Oncofusion

start / graph / mechanism-fusion-clonal-stability-diagnosis-to-relapse

Mechanismstrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.Q1b-genesis2

Are the fusions temporary? No: the KMT2A fusion is the one thing that does not change between diagnosis and relapse

also called clonal stability of the fusion · truncal lesion · is the fusion in flux

Almost everything about this leukaemia can change when it comes back after treatment: which cells it looks like, which markers it carries, which extra mutations it has. The fusion itself does not change, which is why it is the right thing to aim a drug at.

The same thing, for a scientist

The PI asked whether the fusions are temporary or in flux. The answer has two halves that point in opposite directions and must be kept apart. **The fusion is truncal and stable.** It is the founding lesion, present in the dominant clone at diagnosis, and it is retained through relapse, through lineage switch, and through immunophenotypic escape from CD19-directed therapy (`mechanism-lineage-switch-kmt2a`, `risk-lineage-switch-cd19-directed-therapy`). **Everything layered on top of it is in flux.** Andersson et al. sequenced 65 infants and 20 older children and found that the activating kinase-PI3K-RAS mutations present in 47% of cases were often subclonal and were *frequently lost at relapse*, against a dominant-clone burden of only 1.3 non-silent mutations [E1]. So the disease that relapses is the same fusion in a differently configured cell, not a different disease. The practical consequence is that the fusion, and by extension the fusion's obligate protein interfaces, is the only durable target; anything selected from the diagnostic co-mutation profile is a target that may not be there when you need it.

Go deeper
What 'in flux' correctly describes1 deeper

Three things genuinely move between diagnosis and relapse, and all three are downstream of the fusion rather than the fusion itself:

  1. Cooperating signalling mutations. RAS-pathway lesions are common at diagnosis but subclonal, and are frequently absent at relapse 1. A RAS inhibitor chosen on a diagnostic sample can therefore be aimed at a clone that will not be the one that returns.
  2. Lineage and immunophenotype. KMT2A-r blasts can switch from lymphoid to myeloid under CD19-directed pressure while keeping the same fusion, which is escape by changing identity rather than by changing driver (see mechanism-lineage-switch-kmt2a).
  3. Subclonal architecture. Intraclonal diversity is present from before birth in concordant twins 2, so the relapse clone may be a pre-existing minor population rather than a newly evolved one.
Why a stable truncal driver is unusual and valuable

In most adult solid tumours the driver landscape itself shifts under therapy. Here it does not, because the fusion cannot be point-mutated away without losing the oncogene, and because there is no second driver waiting to take over: the dominant clone carries a mean of 1.3 non-silent mutations 1. This is precisely the situation in which an interface-directed drug should work, and it is the strongest structural argument for the menin hypothesis that does not depend on any menin-specific data.

Is the fusion ever genuinely lost?

Two distinct phenomena must not be conflated. Loss of the fusion from a relapsing clone is not something the literature this beat could verify; searches for documented fusion-negative relapse of a KMT2A-rearranged leukaemia returned nothing, and the graph records that as an unresolved gap rather than as a negative result. Loss of a fusion-positive clone before disease is documented: Chuk et al. tracked a clonal KMT2A::MLLT1-positive population in a healthy infant co-twin and watched it disappear 3. That is the one clean human observation that a fusion-positive clone is not a one-way door, and it happened before leukaemia, not after it.

How this could bite us back
  • Stability of the fusion is a reason to target it, but it also means the fusion cannot be used as a marker of therapeutic pressure: it will look identical whether or not a menin inhibitor is working, so response must be judged on program output (HOXA9/MEIS1) rather than on the driver.
  • If relapse clones are selected from pre-existing diversity rather than evolved, then intensifying therapy cannot prevent relapse by preventing mutation, and only a drug that kills the diverse population at diagnosis will help.
  • Choosing a co-target from the diagnostic RAS profile risks building a combination against a clone that is frequently gone by relapse [E1].
Still open
  1. Is there a single published, well-documented case of a KMT2A-rearranged leukaemia relapsing WITHOUT the fusion? This beat could not find one. Either the case series exist and are buried in cytogenetics supplements, or the answer is genuinely no, and those two possibilities have very different implications for how confident an interface-directed strategy can be.
  2. Is the relapse clone selected from pre-existing diagnostic subclones or newly generated? Paired diagnosis-relapse single-cell DNA sequencing in infant KMT2A-r ALL would settle it, has not been published, and is squarely within reach of a lab with access to Interfant biobank samples.
  3. Does the genomic breakpoint sequence itself ever change (secondary rearrangement of the derivative chromosome) between diagnosis and relapse? If it never does, a patient-specific genomic-junction MRD assay is the most specific relapse monitor available and is being under-used relative to RT-PCR.
Every source behind this entry (3)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In infant KMT2A-rearranged ALL the dominant clone carries a mean of 1.3 non-silent mutations, and the cooperating kinase-PI3K-RAS mutations found in 47% of cases are often subclonal and frequently lost at relapse.

    The landscape of somatic mutations in infant MLL-rearranged acute lymphoblastic leukemias · Andersson AK, Ma J, Wang J, Chen X, Gedman AL, Dang J, et al. · Nature Genetics 2015
    PMID 25730765DOI
    What this evidence cannot show

    65 infants, of whom 47 were KMT2A-rearranged; the diagnosis-relapse comparison rests on a subset of paired samples, so 'frequently lost' is a qualitative statement from a modest number of pairs. Bulk sequencing at this depth cannot exclude a persisting minor RAS-mutant subclone at relapse.

    Evidence that argues against this (1)
    • Genomic stability of the fusion is not stability of the disease, and in this exact disease the phenotype can switch and switch back within one patient. Wolfl et al. describe KMT2A-rearranged infant ALL that underwent a blinatumomab-induced ALL-to-AML lineage switch and then spontaneously reverted, which the authors present as illustrating the high plasticity of this phenotype under immunological pressure. The truncal fusion was not what changed. Any clinical inference drawn from fusion stability -- MRD design, antigen choice, or the assumption that a relapse is the same disease -- has to be made at the level of the phenotype, which is unstable on a timescale of weeks. Single case report. Read: abstract only. Wolfl M et al. 2018
  2. E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    Intraclonal genetic diversity on a shared fusion-positive stem line is already present before birth in concordant infant twins.

    Prenatal chromosomal diversification of leukemia in monozygotic twins · Kempski H, Mensa-Bonsu KA, Kearney L, Jalali GR, Hann I, Khurshid M, Greaves M · Genes, Chromosomes & Cancer 2003
    PMID 12800152DOI
    What this evidence cannot show

    A single twin pair described by the authors as unusually diverse; cytogenetic resolution only.

    Evidence that argues against this (1)
    • A fusion-bearing clone's state is reprogrammable end to end, which is the strongest available demonstration that clonal genomic stability does not imply a fixed cellular state. Liu et al. converted transplantable murine MLL-AF9 leukaemia cells into iPS cells that formed teratomas and produced chimeras, with RNA-seq showing reversible global expression patterns between the interchangeable leukaemic and pluripotent states on inactivation or reactivation of MLL-AF9; notably most chimeric mice then spontaneously developed the same AML, and both the reprogramming and the leukaemia-reinitiating paths could descend from the same leukaemia-initiating cell. Murine, MLL-AF9, and with exogenous Yamanaka factors, so it is a statement about what is possible rather than about what happens in patients. Read: abstract only. Liu Y et al. 2014
  3. E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    A clonal KMT2A-fusion-positive population in a healthy infant was documented and subsequently cleared without treatment.

    PMID 19411627DOI
    What this evidence cannot show

    Single case; the clearance mechanism is inferred from temporal association with viral illness, not measured.

    Evidence that argues against this (1)
    • The single twin clearance case should not be generalised, in either direction. Greaves and colleagues' review of twin leukaemia records concordant monozygotic infant pairs as the recurrent observation since 1882, explained by transplacental spread of one initiated clone through a shared monochorionic placenta and confirmed by clonotypic genomic fusion sequences establishing common clonal origin; the review's own conclusion is that latency is variable and often protracted and that further postnatal events are needed in childhood ALL/AML. Neither the concordant-pair literature nor the single discordant case has a denominator, so how often a fusion-positive clone clears remains unmeasured. Read: abstract only. Greaves MF et al. 2003
Connected to