Infants whose leukaemia expresses the reciprocal fusion survive far better - 62.4% versus 11.7% event-free at four years
also called AF4-MLL prognostic · reciprocal expression and outcome · 62.4 versus 11.7
Babies whose leukaemia makes the mirror-image fusion protein do dramatically better than those whose leukaemia does not - more than five times as many are still event-free after four years. That is the opposite of what you would expect if the mirror-image protein were a second cancer driver.
The same thing, for a scientist
In 124 de novo infant B-ALL cases diagnosed and treated uniformly on Interfant 99/06, the reciprocal AF4-MLL transcript was expressed in 45% (19/43) of t(4;11) cases. Reciprocal-expressing patients had 4-year event-free survival of 62.4% versus 11.7% (P = 0.001) and overall survival of 73.7% versus 25.2% (P = 0.016), with the reciprocal transcript retaining prognostic significance in a Cox model adjusted for Interfant-06 risk stratification on age, white cell count and prednisone response [E1]. In the same cohort HOXA cluster genes were expressed exclusively in the reciprocal-expressing patients [E1], so the reciprocal split and the HOXA split are the same split in this dataset and cannot be separated by it. This result cuts against the simplest version of the second-oncoprotein hypothesis. It does not settle it, because 'better survival' is an outcome under a specific chemotherapy protocol, not a statement about the biology of the protein, and the direction is what a differentiating or dose-limiting second protein would also produce.
Reading it against the hypothesis this beat was sent to test
The brief for this beat treats the reciprocal product as the most refutation-relevant object in the graph, on the argument that it is a second KMT2A-derived oncoprotein a menin inhibitor cannot engage. The prognostic data point the other way: the patients who have it do better, not worse. Four readings survive contact with the numbers:
- The reciprocal product is not oncogenic in established disease and its presence marks a less evolved, more chemosensitive clone. Consistent with 2, where removing it from t(4;11) cell lines did nothing.
- The reciprocal product constrains the direct one. One 2026 report has KMT2A::AFF1 driving high expression of a few targets and AFF1::KMT2A counteracting that effect 3.
- It is a marker, not a cause - of breakpoint class, of HOXA status, of clonal age.
- The effect is real, causal and protective, and would be abolished by a drug that removed it. Nobody has proposed removing it, but a Taspase1 inhibitor would.
None of these is excluded. What is excluded is the naive form of 'a second oncoprotein the drug cannot reach, therefore worse outcomes'.
Why the number is weaker than it reads
n = 43 for the t(4;11) comparison, 19 versus 24. A 50-point EFS difference on those numbers has wide confidence limits that the abstract does not give. The Cox adjustment covers age, white cell count and prednisone response and does not cover breakpoint position or HOXA status - the two variables most likely to be the real signal 1. And the exposure is a diagnostic RT-PCR whose negative arm demonstrably contains patients with transcribed reciprocal alleles fused to other partners 4.
The unresolved question that costs the most
Does the direction reverse under a menin inhibitor? Every patient in this analysis was treated with conventional chemotherapy. If the reciprocal product is menin-blind and menin inhibition is the pressure, then reciprocal-positive disease has a compartment the drug cannot touch, and the favourable chemotherapy prognosis says nothing about that. No menin-inhibitor cohort has been stratified this way 5.
| metric | 4-year event-free survival by reciprocal transcript status |
|---|---|
| efs reciprocal positive PCT | 62.4 |
| efs reciprocal negative PCT | 11.7 |
| efs p | 0.001 |
| os reciprocal positive PCT | 73.7 |
| os reciprocal negative PCT | 25.2 |
| os p | 0.016 |
| n t411 | 43 |
| n reciprocal positive | 19 |
| n cohort total | 124 |
| population | de novo infant B-ALL, Interfant 99/06 |
| region | international (Interfant) |
| year | 2019 |
| source | PMID 30679323 |
| treatment era | conventional chemotherapy; no menin inhibitor |
adjusted for
[ "age at diagnosis", "white blood cell count", "prednisone response" ]
not adjusted for
[ "breakpoint position", "HOXA cluster expression", "reciprocal protein level" ]
- If the reciprocal product is protective, any agent that destroys it - a Taspase1 inhibitor, a FYRC decoy - would be actively harmful, and those are exactly the agents the mechanistic literature proposes.
- If reciprocal status is instead a marker of a chemosensitive clone, then using it to select patients for a targeted agent would select the patients who least need one.
- The prognostic estimate is from the chemotherapy era. A menin inhibitor changes the selection pressure in the one direction the reciprocal product is immune to, so the sign of this association could invert under the new standard of care and no protocol is collecting the data that would show it.
- Does the favourable prognosis of reciprocal-positive t(4;11) survive a menin inhibitor? Every number here is from conventional chemotherapy and the reciprocal product is menin-blind by construction.
- Can the reciprocal effect be separated from the HOXA effect in any cohort? In this one they are the same patients, so the dataset cannot do it.
- What are the confidence intervals on 62.4% versus 11.7% at n = 19 versus 24? The abstract gives P values only.
- Is there a second, independent cohort? This beat found none: the partner-stratified paediatric cohort that exists did not measure reciprocal status.
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Among 124 de novo infant B-ALL patients on Interfant 99/06, the reciprocal AF4-MLL transcript was expressed in 45% (19/43) of t(4;11) cases; those patients had significantly better 4-year event-free survival (62.4% vs 11.7%, P = 0.001) and overall survival (73.7% vs 25.2%, P = 0.016), with significance retained in a Cox model adjusting for Interfant-06 risk stratification, and HOXA cluster genes were expressed exclusively in the AF4-MLL-expressing patients.
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
n = 43 for the reciprocal comparison; confidence intervals are not given in the abstract; the Cox model adjusts for age, white cell count and prednisone response only, so breakpoint position and HOXA status are unadjusted and HOXA is perfectly collinear with the exposure in this dataset. All patients received conventional chemotherapy, so the estimate does not transfer to a menin-inhibitor era.
Evidence that argues against this (1)
- A separate paediatric cohort analysing outcome by KMT2A partner found MLL-AF4 itself an independent adverse factor and did not measure reciprocal status at all, so the partner-level and reciprocal-level prognostic literatures have never been reconciled in one model. Qiu KY et al. 2022
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
siRNA knockdown of AF4-MLL in the human t(4;11) lines RS4;11 and SEMK2-M1 had no effect on growth, and in SEMK2-M1 no effect on cell cycle or apoptosis, whereas knockdown of MLL-AF4 significantly inhibited growth of both and caused G2/M arrest and increased apoptosis.
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 cell lines, transient siRNA, in vitro readouts only; a maintenance experiment that cannot speak to initiation, and knockdown efficiency at the protein level cannot be confirmed because no reagent distinguishes the reciprocal-derived fragment.
Evidence that argues against this (1)
- The same laboratory group whose model the knockdown opposes reports that AF4-MLL alone induces acute leukaemia in transplanted murine progenitors where MLL-AF4 alone does not, so initiation and maintenance give opposite answers. Bursen A et al. 2010
- E3weak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.paper
In engineered model systems KMT2A::AFF1 alone drives aberrantly high expression of only a few target genes while AFF1::KMT2A counteracts that effect, and both reciprocal transcripts are bound by IGF2BP3, which reduces KMT2A::AFF1 protein abundance and increases AFF1::KMT2A levels.
Influence of IGF2BP3 on t(4;11) fusion mRNA and protein abundance. · Hanewald T, Benz T, Marschalek R · Leuk Res 2026What this evidence cannot show
Engineered systems, not primary blasts; a 2026 report without independent replication; the counteracting effect is measured on transcript output rather than on leukaemic phenotype.
Evidence that argues against this (1)
- A study co-expressing both fusions in a conditional cell-culture model found the opposite sign: co-transfected cells showed the highest resistance to apoptotic triggers, the greatest cell-cycling capacity and loss of contact inhibition, with synergy rather than antagonism. Gaussmann A et al. 2007
- E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
t(4;11) patients scored as lacking the reciprocal allele were found on genomic analysis to carry reciprocal MLL fusions to other partners including NFKB1 and RABGAP1L that were actively transcribed, as well as out-of-frame joins and solitary 3'-MLL segments.
Complex MLL rearrangements in t(4;11) leukemia patients with absent AF4.MLL fusion allele. · Kowarz E, Burmeister T, Lo Nigro L, et al. · Leukemia 2007What this evidence cannot show
A selected subgroup, not a consecutive cohort; the proportion of a typical reciprocal-negative arm that is misclassified this way is not quantified anywhere.
Evidence that argues against this (1)
- The cohort that produced the survival estimate scored reciprocal status as a binary RT-PCR result and did not sequence its negative arm, so this misclassification is unmeasured in the population where it matters. Agraz-Doblas A et al. 2019
- E5moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
No published study connects menin inhibition to the reciprocal fusion product. Searched 2026-09-19: '(revumenib OR ziftomenib OR menin inhibitor OR SNDX-5613 OR KO-539) AND (reciprocal OR AF4-MLL)' returns 1 record, which concerns JUND and menin-MLL inhibitor resistance and does not mention reciprocal fusion status; 'menin inhibitor AND AF4-MLL' returns 0; 'AF4-MLL AND menin' returns 0.
Menin-MLL inhibitors enhance JUND activity in MLLr leukemic cells contributing to tumorigenesis and therapy resistance. · Ozyerli-Goknar E, Nizamuddin S, Fetsch V, et al. · Blood 2026What this evidence cannot show
A bounded absence over three query spellings. It is bounded by PubMed indexing and by the alias set used; a trial that stratified by reciprocal status in an unpublished analysis, or a paper using 'der(4)' phrasing that the PubMed parser mishandles, would not appear. The cited record is the single hit and is included so the absence can be audited, not because it supports the claim.
Evidence that argues against this (1)
- The one retrieved record shows that menin-inhibitor resistance mechanisms are being actively mapped - through JUND and the AP-1 axis - which makes the absence of any reciprocal-status analysis a choice of framing rather than a gap nobody is in a position to fill. Ozyerli-Goknar E et al. 2026
the t(4;11) effect size this fusion's 153-versus-101 comparison is powered to look for, and its opposite direction
The clinical outcome associated with expressing this protein, and it runs against the second-oncoprotein reading.
Measured in 124 Interfant 99/06 infants, 43 of them t(4;11).
The simplest form of that hypothesis predicts worse outcomes for reciprocal-positive patients; the observed direction is the opposite and strongly so.
Supplies the effect size and the era caveat that make the proposed reanalysis worth running.
Offers an explanation for the survival split that does not require the reciprocal product to be protective.
Offers a mechanism by which the reported prognostic effect is confounded rather than causal.
The confound that the prognostic estimate cannot adjust for, because it is the same partition.
The partner-constant comparison that could confirm or break the t(4;11) prognostic association.