Oncofusion

start / graph / hyp-q147-irx1-occupancy-is-a-lineage-switch-state-confounded-with-one-institution

Hypothesisweak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.Q147-hoxa-irx1-across-partners

IRX1 occupancy is a lineage-switch cell state, and in GSE252378 it is confounded with the AFF1 partner and one institution

The three samples where the fusion sits on IRX1 share three things, not one: the same fusion partner, the same hospital, and the same unusual disease behaviour where the leukaemia changes lineage. Any of the three could be the reason.

The same thing, for a scientist

**The competing reading of this beat's own result, written out so it cannot be quietly dropped.** The three IRX1-occupied samples in GSE252378 are p279, p318 and p454. All three are KMT2A::AFF1. **All three are from Texas Children's Hospital.** All three are infant lineage-switch-associated cases [E1]. **And the deposit's own paper predicts the cell-state version.** It reports that in lineage-switching samples the oncoprotein levels are reduced and the oncoproteins preferentially activate granulocyte-monocyte-progenitor genes rather than the ALL or AML programmes they activate at high levels [E2]. A HOXA-low, alternative-address state is exactly what that model produces - **with no partner effect required.** Under this hypothesis the IRX1 address would be reachable in any KMT2A partner given the same cell state, and GSE252378 simply has no untreated non-AFF1 sample in that state. **The one non-AFF1 sample that is HOXA-negative is not lineage-switch-associated:** A73051 is a paediatric AML from Fred Hutch with no switch annotation, and it is IRX1-negative [E1]. That is weak evidence against this hypothesis - its cell state is not matched either - and it is the only relevant data point the deposit provides. **Which hypothesis is currently better supported: this one.** It requires no new mechanism, it is the deposit paper's own model, and it explains the institution clustering, which the partner hypothesis does not. The partner hypothesis has one thing this one does not: an independent published mechanism in the right disease [Kuhn's t(4;11)-specific IRX1/MLL-AF4 interaction]. Neither is established.

Go deeper
What an institution effect would even be

Institution is a proxy, not a cause. It stands in for referral pattern (Texas Children's contributed the lineage-switch cases), collection protocol, time from aspiration to fixation, blast percentage at collection, and patient population. Any of these could shift a chromatin measurement.

The specific worry that would be cheapest to rule out: all three IRX1-occupied samples have the deposit's strongest positive controls (MEIS1 16.7-27.1x, CDK6 9.8-21.5x) and p279 has the deepest libraries in the deposit. If library quality tracks institution, then 'IRX1 detected' could partly be 'IRX1 detectable'. Against that: p179 and p247 are also Texas Children's with equally strong controls (MEIS1 16.8-25.5x) and both read IRX1 at 0.021x and 0.068x - so within the same institution and at the same library quality, IRX1 is on in three samples and off in two. That is the strongest available argument that the IRX1 signal is not an institution artefact.

Facts and figures
claimthe IRX1 occupancy state is a lineage-switch-associated low-oncoprotein cell state reachable in any KMT2A partner, and its apparent AFF1-restriction in GSE252378 is a confound with partner and institution
confounded variables
[
  "fusion partner (all AFF1)",
  "institution (all Texas Children's Hospital)",
  "lineage-switch association (all three)"
]
irx1 occupied
[
  "p279",
  "p318",
  "p454"
]
same institution irx1 negative
{
  "p179": 0.021,
  "p247": 0.068
}
experiments
[
  "exp-q147-kmt2a-n-cut-and-run-in-an-untreated-non-aff1-lineage-switch-blast",
  "exp-q147-swap-the-aff1-c-terminus-onto-kmt2a-af9-and-score-irx1-occupancy",
  "exp-q147-irx1-occupancy-under-partial-menin-inhibition-in-kmt2a-aff1"
]
analyses
[
  "ana-q147-irx1-fusion-occupancy-is-confined-to-kmt2a-aff1",
  "ana-q147-a73051-is-a-hoxa-negative-irx1-negative-kmt2a-af10-blast"
]
falsifiers
[
  "An untreated non-AFF1 infant KMT2A-r lineage-switching blast, depth-adequate (p_dz < 0.05) and passing a MEIS1/CDK6 readability gate, with KMT2A-N at the IRX1 gene body BELOW 2.0x an average promoter in both N antibodies while its HOXA-low state is confirmed. That refutes the cell-state reading and leaves the partner reading. NOT YET RUN - the only such pair in GSE252378 (148752/152985, KMT2A-ENL, Texas Children's) has its myeloid aliquot on revumenib with the oncoprotein globally evicted.",
  "A KMT2A::AFF1 IRX1-occupied sample from an institution other than Texas Children's, which removes the institution leg of the confound. CURRENT STATUS: the deposit has three St. Jude KMT2A::AFF1 samples (TB051939, TB920488, TB986292) and all three read IRX1 below 2.0x - but all three are depth-MARGINAL or INADEQUATE, so they neither support nor refute.",
  "A KMT2A-AF9 construct carrying the AFF1 C-terminus acquiring IRX1 occupancy at matched oncoprotein level in a fixed cell background, which would show the partner is sufficient without any change of cell state. NOT YET RUN."
]
How this could bite us back
  • If this hypothesis is right, a menin inhibitor can drive a patient into the IRX1 state - the deposit paper shows lineage switching under revumenib - so the drug would create the address rather than find it. That is a far more serious clinical claim than the partner reading and it is the reason this hypothesis must not be dropped merely because it is the deflating one.
  • It also implies the IRX1 state is transient and sampling-dependent, which would make any single-timepoint stratification unreliable and would call for serial rather than diagnostic profiling.
Still open
  1. Is 'lineage-switch-associated' even the right cell-state variable, or is it oncoprotein level? The deposit paper treats them as linked, and q144 measured oncoprotein-target promoter counts per sample, so the two could be separated with the data already in hand.
  2. Do the three depth-inadequate St. Jude KMT2A::AFF1 samples become informative if resequenced? They are the only material that could break the institution leg of the confound.
Every source behind this entry (2)
  1. E1moderate confidence 3 of 5: One strong study, or several smaller consistent ones.dataset

    All three IRX1-occupied GSE252378 samples (p279, p318, p454) are KMT2A::AFF1 from Texas Children's Hospital and lineage-switch-associated; the only HOXA-negative non-AFF1 sample (A73051, KMT2A-AF10, Fred Hutch) has no lineage-switch annotation and is IRX1-negative (0.249x against a 0.255x floor). Two further Texas Children's KMT2A::AFF1 samples with equally strong positive controls, p179 and p247, read IRX1 at 0.021x and 0.068x.

    What this evidence cannot show

    The confound is complete and is stated as such: with three positive samples sharing three variables, no analysis inside this deposit can apportion them. The p179/p247 argument bounds the institution explanation but not the cell-state explanation, because both are Texas Children's and neither is annotated as lineage-switching. Every number is a WITHIN-SAMPLE share: an interval's mean coverage divided by that same library's mean coverage over all 19,404 hg38 RefSeq promoters, so 1.0 = an average promoter in that library. GSE252378 has no spike-in, so no between-library absolute statement is licensed; the only cross-sample quantity used is a ratio of two loci measured inside one library, where the unknown per-library scale cancels. The occupancy call is absolute, not rank-based: >=2.0x an average promoter in BOTH KMT2A-N antibodies AND above that library's own floor, where the floor is the 95th percentile of eight width-matched windows at loci that cannot be occupied in a leukaemia (ALB, RHO, MYH7, MYOD1, NEUROD1, KRT14, GFAP, INS). IgG is not the gate: q145 measured GSE252378 IgG libraries as 57-94% zero at promoters, and one of them (TB051939) reads 13.1x an average promoter at a locus that cannot be occupied.

    Evidence that argues against this (1)
    • Lineage-switch annotation in this deposit is free text in the GEO description and is not a structured field; 'lineage-switch-associated' for p454 rests on the paper's description rather than on a characteristic, so the exposure variable itself is imprecisely defined. Janssens DH et al. 2024
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    The deposit's own paper reports that in lineage-switching samples the oncoprotein levels are reduced and the oncoproteins preferentially activate GMP genes, and proposes that MLL oncoproteins promote lineage-switching events through dynamic chromatin binding at lineage-specific target genes.

    MLL oncoprotein levels influence leukemia lineage identities · Janssens DH, Duran M, Otto DJ, Wu W, Xu Y, Kirkey D, Mullighan CG, Yi JS, Meshinchi S, Sarthy JF, Ahmad K, Henikoff S · Nat Commun 15:9341 2024
    PMID 39472576DOI
    What this evidence cannot show

    The paper's central model, cited as the mechanistic basis of this hypothesis rather than as evidence for IRX1 specifically - it does not mention IRX1. The step from 'low oncoprotein redirects to GMP genes' to 'low oncoprotein redirects to IRX1' is an extrapolation and IRX1 is not a GMP gene.

    Evidence that argues against this (1)
    • If the low-oncoprotein state redirects to GMP genes specifically, then IRX1 - a pro-B and megakaryocyte-erythroid factor, not a GMP factor - is the wrong destination for that model, which weakens this hypothesis rather than supporting it. Nagel S 2023
Connected to