Oncofusion

start / graph / hyp-q146-the-kmt2a-aff1-target-promoter-set-is-a-fusion-level-property-and-patient-identity-adds-nothing-measurable-to-it

Hypothesisspeculative confidence 1 of 5: Our own inference or analogy. Not shown to be true.Q146-one-universe-jaccard

The KMT2A::AFF1 oncoprotein-target promoter set is a fusion-level property, and patient identity adds nothing measurable on top of it

Where this cancer protein lands on DNA may be decided by the fusion itself rather than by the individual child. If so, two samples from the same child are no more alike than samples from two different children - which is what the data show.

The same thing, for a scientist

Every KMT2A::AFF1 sample pair in GSE252378 overlaps its structure-respecting null by a similar amount - 1.19-1.62x under the paper's calling rule, permutation p = 0.0005 - and the one pair that is two states of a single patient sits at the 37th percentile of that distribution, inside the central 80%. The hypothesis is that the shared component is entirely fusion-level (the oncoprotein plus its cofactors addressing a characteristic class of active promoters) and that there is no additional patient-level component detectable at this sample size and this assay resolution. The alternative - a real patient-level component of the target set that this design is underpowered to see - is not excluded, and the post-hoc institution-matched analysis leans the other way, toward the same-patient pair being the LEAST similar rather than merely average.

Go deeper
What the hypothesis predicts
  1. A serial pair from the same patient with no lineage change (diagnosis and a same-lineage relapse) should overlap no more than an unrelated same-lineage pair at the same institution. If a same-patient non-switching pair DOES overlap more, the patient-level component exists and the lineage switch destroys it.
  2. Excess over the matched null should not differ between same-fusion and cross-fusion pairs beyond what promoter activity explains. If KMT2A::AFF1 pairs overlap more than AFF1-versus-ENL pairs at matched activity, the shared component is partner-specific, which is a weaker version of the hypothesis.
  3. A cell line carrying KMT2A::AFF1 (SEM) should overlap primary KMT2A::AFF1 patients at the same excess as those patients overlap each other. A line shares no patient with anyone, so if it reaches the same excess, patient identity is contributing nothing.
What it would mean if true

It bears directly on the corpus's menin-interface argument. If the target set is fusion-level and reproducible across children, then the addressing machinery - which includes the menin interface - is doing the same job in every patient, and between-patient differences in oncoprotein occupancy are differences in amount rather than in destination. That is the 'permission not a dial' reading this corpus already holds from a different measurement, and this would be independent support for it.

Clinically it argues AGAINST carrying a patient's diagnostic target list into salvage, and FOR a fusion-level target list - the opposite of the usual personalisation instinct.

Facts and figures
statementThe KMT2A::AFF1 oncoprotein-target promoter set is determined at the level of the fusion and its cofactor complex, and patient identity contributes no additional detectable component to it.
origindata-pattern
prior probability0.35
predictions
[
  "A same-patient, same-lineage serial pair overlaps no more than an institution-matched unrelated same-lineage pair.",
  "SEM (a KMT2A::AFF1 cell line) reaches the same excess-over-matched-null against primary KMT2A::AFF1 patients as those patients reach against each other.",
  "Excess over the matched null does not differ between same-partner and cross-partner pairs once promoter activity is matched."
]
experiments
[
  "exp-q146-a-pre-registered-institution-matched-serial-pair-replication-of-the-target-set-overlap"
]
analyses
[
  "ana-q146-on-one-universe-with-one-rule-the-lineage-switch-pair-is-indistinguishable-from-an-unrelated-infant",
  "ana-q146-the-excess-overlap-survives-a-structure-respecting-null-in-both-beats-but-shrinks-two-to-four-fold",
  "ana-q146-post-hoc-against-institution-matched-anchored-comparators-the-same-patient-pair-is-the-lowest-in-all-six-cells"
]
falsifiers
[
  "A same-patient, same-lineage serial pair (diagnosis and same-lineage relapse, same institution, matched depth) whose excess over the activity-stratified null exceeds the 90th percentile of institution-matched unrelated pairs at n >= 4 pairs. That would establish a patient-level component that the lineage switch destroys.",
  "SEM versus primary KMT2A::AFF1 patients showing excess over the matched null below the 10th percentile of patient-versus-patient pairs, in the U_pair universe under the paper's rule. That would mean a cell line cannot substitute for a patient, i.e. there is a patient- or primary-tissue-level component the fusion does not supply.",
  "ALREADY PARTLY RESOLVED IN THE WRONG DIRECTION FOR THIS HYPOTHESIS: the post-hoc institution-matched anchored comparison places the same-patient pair below ALL six matched comparators in all six factorial cells (sign p = 0.031 in the three absolute-rule cells). 'Adds nothing' predicts average, not lowest. If a pre-registered replication reproduces the lowest-of-all result, this hypothesis is refuted in favour of the lineage switch actively REPLACING the target set."
]
How this could bite us back
  • If the target set is fusion-level, a menin inhibitor's target list is the same in every child - which is good for drug development and bad for predicting which child will respond, because response then cannot be read off the target set at all.
  • The same hypothesis predicts that resistance cannot be anticipated from a patient's own diagnostic target map, since the map at relapse is no more that patient's than anyone else's.
Still open
  1. Is the excess-over-matched-null the same for cross-fusion pairs as for same-fusion pairs? This beat computed the all-pairs distribution (median 1.404 over 319 pairs versus 1.361 over 27 AFF1 pairs) but did not test the difference formally.
  2. SEM libraries exist in this deposit's AFF1 arm but not in the N/C arm. Is there any KMT2A::AFF1 cell line with paired N- and C-terminal AutoCUT&RUN anywhere in the public record?
Every source behind this entry (2)
  1. E1moderate confidence 3 of 5: One strong study, or several smaller consistent ones.dataset

    Across 27 unrelated KMT2A::AFF1 sample pairs the excess over an activity-stratified null is 1.19-1.62x with permutation p = 0.0005, and the one same-patient pair sits at the 37th percentile of that distribution.

    What this evidence cannot show

    Seven patients, one deposit, one assay. The 27 pairs are not independent - they are drawn from 7 patients, and the two same-patient samples both contribute - so the distribution's width is understated.

    Evidence that argues against this (1)
    • The post-hoc institution-matched analysis places the same-patient pair BELOW all six matched comparators in all six factorial cells, which is not 'adds nothing' but 'subtracts something' - a different hypothesis that this entry does not adopt but cannot exclude. Janssens DH et al. 2024
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    The publication's own finding is that oncoprotein genomic enrichment is highly variable between samples and dynamically regulated, and that fusion-partner-specific binding patterns over lineage gene sets correlate with each mutation's prevalence in ALL versus AML - i.e. that the partner, not the patient, is the organising variable.

    MLL oncoprotein levels influence leukemia lineage identities · Janssens DH, Duran M, Otto DJ, Wu W, Xu Y, Kirkey D, Mullighan CG, Yi JS, Meshinchi S, Sarthy JF, Ahmad K, Henikoff S · Nature Communications 15:9341 2024
    PMID 39472576DOI
    What this evidence cannot show

    The paper's partner-level claim is about binding patterns over curated lineage gene sets, not about genome-wide set overlap, so it is consistent with this hypothesis rather than a test of it.

    Evidence that argues against this (1)
    • This corpus has separately measured that oncoprotein occupancy does NOT sort by fusion partner across six partners and 27 samples, which is evidence against a strong partner-determines-targets reading and therefore complicates the fusion-level claim. Janssens DH et al. 2024
Connected to