Hypothesis: the KMT2A break can be prevented rather than the fusion treated - stated so it can be killed, and it probably should be
also called upstream prevention of KMT2A breakage · block the break not the fusion · APE1 prophylaxis hypothesis
If we knew what cuts KMT2A, could we stop the cut instead of treating the leukaemia it causes? The honest answer is that we know at least one molecule whose blockade protects the right human cells, and that almost everything else about the idea is wrong: the cut happens before birth, it happens in healthy people all the time, and the step that is actually rare is not the cut.
The same thing, for a scientist
Two entries in this graph point here and neither states the hypothesis, so this entry states it and then argues against it. **The claim:** a pharmacological intervention that reduces the rate of DNA cleavage at the KMT2A breakpoint cluster region, applied during the window in which the founding translocation occurs, would reduce the incidence of KMT2A-rearranged infant leukaemia. **The one real asset:** inhibition of the base-excision-repair endonuclease APE1 protected human cord-blood-derived haematopoietic stem/progenitor cells from MLLbcr cleavage [E1] - a pharmacologically tractable node whose blockade demonstrably reduces cleavage at the right locus in nearly the right cell type. **Prior: 0.05, and most of that is the chance that the idea yields a mechanistic probe rather than a prophylactic.** Four arguments push it down. (1) **The window.** The fusion is present in the patient's own neonatal blood spot, so any preventive agent must be given to a pregnant woman [E2]. (2) **The base rate.** KMT2A rearrangements were detectable by inverse PCR in the peripheral lymphocytes of 49% of 100 healthy adults, with no t(4;11) transcript expressed in any of them [E3], and rearrangement bands appear in cytokine-stimulated normal CD34+ cells with no genotoxin at all, more at day 7 than in etoposide-treated cells, without producing a leukaemogenic oncogene or an expanding clone [E4]. Breaking KMT2A is common; leukaemia is not. (3) **The arithmetic.** At roughly 41 infant leukaemias per million infants per year with 70-80% of the ALL and about 50% of the AML KMT2A-rearranged [E5], KMT2A-rearranged infant leukaemia runs at roughly 26 per million infant-years - about one in 38,000 pregnancies would have to be exposed to prevent one case at 100% efficacy. (4) **The targets are all essential.** APE1 is the central abasic-site endonuclease; c-NHEJ is required for V(D)J recombination; TDP2 loss increases translocation and causes a paediatric neurodegenerative syndrome. There is no node on this pathway that can be occupied in a fetus.
What would have to be true1 deeper
Stated as conditions so that each can be checked and each can fail:
- A dominant mechanism exists. One cleavage route accounts for most de novo infant KMT2A breaks. Status: unknown. At least four routes are on record - TOP2 poisoning, apoptotic nuclease, replication stress via AID/BER/EndoG, and transcription-associated fragility - and this graph carries entries for all four without a way to apportion them.
- That route has a node that can be blocked without unacceptable harm. Status: no candidate. APE1 is the best-evidenced and is essential.
- Blocking it reduces FUSION formation, not just cleavage signal. Status: never tested. The APE1 result is a cleavage assay 1;
exp-human-cd34-kmt2a-breakage-assayis the system in which the rearrangement-frequency version could be run. - The break, not a later step, is rate-limiting for the disease. Status: evidence points the other way 34.
- The at-risk population can be identified before the event. Status: no. There is no published denominator-based estimate of KMT2A fusion prevalence in newborns, and no known germline risk factor of sufficient effect size.
- The exposure window is reachable. Status: only in utero 2.
Condition 4 is the one that matters most and is least often stated. If the break is common and the leukaemia is rare, an intervention on the break rate is acting on the abundant step of the causal chain, where the achievable absolute risk reduction is bounded by whatever fraction of the rarity is attributable to break rate at all.
The arithmetic, shown so it can be disputed
From 5: about 41 infant acute leukaemias per million infants per year, roughly 160 US cases per year split about 90 ALL and 70 AML, with KMT2A rearrangement in 70-80% of infant ALL and about 50% of infant AML. Taking 75% and 50%: 41 x [ (90/160) x 0.75 + (70/160) x 0.50 ] = 41 x [0.422 + 0.219] = about 26 KMT2A-rearranged infant leukaemias per million infants per year, i.e. roughly 1 in 38,000.
This is the entry's own arithmetic from 5's figures, not a published number. It means a prophylactic given to all pregnancies would expose about 38,000 fetuses per case prevented, at perfect efficacy and zero fetal toxicity - neither of which any candidate on this pathway can offer. Restricting to a high-risk subgroup would change the number, and no way to define such a subgroup exists.
The two interventions that have been proposed, and why both are wrong-way
Poltheta inhibition. Selective allosteric inhibitors of the POLQ polymerase domain exist and are clinically advancing 6. They are aimed at alternative end joining, and human translocations are made by classical NHEJ, so this targets the pathway that is not doing the joining. Worse, POLQ SUPPRESSES genome rearrangement: Polq-defective mice show a marked increase in Myc/IgH translocations, and the authors state explicitly that this departs from the prevailing view that alt-EJ is generically translocation-prone 7. A POLQ inhibitor is a hazard here, not a prophylaxis. This graph's mechanism-nhej-vs-alt-ej-at-fusion-junctions reached the same conclusion independently.
TDP2 inhibition. The intuition that removing the enzyme that finishes a TOP2-induced break would prevent the fusion is backwards: TDP2 loss INCREASES etoposide-induced chromosome rearrangements and translocations including breakage at KMT2A 8, and biallelic human TDP2 loss causes a paediatric-onset spinocerebellar ataxia with drug-resistant epilepsy. See mechanism-tdp2-repair-of-trapped-top2-complexes.
Both failures share a shape worth naming: inhibiting a repair pathway does not remove a break, it changes which repair pathway handles it, and in a fetal progenitor the alternative is usually worse.
Where the hypothesis survives, narrowly
Two reframings keep something alive.
As a probe, not a prophylactic. If APE1 inhibition reduces measured KMT2A REARRANGEMENT frequency (not just cleavage) in human fetal liver HSC, that result apportions the de novo mechanism - it would be the first quantitative statement about how much of endogenous KMT2A breakage runs through the AID/BER/EndoG route. That is worth doing for mechanism even though it will never be given to a pregnant woman.
As a therapy-related-leukaemia question, where the window is reachable. The one population in which the break demonstrably happens under medical supervision, at a known time, with a known cause, is children and adults receiving epipodophyllotoxins. There the window is not in utero, the exposure is scheduled, and the counterfactual is measurable. A TOP2B-sparing topoisomerase poison, or a scheduling change, is a real target - this graph's mechanism-top2-cleavage-at-kmt2a already asks whether a TOP2B-sparing poison can be designed. The hypothesis as written is for infant disease and fails; the same hypothesis restricted to therapy-related disease is not obviously dead.
| statement | A pharmacological intervention that reduces the rate of DNA cleavage at the KMT2A breakpoint cluster region, delivered during the window in which the founding translocation occurs, would reduce the incidence of KMT2A-rearranged infant leukaemia. |
|---|---|
| origin | literature-gap |
| prior probability | 0.05 |
| prior reasoning | The mechanistic asset is real (APE1 protection of human cord-blood HSPC) but every other condition fails: the window is prenatal, the break is common in healthy people, the rate-limiting step is downstream of the break, the targets are essential, and the number needed to expose is about 38,000 pregnancies per case prevented on this entry's own arithmetic from E5. |
conditions that must hold
[ "one cleavage route dominates de novo infant KMT2A breakage - UNKNOWN, four routes on record", "that route has a blockable node with acceptable fetal safety - NO CANDIDATE, APE1 is essential", "blockade reduces fusion formation not just cleavage signal - NEVER TESTED", "the break is rate-limiting for the disease - EVIDENCE AGAINST (E3, E4)", "an at-risk population can be identified prospectively - NO", "the exposure window is reachable - ONLY IN UTERO (E2)" ]
arithmetic
{
"infant_acute_leukaemia_per_million_infant_years": 41,
"split": "about 90 ALL and 70 AML of about 160 US cases per year",
"kmt2a_r_fraction_all": "70-80%",
"kmt2a_r_fraction_aml": "about 50%",
"derived_kmt2a_r_infant_leukaemia_per_million_infant_years": 26,
"derived_pregnancies_exposed_per_case_prevented_at_100pct_efficacy": 38000,
"note": "derived by this entry from E5's figures using 75% and 50%; not a published number"
}interventions assessed and rejected
[
{
"agent_class": "POLQ inhibitor (ART558-class)",
"verdict": "wrong pathway and wrong direction; POLQ loss increases translocation",
"refs": [
"E6",
"E7"
]
},
{
"agent_class": "TDP2 inhibitor",
"verdict": "wrong direction; TDP2 loss increases KMT2A breakage and translocation, and germline loss causes SCAR23",
"refs": [
"E8"
]
},
{
"agent_class": "c-NHEJ inhibitor",
"verdict": "would cause severe combined immunodeficiency; c-NHEJ is required for V(D)J recombination",
"refs": []
},
{
"agent_class": "APE1 inhibitor",
"verdict": "best mechanistic evidence, unusable as prophylaxis; essential BER node, mutagenic and myelosuppressive",
"refs": [
"E1"
]
},
{
"agent_class": "dietary/environmental TOP2-inhibitor avoidance",
"verdict": "epidemiology does not support it; recorded as NOT FEASIBLE in target-prevention-exposure-avoidance-infant-leukemia",
"refs": [
"E5"
]
}
]surviving reframings
[ "APE1 inhibition as a MECHANISTIC PROBE to apportion how much endogenous KMT2A breakage runs through the AID/BER/EndoG route", "the same hypothesis restricted to THERAPY-RELATED disease, where the exposure is scheduled, supervised and counterfactually measurable" ]
experiments
[ "exp-human-cd34-kmt2a-breakage-assay", "exp-hic-kmt2a-aff1-proximity-fetal-liver" ]
analyses
[ "ana-junction-microhomology-signature-mining", "ana-repeat-feature-enrichment-kmt2a-breakpoints" ]
falsifiers
[ "APE1 inhibition (or any candidate) fails to reduce KMT2A REARRANGEMENT frequency in human fetal liver HSC even though it reduces cleavage signal - the protective result would then be an artefact of the readout", "the junction-signature analysis shows de novo infant junctions carry the same repair signature as therapy-related ones, implying no distinct de novo mechanism exists to target", "a denominator-based screen shows KMT2A fusion-positive newborns are common and nearly all healthy, confirming the break is not rate-limiting and capping the achievable benefit at near zero", "no candidate node is identified whose partial inhibition is tolerated in a developing haematopoietic system - the absence of such a node after a systematic search would close the hypothesis on feasibility rather than on biology" ]
- Every candidate node on this pathway is essential. APE1 is the central abasic-site endonuclease, c-NHEJ is required for V(D)J recombination and its loss causes SCID, TDP2 loss causes a paediatric neurodegenerative syndrome, and cohesin and CTCF build the regulatory architecture of every cell.
- Inhibiting a repair pathway does not remove a break; it hands the break to a different pathway. Both agents ever proposed for this purpose - POLQ and TDP2 inhibitors - would be expected to INCREASE translocation.
- The intervention would have to be given to a pregnant woman with no identifiable risk factor, to prevent a disease affecting roughly 1 in 38,000 infants. Any fetal toxicity at all inverts the risk-benefit, and no agent on this pathway is free of it.
- KMT2A breaks are detectable in half of healthy adults and arise in normal progenitors under nothing more than cytokine stimulation. Halving the break rate may halve nothing that matters, because the rare step is not the break.
- Publicising a break-prevention mechanism creates pressure for unvalidated maternal avoidance advice, which this graph has already judged not feasible on the epidemiology. The reputational and public-health cost of an over-claimed prevention story in this disease is real.
- Does APE1 inhibition reduce actual KMT2A rearrangement frequency, measured by inverse PCR, in human FETAL LIVER HSC? This is the single experiment that would move the prior, and it has never been run in the right cell with the right readout.
- What fraction of KMT2A breakage in an unperturbed human fetal progenitor is attributable to each of the four candidate routes? Without an apportionment there is no target to choose, and no study has attempted one.
- Is there a denominator-based estimate of KMT2A fusion prevalence in newborns? There is not, and the graph's `epi-fusion-positive-newborns-versus-leukemia-risk` argues the ETV6::RUNX1 answer may not transfer.
- In therapy-related disease, where the window IS reachable, can a TOP2B-sparing regimen or schedule be designed that preserves anti-tumour efficacy? This is the version of the hypothesis that is not dead and nobody in this graph is working on it.
- E1moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
Inhibition of the base-excision-repair factor APE1 protected human cord-blood-derived haematopoietic stem/progenitor cells, as well as tumour cells, from cleavage at the KMT2A breakpoint cluster region under replication stress.
Base excision repair proteins couple activation-induced cytidine deaminase and endonuclease G during replication stress-induced MLL destabilization · Gole B, Mian E, Rall M, Wiesmuller L · Leukemia 2018What this evidence cannot show
Single laboratory, never independently replicated. The readout is a cleavage assay, not fusion formation. Cord blood is one developmental step past the fetal liver compartment this disease arises in. Aphidicolin-induced replication stress at experimental doses. Read from the abstract.
Evidence that argues against this (1)
- APE1 is the central abasic-site endonuclease of base-excision repair and is essential; chronic inhibition would be mutagenic and myelosuppressive, so even a confirmed protective effect does not yield a usable prophylactic. The graph's replication-stress entry states this in its own bite-back, and the competing account of the endogenous break places it at a transcription initiation region rather than at a deaminase-dependent lesion, with etoposide-induced breaks across the BCR colocalising with the RNA polymerase II binding site of a gene-internal MLL promoter. Scharf S et al. 2007
- E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
Patient-specific clonotypic KMT2A::AFF1 genomic breakpoint junctions are recoverable from the patient's own archived neonatal blood spot, establishing that the translocation is already present at birth in children diagnosed between five months and two years of age.
Backtracking leukemia to birth: identification of clonotypic gene fusion sequences in neonatal blood spots · Gale KB, Ford AM, Repp R, et al. · Proceedings of the National Academy of Sciences of the United States of America 1997What this evidence cannot show
A small backtracking series with a method that can only detect the clone it was designed around. It establishes that the fusion is present at birth in the cases tested; it does not date the event within gestation, so the OPENING of the window is unknown even though its closing is not.
Evidence that argues against this (1)
- Prenatal origin does not by itself close the postnatal window for every case: the same backtracking method demonstrated protracted postnatal natural histories in childhood leukaemia, with clonotypic fusion sequences recovered from the birth spots of children diagnosed many years later. For KMT2A the interval is short, which is the contrast that matters, but a blanket 'there is no postnatal window' overstates the general result. Maia AT et al. 2004
- E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
Inverse PCR screening of peripheral lymphocytes from 100 healthy individuals detected KMT2A rearrangements in 49%, with sequence analysis assigning 66% of them to t(4;11)(q21;q23) and 20% to t(9;11), while RT-PCR failed to detect any t(4;11) expression in the population.
MLL leukemia-associated rearrangements in peripheral blood lymphocytes from healthy individuals · Brassesco MS, Montaldi AP, Gras DE, et al. · Genetics and molecular biology 2009What this evidence cannot show
Inverse PCR on lymphocytes is prone to artefact and a 49% positive rate is high enough to warrant that concern; the paper itself frames the assay as a biomonitoring tool for genotoxic exposure. Sequence assignment of a rearrangement to 'corresponding to t(4;11)' is not the same as recovering a genuine reciprocal fusion. The negative RT-PCR is the load-bearing half and it is a clean one: whatever was detected, it was not producing fusion transcript.
Evidence that argues against this (1)
- An assay's positivity rate at this level invites the opposite reading - that much of it is technical. The comparator literature shows how sharply such estimates fall under stricter confirmation: screening 1,417 cord bloods for a different leukaemia fusion with highly sensitive real-time RT-PCR, and flow-sorting and re-testing every initial positive, produced NO confirmed positives and the conclusion that prevalence is markedly lower than previously suggested. A 49% rearrangement rate that yields zero transcript is consistent with a common harmless event and equally consistent with a noisy assay. Lausten-Thomsen U et al. 2011
- E4moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
KMT2A and RUNX1 rearrangement bands appear by inverse PCR in cytokine-cultured normal CD34+ cells with no etoposide - more at day 7 than in etoposide-treated cells - and in cytokine-mobilised peripheral blood stem cells, yet none of the rearranged genes encoded a leukaemogenic oncogene and the cells carrying them did not expand.
Gene rearrangements of MLL and RUNX1 sporadically occur in normal CD34+ cells under cytokine stimulation · Harada Y, Shingai N, Ding Y, et al. · Cancer science 2020What this evidence cannot show
Adult CD34+ cells and donor mobilisation products, not fetal liver. Inverse PCR band counts are a coarse measure. The treated and untreated arms are compared across timepoints. 'None encoded a leukaemogenic oncogene' is bounded by what the assay recovers.
Evidence that argues against this (1)
- A high background of non-oncogenic rearrangement does not make the break rate irrelevant, because exposure demonstrably shifts the spectrum: in the same cell type etoposide promotes rearrangements matching the full range of oncogenic events seen in patients, and those rearrangements persist through 10-14 days of clonal expansion. The prevention hypothesis is weakened by the background, not eliminated by it. Libura J et al. 2005
- E5moderate confidence 3 of 5: One strong study, or several smaller consistent ones.review
Infant acute leukaemia occurs at about 41 cases per million infants per year in the United States, roughly 160 cases annually split about 90 ALL and 70 AML, with KMT2A rearrangement in 70-80% of infant ALL and about 50% of infant AML.
Treatment of infant leukemias: challenge and promise · Brown P · Hematology. American Society of Hematology. Education Program 2013What this evidence cannot show
A review, used as a map for incidence figures rather than as a primary source. US figures from 2013; the denominator arithmetic in this entry is the entry's own and is sensitive to the 70-80% and about-50% ranges. A registry-based incidence citation would be preferable and this graph's `epi-infant-leukemia-incidence` carries the same figure.
Evidence that argues against this (1)
- Incidence figures constrain the achievable absolute benefit and say nothing about the causal fraction attributable to a modifiable break rate. This graph's own prevention entry records the verdict on the only modifiable exposure ever proposed as NOT FEASIBLE, with the largest case-control test finding little evidence of association and a protective association with maternal fruit and vegetable intake - so the numerator of any prevention calculation is not just small, it has no demonstrated modifiable component. Spector LG et al. 2005
- E6strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Potent selective allosteric inhibitors of the POLQ polymerase domain exist, block theta-mediated end joining without inhibiting NHEJ, and elicit BRCA-gene synthetic lethality.
Poltheta inhibitors elicit BRCA-gene synthetic lethality and target PARP inhibitor resistance · Zatreanu D, Robinson HMR, Alkhatib O, et al. · Nature communications 2021What this evidence cannot show
Developed as a synthetic-lethal oncology agent in homologous-recombination-deficient tumours. There is no published evidence that POLQ inhibition reduces chromosomal translocation frequency in human cells, and citing it as an anti-translocation strategy would be an unsupported extrapolation.
Evidence that argues against this (1)
- POLQ suppresses rather than causes genome rearrangement at DSBs: Polq-defective mice show a MARKED INCREASE in Myc/IgH translocations, and the authors state their findings depart from the prevailing view that alternative end joining is generically translocation-prone. A POLQ inhibitor used prophylactically would be expected to increase, not decrease, translocation. Yousefzadeh MJ et al. 2014
- E7strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
POLQ suppresses chromosomal instability: Polq-defective mice show a marked increase in Myc/IgH translocations, and the authors state explicitly that this departs from the prevailing view that alternative end joining is generically translocation-prone.
Mechanism of suppression of chromosomal instability by DNA polymerase POLQ · Yousefzadeh MJ, Wyatt DW, Takata K, et al. · PLoS genetics 2014What this evidence cannot show
Mouse genetics at the Myc/IgH locus pair, not at KMT2A. The direction of the effect is the transferable part; the magnitude is not.
Evidence that argues against this (1)
- The translocation-suppressing role of an end-joining factor is not general and depends on the break chemistry: designer-nuclease-induced translocations in human cells are GENERATED by classical NHEJ, with LIG4 or XRCC4 deficiency altering rather than abolishing them. Whether a given repair factor protects or produces a fusion is therefore a property of the lesion, and at KMT2A the lesion type is not settled. Ghezraoui H et al. 2014
- E8strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
TDP2 protects the KMT2A locus from TOP2-induced breakage and suppresses etoposide-induced chromosome rearrangements and translocations, so loss of TDP2 increases rather than decreases them.
TDP2 suppresses chromosomal translocations induced by DNA topoisomerase II during gene transcription · Gomez-Herreros F, Zagnoli-Vieira G, Ntai I, et al. · Nature communications 2017What this evidence cannot show
Full text read via Europe PMC fullTextXML (PMC5550487). hTERT-RPE-1 and MCF7 cells under etoposide; not haematopoietic and not drug-free. Breakage at KMT2A is scored by FISH probe separation.
Evidence that argues against this (1)
- The same enzyme is required for the CONVERSION of a trapped TOP2 cleavage complex into an irreversible break, so in a different readout TDP2 activity is a prerequisite for the lesion rather than a protection against it. The two statements are compatible only if 'break' and 'translocation' are kept apart, and an intervention argument that conflates them will pick whichever sign it wants. Canela A et al. 2019
the APE1 protection result is the hypothesis's only real mechanistic asset
independently reaches the same verdict that POLQ inhibition is the wrong pathway and the wrong direction
- depends on →The prevention window: the break happens in a fetus, and we know when it CLOSES but not when it opens
the scheduling constraint that bounds every version of this hypothesis
the other prevention route, already judged not feasible on the epidemiology
the system in which the rearrangement-frequency version of the APE1 result could be run
TDP2 inhibition, the obvious repair-blocking move, increases KMT2A breakage and translocation
the missing denominator: no KMT2A fusion prevalence estimate in newborns exists
about 26 KMT2A-rearranged infant leukaemias per million infant-years, roughly 1 in 38,000
a prevention target whose blockade runs the wrong way - TDP2 loss increases rearrangement
identifies an enzyme without proposing a prevention strategy; that entry sets out why the class of argument is probably wrong