In the only cohort that measured it, HOXA genes are ON in exactly the reciprocal-positive patients and OFF in the rest
also called HOXA and AF4-MLL · HOXA-positive t(4;11) · reciprocal-HOXA collinearity
In the one study that looked, the HOXA genes - the ones this whole field treats as the readout of the disease - were switched on only in the patients whose leukaemia made the mirror-image fusion. The two things are the same split in the same patients.
The same thing, for a scientist
In 124 Interfant-treated infants, HOXA cluster genes were expressed exclusively in the AF4-MLL-expressing patients [E1]. That is not a correlation to be adjusted for; in that dataset it is the same partition. The consequences run in two directions at once. For the prognostic result, reciprocal status and HOXA status cannot be separated, so the survival effect could be a HOXA effect [E1]. For the menin-interface hypothesis, it is worse than that: HOXA output is the principal pharmacodynamic readout of menin inhibition, and it is reported here as belonging to the reciprocal-positive subgroup - yet the reciprocal product has no menin-binding motif [E2]. The menin-blind KMT2A architecture for which independent evidence exists, NUP98::KMT2A, likewise shows reduced rather than elevated HOXA and is reported menin-inhibitor resistant [E3]. Two independent observations therefore associate HOXA status with the presence or absence of a menin-binding KMT2A product in ways that do not line up simply, and this graph should stop treating HOXA as a clean proxy for menin-dependent fusion activity in t(4;11).
What would have to be checked before this is load-bearing
The collinearity is reported in one sentence of one paper's abstract, in one cohort, at n = 43 for the t(4;11) subgroup 1. Whether it is exact or approximate is not stated; whether the HOXA measurement is RNA-seq derived from the same samples is stated only implicitly. Before this is used to argue anything about menin, someone should read the paper's supplementary tables and count patients in the four cells of the HOXA-by-reciprocal table. This beat did not do that.
Why it matters for pharmacodynamics
If HOXA is high only where the reciprocal product is present, then a menin inhibitor's HOXA readout in a t(4;11) patient is being measured in exactly the patients who also carry a KMT2A oncoprotein the drug cannot bind. A HOXA fall would then not be attributable to the direct product with any confidence. This is the same class of error as measuring total H3K79me2 as a pinometostat endpoint: the signal is dominated by a unit the drug was not aimed at.
| observation | HOXA cluster expression exclusive to reciprocal-expressing patients |
|---|---|
| cohort | Interfant 99/06, 124 infants, 43 t(4;11) |
| collinearity | complete in that dataset as reported |
| consequence | the reciprocal prognostic effect and any HOXA effect cannot be separated in the only cohort that measured either |
| cross tabulation read | false |
- A menin inhibitor's HOXA pharmacodynamic readout in t(4;11) may be reporting on a subgroup defined by a protein the drug cannot bind.
- If HOXA-negative t(4;11) is the majority and the worse-outcome group, then a therapy whose rationale is HOXA suppression is aimed at the patients who do better anyway.
- What does the four-cell HOXA-by-reciprocal table actually contain? The abstract says 'exclusively' and the supplement would say how exclusive.
- Does any second cohort measure both? None was found for this beat.
- If HOXA output belongs to the reciprocal-positive subgroup, what drives HOXA in the reciprocal-negative half, which is the majority and has the worse outcome?
- Is HOXA a valid pharmacodynamic endpoint for menin inhibition in t(4;11) at all, given this association?
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
In 124 infant B-ALL patients treated on Interfant 99/06, HOXA cluster genes were expressed exclusively in the AF4-MLL-expressing patients, who were 45% (19/43) of t(4;11) cases and had significantly better event-free and overall survival.
Unraveling the cellular origin and clinical prognostic markers of infant B-cell acute lymphoblastic leukemia using genome-wide analysis. · Agraz-Doblas A, Bueno C, Bashford-Rogers R, et al. · Haematologica 2019What this evidence cannot show
One cohort, n = 43 for the t(4;11) subgroup; the word 'exclusively' is the abstract's and the underlying cross-tabulation is not given there. Whether the association is exact, and whether it survives in other cohorts, is untested.
Evidence that argues against this (1)
- A partner-stratified paediatric cohort reporting outcome by KMT2A fusion partner measured neither HOXA status nor reciprocal status, so no independent dataset can currently confirm or break the collinearity. Qiu KY et al. 2022
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Read in full text (PMC8530991): the reciprocal t(4;11) fusion gene comprises AF4 exons 1-3 fused to MLL exons 12-37, and this infant-typical intron-11 version was the construct used throughout that study.
The role of reciprocal fusions in MLL-r acute leukemia: studying the chromosomal translocation t(4;11). · Wilhelm A, Marschalek R · Oncogene 2021What this evidence cannot show
Read in full text via Europe PMC fullTextXML (40,194 characters), not abstract only. Cited here for the exon arithmetic alone. That a product beginning at KMT2A exon 12 carries no menin-binding motif is an inference from UniProt coordinates placing that motif at residues 6-25, not a statement in this paper, and no menin-binding experiment has been performed on any reciprocal KMT2A product.
Evidence that argues against this (1)
- Absence of a binding motif is not absence of a functional relationship: menin has other interactors, including JUND through the same MLL-binding pocket, so a menin-independent architecture is not necessarily a menin-irrelevant one. Ozyerli-Goknar E et al. 2026
- E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
NUP98::KMT2A, which lacks the menin-binding exon while retaining the WIN motif and SET domain, produces leukaemia with very low Hoxa5, Hoxa7, Hoxa9, Hoxa10 and Meis1 and is reported resistant to menin inhibition.
Transforming activities of the NUP98-KMT2A fusion gene associated with myelodysplasia and acute myeloid leukemia. · Fisher JN, Thanasopoulou A, Juge S, et al. · Haematologica 2020What this evidence cannot show
Murine inducible transgenic with long latency; resistance shown with a tool compound; a single rearrangement rather than a class, and the HOXA direction here is the opposite of the t(4;11) observation, which is why the two together are confusing rather than confirmatory.
Evidence that argues against this (1)
- The direct KMT2A::AFF1 product is menin-binding and is associated with HOXA activation in most of the literature, so an account in which HOXA tracks the menin-blind product in one system and the menin-competent product in another needs more than two data points. Agraz-Doblas A et al. 2019
collinear cluster activation is the null expectation against which any HOXA7-specific claim must be read
The confound that the prognostic estimate cannot adjust for, because it is the same partition.
- related to →HOXA9 — required for survival of established MLL-rearranged leukemia, and dispensable for starting it
HOXA cluster expression partitions with reciprocal transcript status in the Interfant cohort.
Challenges HOXA as a clean pharmacodynamic proxy for menin-dependent fusion activity in t(4;11).
Adds a pharmacodynamic consequence to that hypothesis: the HOXA readout tracks the menin-blind subgroup.
Independently connects reciprocal presence to HOXA output, via H3K79 methylation in double-fusion human cells.