The HOXA-negative, IRX1-negative KMT2A-rearranged blast
Some KMT2A-driven leukaemias have the fusion protein on neither HOXA nor IRX1. That breaks the idea that these two are the only two options, and nobody knows what the fusion is doing in those cells instead.
The same thing, for a scientist
The published subgroup model for infant t(4;11) is a **dichotomy**: iALL-HOXA9 or iALL-IRX1, mutually exclusive by expression. Measured as oncoprotein occupancy across GSE252378's 27 samples, a **third state exists and is not rare**: the fusion clears the absolute floor at neither locus. Two clear instances, both depth-usable and both with working libraries: - **TB920488** (KMT2A::AFF1, St. Jude): HOXA 0.97x, IRX1 0.59x, against strong positive controls (MEIS1 5.9-11.2x, CDK6 5.4-23.2x). Depth MARGINAL (p_dz 0.135) [E1]. - **A73051** (KMT2A-AF10, Fred Hutch): HOXA 0.87x, IRX1 0.25x, depth ADEQUATE (p_dz 0.0120) but with the deposit's weakest positive controls [E1]. And a third, drug-associated: **152985** (KMT2A-ENL, on revumenib) is HOXA 0.97x, IRX1 0.68x - but it fails the readability gate entirely, which is itself the expected consequence of menin inhibition and means it cannot be counted as this state [E1]. **What the fusion is doing in these cells is not measured here.** The deposit's own paper supplies the most likely answer: low-oncoprotein samples preferentially activate granulocyte-monocyte-progenitor genes rather than the lymphoid or stem programmes [E2]. That makes the double-negative state a candidate *third programme* rather than an absence, and it is testable in this deposit with a GMP gene set.
Why a third state matters for the two-subgroup model
Symeonidou & Ottersbach's split is an expression dichotomy derived from clustering two RNA-seq cohorts, and a clustering that offers two labels will assign every sample one of them. An occupancy measurement can return neither, and does.
The practical consequence: a stratification built on 'HOXA-high or else IRX1-high' mis-assigns the double-negative patients, and in this deposit they are 2 of 26 untreated samples by occupancy - and 4 of 8 KMT2A::AFF1 samples are HOXA-negative, of which 2 are IRX1-occupied, 1 equivocal and 1 negative.
What has to be true for this state to be real rather than an instrument artefact
Three things, and only the first two are established here:
- The library reports occupancy somewhere. TB920488 does (CDK6 up to 23.2x); A73051 barely does (MEIS1 2.4-2.6x only).
- The zeros clear a floor test rather than being shallow. A73051 is depth-ADEQUATE. TB920488 is MARGINAL, so its IRX1 0.59x is weakly measured.
- Something else is occupied instead. NOT measured. This is the gap, and the GMP gene set from the deposit paper is the obvious first test.
| excluded instance | 152985 (fails readability under revumenib) |
|---|---|
| candidate third programme | granulocyte-monocyte progenitor gene set (not scored here) |
instances
[ "TB920488", "A73051" ]
tb920488
{
"partner": "KMT2A-AFF1",
"hoxa": 0.968,
"irx1": 0.588,
"meis1_range": [
4.45,
11.2
],
"depth": "MARGINAL"
}a73051
{
"partner": "KMT2A-AF10",
"hoxa": 0.87,
"irx1": 0.249,
"depth": "ADEQUATE"
}analyses
[ "ana-q147-irx1-fusion-occupancy-is-confined-to-kmt2a-aff1", "ana-q147-a73051-is-a-hoxa-negative-irx1-negative-kmt2a-af10-blast" ]
- A patient in the double-negative state has no fusion-occupied HOXA address for a menin inhibitor to act on, and in this deposit that state coincides with low oncoprotein overall. If it is a resistance state rather than a low-burden state, stratifying trials on HOXA occupancy would systematically exclude the patients who most need an alternative.
- Conversely, if the double-negative state is simply a low-tumour-burden aliquot, treating it as a biological subgroup would create a phantom resistance mechanism out of a sampling artefact. The two are not distinguishable without a blast-percentage or a spike-in, and GSE252378 has neither.
- Is a GMP gene set fusion-occupied in TB920488 and A73051? The deposit paper's own gene sets are published and the bigWigs are in hand, so this is directly answerable and was not done.
- How common is the double-negative state in an unselected KMT2A-r cohort? 2 of 26 here, but the AFF1 arm of this deposit was selected for lineage switching.
- E1moderate confidence 3 of 5: One strong study, or several smaller consistent ones.dataset
In GSE252378, TB920488 (KMT2A::AFF1) has KMT2A-N shares of 0.97x at HOXA9-HOXA10 and 0.59x at the IRX1 gene body while CDK6 reaches 23.2x and MEIS1 11.2x in the same libraries; A73051 (KMT2A-AF10) has 0.87x and 0.25x at depth-ADEQUATE (p_dz 0.0120). Neither clears the pre-registered occupancy floor at either locus.
q147 re-analysis of GSE252378: HOXA and IRX1 oncoprotein occupancy across all six KMT2A fusion partners, 27 samples, 133 libraries (analyses/q147-hoxa-irx1-across-partners) · Iqbal Lab oncofusion knowledge graph, beat q147 · internal analysis over deposited GSE252378 bigWigs 2026What this evidence cannot show
Two samples, from two partners and two institutions, one depth-MARGINAL and one with the deposit's weakest positive controls. Neither is a clean example and the state is therefore proposed rather than established. A third candidate (152985) is excluded because it fails the readability gate under revumenib. 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)
- The published subgroup model is an EXPRESSION dichotomy and would assign both these samples to one of its two clusters; no chromatin measurement exists in that paper, so a third occupancy state is not in conflict with it - it is a different measurement that the two-label model has no slot for. Symeonidou V et al. 2021
- 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 granulocyte-monocyte progenitor genes, rather than the pro-B-cell-enriched ALL programme or the haematopoietic-stem-cell-enriched AML programme they activate at high levels.
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 2024What this evidence cannot show
This is the paper's own model for low-oncoprotein samples and is cited to name the most likely third programme, not to assert that these two samples are in it - no GMP gene set was scored in this beat.
Evidence that argues against this (1)
- The GMP redirection in that paper is described for lineage-switching samples specifically. Neither TB920488 nor A73051 is annotated as lineage-switching in the deposit, so importing the GMP explanation to them is an extrapolation. Janssens DH et al. 2024
answers that entry's own open question: is a GMP gene set fusion-occupied in TB920488 and A73051
scores the GMP gene set in the second of that entry's two named instances
proposes a mechanism for that cell state: dosage rather than redirection