infant metabolic phenotype