start / graph / ana-q147-prom1-occupancy-does-not-track-hoxa-outside-kmt2a-aff1-and-is-absent-in-af9-and-afdn

AnalysismoderateQ147-hoxa-irx1-across-partners

PROM1 occupancy tracks HOXA at +0.93 in KMT2A::AFF1 and not at all elsewhere; it is near-absent in KMT2A-AF9 and -AFDN despite high HOXA

PROM1 (CD133) was the second marker that seemed to move with HOXA in t(4;11). Across the other fusion partners it does not - and in two partner groups the fusion sits heavily on HOXA while ignoring PROM1 entirely.

The same thing, for a scientist

q145 reported PROM1, Symeonidou's independent iALL-HOXA9 marker, tracking HOXA occupancy across seven KMT2A::AFF1 patients at Spearman **+0.952**. Reproduced here exactly at promoter scale (+0.9524, exact two-sided p = 0.0011) and **+0.9286 at gene scale** (p = 0.0022) [E1]. **It does not reproduce in KMT2A-ENL**, the only other group that can carry a correlation: **+0.086 at gene scale and +0.086 at 2 kb promoter scale, exact two-sided p = 0.919 over all 720 permutations** - flat at both windows [E1]. 🔴 **KING CORRECTION (cycle 131).** This entry originally reported **+0.400 (p = 0.517) at gene scale and -0.100 (p = 0.950) at promoter scale**, and called it a sign flip. Those come from the **superseded n = 5 ENL sample set**, which dropped `148752` as menin-inhibitor-treated - but `148752` is the **untreated diagnosis sample** (the deposit's *treated* string is a case-level label, recorded in this corpus two cycles earlier), so the correct group is n = 6. The beat used n = 6 for its IRX1 numbers and n = 5 here, **mixing two runs inside one entry.** On n = 6 the value is **+0.086 at both scales**, so **there is no sign flip** and the PROM1 half of that argument is withdrawn. The -0.100 matches no intended window pairing at all. **The conclusion is unchanged and strengthened:** +0.086 with p = 0.919 is flatter than +0.400 was. And the stronger, simpler observation, which needs no correlation: **PROM1 is essentially unoccupied in KMT2A-AF9 (0.069-0.430x) and KMT2A-AFDN (0.090-0.308x) while HOXA in those same samples is 5.7-29.1x** [E1]. PROM1 is high (>4x) in AFF1, ENL and AF10 samples and flat in AF9, AFDN and EPS15. **So PROM1 is not a general HOXA co-target; its co-occupancy with HOXA is restricted to a subset of partners.** This matters beyond PROM1: it means the q145 result generalises to **neither** of its two axes. The IRX1 anti-correlation and the PROM1 co-correlation are both KMT2A::AFF1 findings.

Go deeper

PROM1 by partner

Gene-scale, 8 kb PROM1 promoter window, KMT2A-N mean of two antibodies, untreated samples.

partner	n	PROM1 range	HOXA range	PROM1 >= 2.0x
KMT2A-AFF1	8	1.02 - 7.72x	0.11 - 11.78x	6/8
KMT2A-ENL	6	0.89 - 12.30x	7.10 - 17.65x	4/6
KMT2A-AF10	3	0.43 - 7.97x	0.87 - 20.17x	2/3
KMT2A-AF9	3	0.07 - 0.43x	5.69 - 29.11x	0/3
KMT2A-AFDN	4	0.09 - 0.31x	11.03 - 21.63x	0/4
KMT2A-EPS15	2	0.31 - 0.44x	2.94 - 5.74x	0/2

TB040576 (KMT2A-AF9) has the highest HOXA in the whole deposit at 29.1x and PROM1 at 0.069x - a 420-fold ratio between two loci the published model treats as co-regulated. That single sample is the cleanest refutation of PROM1 as a general HOXA co-target.

A lineage confound is unresolved and may be the whole explanation: the AF9, AFDN and EPS15 samples are 9 of 9 from Fred Hutch or Seattle paediatric AML / infant B-ALL lineage-switch cases, while the PROM1-high samples are concentrated in the B-ALL arms. PROM1/CD133 is a stem and progenitor surface marker, and partner and lineage are confounded in this deposit by design - AF9 and AFDN are collected as AML, AFF1 largely as B-ALL.

What this does to the q145 entry

Nothing is retracted. finding-prom1-occupancy-tracks-hoxa-occupancy-across-seven-patients is a statement about seven KMT2A::AFF1 patients and it reproduces exactly. What this adds is a scope boundary: the tracking is not a property of the HOXA programme, it is a property of the HOXA programme in these partners. The recommended amendment, reported not applied, is a caveat line naming the AF9 and AFDN measurements.

Facts and figures

prom1 window	PROM1 TSS +/- 4 kb (8,000 bp, hg38 chr4:16,080,023-16,088,023)
rho aff1 promoter	0.9524
p aff1 promoter	0.00114
rho aff1 gene	0.9286
p aff1 gene	0.00223
rho enl gene	0.0857
p enl gene	0.9194
rho enl promoter	0.0857
p enl promoter	0.9194
entry to scope	finding-prom1-occupancy-tracks-hoxa-occupancy-across-seven-patients
n enl	6
corrected cycle 131	rho_enl_gene/promoter were 0.400 and -0.100 from the superseded n=5 ENL set that wrongly excluded the untreated diagnosis sample 148752; king recomputed on n=6 from results_regions147.json and results_gene_scale147.json: +0.0857 at both scales, exact two-sided p = 662/720 = 0.9194.

prom1 range af9

[
  0.069,
  0.43
]

hoxa range af9

[
  5.689,
  29.108
]

prom1 range afdn

[
  0.09,
  0.308
]

hoxa range afdn

[
  11.027,
  21.631
]

extreme sample

{
  "id": "TB040576",
  "partner": "KMT2A-AF9",
  "HOXA": 29.108,
  "PROM1": 0.069,
  "ratio": 421
}

analyses

[
  "ana-q147-irx1-fusion-occupancy-is-confined-to-kmt2a-aff1"
]

How this could bite us back

CD133/PROM1 is being pursued as a surface target in several leukaemias. If its fusion occupancy is restricted to a subset of KMT2A partners, a CD133-directed agent selected on KMT2A-rearrangement alone would be aimed at a target the fusion is not driving in AF9 and AFDN cases - and PROM1 is also expressed on normal haematopoietic and epithelial stem cells, so the on-target cost does not shrink with the benefit.

Still open

Is the PROM1 absence in AF9/AFDN a lineage effect or a partner effect? A KMT2A-AF9 B-ALL sample would separate them; there is none in this deposit.
Does PROM1 surface expression by flow agree with its occupancy here? CD133 is routinely measured in these diseases and the deposit's samples are clinical aliquots.

Every source behind this entry (1)

E1moderate confidence 3 of 5: One strong study, or several smaller consistent ones.dataset
In GSE252378: Spearman(HOXA, PROM1) is +0.952 (promoter scale, exact two-sided p = 0.0011) and +0.929 (gene scale, p = 0.0022) across the eight KMT2A::AFF1 sample records; in the six untreated KMT2A-ENL samples it is +0.086 (gene, p = 0.919) and -0.100 (promoter, p = 0.950). PROM1 shares are 0.069-0.430x in the three KMT2A-AF9 samples and 0.090-0.308x in the four KMT2A-AFDN samples, while HOXA in those same samples is 5.69-29.11x; TB040576 has HOXA 29.11x with PROM1 0.069x.
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 2026
What this evidence cannot show
Partner and lineage are confounded in this deposit: all AF9 and AFDN samples are paediatric AML from Fred Hutch, and PROM1/CD133 expression is lineage- and stage-dependent, so the PROM1 absence in those groups may be a lineage property and not a partner property. This analysis cannot separate them and does not claim to. 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 (2)
- PROM1 was proposed as an iALL-HOXA9 marker on EXPRESSION grounds in MLL-AF4 specifically, and Symeonidou & Ottersbach make no claim about other partners, so a null outside AFF1 does not contradict them. Symeonidou V et al. 2021
- The deposit's own paper reports that the oncoprotein activates a haematopoietic-stem-cell-enriched AML programme in AML samples and a pro-B-enriched ALL programme in ALL samples; PROM1 sitting with HOXA in the B-lineage arms and not in the AML arms is consistent with that lineage split rather than with a partner effect. Janssens DH et al. 2024

Connected to