Competing model 3: reciprocal status is a proxy for where the break fell, and the thing that changed is the DIRECT product
also called breakpoint position as the real variable · direct product length hypothesis · reciprocal status as proxy
A third explanation: the mirror-image gene is not doing anything. Whether a patient has it depends on exactly where the chromosome broke - and that same break also decides how much of KMT2A ends up in the famous fusion protein. The break, not the mirror image, may be what matters.
The same thing, for a scientist
The two t(4;11) products are complementary: every residue the direct product gains, the reciprocal product loses. A break in KMT2A intron 9 gives a shorter direct product and a longer reciprocal one; a break in intron 11 gives a longer direct product and a reciprocal one whose PHD1 is truncated [E1]. UniProt places recurrent translocation breakpoint sites at Q03164 1362/1363, 1406/1407 and 1444/1445, spanning PHD1 at 1431-1482 [E2]. Breakpoint position also determines whether a reciprocal transcript is produced at all - about half of KMT2A::MLLT1 cases break upstream of the partner's first exon and make none [E3]. So reciprocal status is, mechanically, partly a readout of breakpoint class. This hypothesis proposes that the 62.4%-versus-11.7% survival split [E4] is carried by the direct product's composition, not by the reciprocal product's presence, and that reciprocal status is a marker of it. It differs from the existing 'enforceable remainder' hypothesis in this graph, which proposes that the reciprocal product protects when it is structurally complete: here the reciprocal product does nothing at all and the causal agent is on the other allele.
Why this is not idle
The strongest direct test of whether the reciprocal product does anything in established disease is negative: siRNA against AF4-MLL in RS4;11 and SEMK2-M1 changed nothing, while siRNA against MLL-AF4 inhibited growth in both 5. If the reciprocal product is inert in maintenance, then a 50-point survival difference attributed to it must be carried by something correlated with it, and the direct product's own composition is the most mechanically inevitable candidate, because the same break defines both.
What kills it
A cohort in which breakpoint intron and reciprocal transcript status are both recorded, showing the survival effect persists within breakpoint stratum. Or the converse: showing breakpoint intron alone reproduces the split with reciprocal status adding nothing. Interfant-06's multivariable model contained neither variable 4, and an absence of testing is not a negative result.
| statement | The prognostic effect attributed to reciprocal transcript expression in t(4;11) is carried by KMT2A breakpoint position, which simultaneously determines the composition of the direct KMT2A::AFF1 product and whether a reciprocal transcript is produced; the reciprocal product itself is causally inert in established disease. |
|---|---|
| origin | data-pattern |
| prior probability | 0.3 |
predictions
[ "Within a breakpoint stratum, reciprocal status will add little or no prognostic information.", "Breakpoint intron alone will reproduce a substantial part of the 62.4/11.7 split.", "Reciprocal-negative patients will be enriched for one breakpoint class rather than distributed evenly." ]
experiments
[ "exp-reciprocal-status-stratified-menin-inhibitor-reanalysis" ]
analyses
[ "ana-reciprocal-product-enumeration" ]
falsifiers
[ "A cohort with both breakpoint intron and reciprocal transcript status showing the survival effect undiminished within breakpoint stratum kills this hypothesis.", "A conditional experiment removing the der(4) allele from an established human t(4;11) leukaemia and changing its phenotype kills the 'causally inert' clause.", "Demonstration that reciprocal-positive and reciprocal-negative patients have indistinguishable breakpoint distributions removes the proxy mechanism entirely." ]
- If breakpoint position is the real variable, a reciprocal-status biomarker would select patients correctly for the wrong reason and would fail the moment a cohort with a different breakpoint distribution was tested.
- If the direct product's composition is what matters, then menin-inhibitor response may also vary with breakpoint position - a stratification variable nobody records and which is cheap to obtain.
- Does any dataset hold KMT2A breakpoint intron and reciprocal transcript status on the same t(4;11) patients? The Frankfurt archive has the breakpoints and Interfant has the outcomes; whether the patients overlap is unknown to this beat.
- Does direct-product length correlate with anything measurable - HOXA output, menin-inhibitor sensitivity, transcriptional signature?
- Is there a published outcome analysis of KMT2A breakpoint position in t(4;11) at all? This beat found none holding partner constant.
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
KMT2A breakpoints in t(4;11) cluster in introns 9-11 and AF4 breakpoints in introns 3-4; intron 11 breaks are commonest in infant ALL and disrupt the first PHD finger of the reciprocal protein, and the direct and reciprocal constructs are complementary (MLL exons 1-10 :: AF4 exons 4-20, and AF4 exons 1-3 :: MLL exons 12-37).
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
Describes cloned constructs and the distribution of breakpoints, not a per-patient analysis linking breakpoint to outcome. The complementarity is structural and not in dispute; its clinical consequence is entirely untested.
Evidence that argues against this (1)
- No published cohort has analysed KMT2A breakpoint position against outcome in t(4;11) with reciprocal status held constant, so the correlation this hypothesis rests on has never been measured. Agraz-Doblas A et al. 2019
- E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database
UniProt Q03164 annotates recurrent translocation breakpoint sites at residues 1362/1363, 1406/1407 and 1444/1445, with PHD-type zinc finger 1 at 1431-1482 - so breaks within the cluster region fall on either side of, or inside, PHD1.
UniProt Q03164 (KMT2A_HUMAN), feature table · UniProt Consortium · UniProtKB 2026What this evidence cannot show
Fetched 2026-09-19; primary gene name read back as KMT2A, length 3969. The annotated sites are named for other partners, so they bound the region rather than locate the AFF1 join.
Evidence that argues against this (1)
- Almost none of the breakpoint cluster region has been structurally observed, so the functional consequence of a break at 1444 versus 1406 is inferred from domain boundaries rather than measured. PDBe 2026
- E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Read in full text (PMC10169636): in ALL patients the infant group shows a clear preference for KMT2A intron 11 fusions, a preference that switches at about 6 months to intron 9, while AML patients preferentially break in intron 9; and breakpoints upstream of exon 11 retain PHD domain I structurally intact for the reciprocal fusion protein, whereas breakpoints within exon 11 or downstream result in different folding of PHD domain I with impaired CYP33 binding and impaired homodimerisation capacity of that domain.
The KMT2A recombinome of acute leukemias in 2023. · Meyer C, Larghero P, Almeida Lopes B, et al. · Leukemia 2023What this evidence cannot show
Read in full text via Europe PMC fullTextXML (86,330 characters). The breakpoint-by-age analysis is the recombinome's own data on 3,401 referred patients; the PHD-folding consequence is carried by that paper from two cited references rather than measured in it, and the paper states explicitly that whether these findings affect clinical outcome is yet unclear.
Evidence that argues against this (1)
- The paper itself says the clinical consequence is unknown, and no cohort has tested breakpoint position against outcome with reciprocal status held constant, so the confounding this hypothesis proposes is mechanistically plausible and empirically untested. Agraz-Doblas A et al. 2019
- E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Reciprocal transcript expression predicted 4-year EFS of 62.4% versus 11.7% and OS of 73.7% versus 25.2% in 43 t(4;11) infants, in a Cox model adjusted for age, white cell count and prednisone response.
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
Breakpoint position is not among the adjustment covariates and was not measured in that analysis, so this hypothesis's confound is simply unexamined rather than excluded.
Evidence that argues against this (1)
- HOXA cluster expression is perfectly collinear with reciprocal status in the same cohort, so there are at least two unadjusted candidate mediators, not one. Agraz-Doblas A et al. 2019
- E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
siRNA knockdown of AF4-MLL in RS4;11 and SEMK2-M1 had no effect on growth, cell cycle or apoptosis, while knockdown of MLL-AF4 inhibited growth in both and caused G2/M arrest and apoptosis in SEMK2-M1.
t(4;11) leukemias display addiction to MLL-AF4 but not to AF4-MLL. · Kumar AR, Yao Q, Li Q, et al. · Leuk Res 2011What this evidence cannot show
Two lines, transient knockdown, in vitro; speaks to maintenance only, and knockdown of the reciprocal transcript cannot be confirmed at the protein level because no discriminating reagent exists.
Evidence that argues against this (1)
- The initiation experiment gives the opposite answer: AF4.MLL alone induced leukaemia in transplanted murine progenitors where MLL.AF4 alone did not. Bursen A et al. 2010
- related to →KMT2A::AFF1 (MLL-AF4), t(4;11)(q21;q23)
Proposes that the direct product's composition, set by the same break, is the causal variable.
Offers a mechanism by which the reported prognostic effect is confounded rather than causal.
Rests on the complementarity of the two products across the break.
That reanalysis would have to record breakpoint intron as well, or it cannot separate these models.
Holding the partner constant removes the confound this hypothesis depends on.