Oncofusion

start / graph / hyp-lineage-switch-is-competitive-release-of-preexisting-clone

Hypothesismoderate confidence 3 of 5: One strong study, or several smaller consistent ones.Q12-analogy

Hypothesis: the myeloid clone that appears after CD19-directed therapy was already there at diagnosis, and its abundance is a pre-treatment biomarker

When a B-cell leukemia turns into a myeloid one after immunotherapy, the myeloid cells may not be newly converted; they may have been present in small numbers all along and simply grown once their competitors were killed. If so, you could count them before treatment starts.

The same thing, for a scientist

Lineage switch after CD19-directed therapy is concentrated in KMT2A-rearranged disease, occurs at a median of 1.5 months and most often from documented MRD-negative remission of the B-ALL [E1]. We hypothesise that this is competitive release of a fusion-positive, myeloid-primed subpopulation present at diagnosis at a frequency of roughly 10-3 or above, rather than de novo transdifferentiation, and therefore that the abundance of that subpopulation at diagnosis predicts switch risk. Supporting evidence is phylogenetic and indirect: a post-immunotherapy switch AML derived from the initial ALL and carried driver mutations detectable before the AML emerged [E2], MLL/AF4 myeloid relapses share fusion breakpoints with their lymphoid presentations and arise from a range of differentiation stages [E3], and in one infant case three late immunoglobulin clones were already present at low abundance diagnostically [E4]. **CORRECTED 2026-09-23 (cycle 120): the single-cell demonstration this entry said did not exist was already cited five times in this corpus.** Chen et al., *Blood* 2022 (PMID 34864916) applied single-cell multiomics to KMT2A-r ALL and report *preexisting lymphomyeloid primed progenitors and myeloid blasts at initial diagnosis of B-ALL*, then tracked those clones in two patients whose leukaemia underwent lineage switch and documented their evolution into frank AML [E7]. That is this hypothesis's own prediction, measured directly rather than inferred phylogenetically. The CHD4 reprogramming result [E3] remains a serious competing account and n=2 is not a frequency estimate, so the hypothesis moves from speculative to moderate rather than to supported.

Go deeper
What would make this a usable biomarker, and what would not

A pre-treatment frequency that separates switchers from non-switchers is only useful if the mitigation is proportionate. It is. The recorded mitigation for lineage switch is already cheap and harmless: full multi-lineage immunophenotyping rather than B-gated flow at every marrow, a lineage-independent molecular assay (fusion-transcript RT-PCR or clonoSEQ) alongside flow, and heightened vigilance in months 0-6.

What the biomarker must never be used for is withholding CD19-directed therapy. Frontline blinatumomab in 30 KMT2A-rearranged infants gave 4-year disease-free survival of 83.3% versus 44.0% 5. Trading that against a switch risk in the 0-28.6% reported range would be a clear net harm, and any publication of a switch-risk score must say so explicitly in its abstract rather than in a limitations paragraph.

The competing account is strong and should not be minimised

Lineage switching in MLL/AF4 involves substantial chromatin-accessibility change and transcriptional rewiring recurrently associated with altered chromatin modifiers, and perturbing CHD4 alone induces myeloid gene expression in MLL/AF4 models 3. That is reprogramming, and it predicts nothing detectable at diagnosis.

The two accounts are not exclusive. The most likely truth is a bipotential or lineage-ambiguous subpopulation present at diagnosis whose conversion is then enabled by a chromatin-modifier state — pre-existence of the substrate plus induction of the transition. If so, the diagnostic frequency of the substrate would still be predictive, but weakly, and the CHD4-state measurement would add independently. The experiment is designed to measure both and apportion between them, not to pick a winner.

Why the detection threshold matters more than the hypothesis

If the precursor population sits at 10-3, standard single-cell approaches sampling a few thousand cells will miss it entirely, and a negative result will mean nothing. The experiment must therefore be designed around detection limit first: enrichment before profiling, or a targeted assay with the sensitivity of MRD methods rather than the breadth of discovery single-cell.

This is the practical reason the question is still open despite being obvious. Everyone who has looked has looked with an assay whose floor is above the frequency of interest.

Facts and figures
statementLineage switch after CD19-directed therapy in KMT2A-rearranged leukemia is competitive release of a fusion-positive, myeloid-primed or lineage-ambiguous subpopulation present at diagnosis at a frequency at or above 10-3, rather than de novo transdifferentiation; the abundance of that subpopulation in the diagnostic sample predicts subsequent switch and adds to KMT2A genotype alone as a risk marker.
originanalogy
source analogyanalogy-ecology-competitive-release-and-adaptive-therapy
prior probability0.4
prior reasoningSupported by the speed of switch, by phylogenetic evidence of pre-existing diversification in one KMT2A-wild-type case, and by shared breakpoints across a range of differentiation stages in MLL/AF4. Held down substantially by the CHD4 reprogramming result, which supplies a complete alternative mechanism. CORRECTED cycle 120: the clause that previously read 'and by the absence of any single-cell demonstration' was false - Chen 2022 (PMID 34864916) is that demonstration, was already cited five times elsewhere in this corpus, and tracked pre-existing myeloid-primed clones through lineage switch into AML in two patients [E7]. The 0.4 stated below predates that correction and is now too low; the competing CHD4 account and n=2 keep it short of supported. We expect the truth to be a pre-existing substrate plus an induced transition, in which case the diagnostic frequency is predictive but weakly, and 0.4 reflects the hypothesis as stated rather than that softer version.
decision valueWould give the first pre-treatment risk marker for lineage switch beyond genotype, with a mitigation (multi-lineage monitoring) that is already established, cheap and harmless. The sanctioned use is intensified monitoring only, never withholding CD19-directed therapy.
predictions
[
  "Deep single-cell multi-omic profiling of diagnostic samples from patients who later switched detects a fusion-positive myeloid-primed or lineage-ambiguous population at or above 10-3, absent or rarer in matched non-switchers.",
  "Phylogenetic reconstruction places the switch clone branching before the start of immunotherapy in KMT2A-rearranged cases, as it does in the published KMT2A-wild-type case.",
  "The diagnostic frequency of that population adds predictive value to KMT2A genotype alone in a case-control comparison.",
  "The same population is detectable, at lower frequency, in KMT2A-rearranged patients who never switch, making it a graded risk marker rather than a binary one.",
  "CHD4 or chromatin-modifier status at diagnosis adds independently to the subpopulation frequency, consistent with substrate plus induced transition."
]
experiments
[
  "exp-diagnostic-subclone-archaeology-lineage-switch"
]
analyses
[
  "ana-lineage-switch-signature-scoring"
]
falsifiers
[
  "No fusion-positive myeloid-primed population is detectable at diagnosis in switchers, at an assay sensitivity demonstrably below 10-3.",
  "Phylogenetic reconstruction shows the switch clone acquired its defining somatic changes after therapy began, in KMT2A-rearranged cases.",
  "The population is present at equal frequency in switchers and non-switchers, making it real but non-predictive.",
  "Switch is fully accounted for by CHD4 or another chromatin-modifier state with no residual contribution from diagnostic subpopulation frequency."
]
How this could bite us back
  • If the precursor pre-exists, deeper CD19-directed responses cause more switch, and MRD-negativity becomes a hazard signal rather than a success measure — which inverts how these trials are monitored.
  • A risk score for switch could be used to withhold a therapy that raised 4-year disease-free survival from 44.0% to 83.3% in exactly this population. The harm from misuse of this biomarker is larger than the harm it prevents, and that has to be stated wherever the score appears.
  • Multi-antigen targeting to close the escape route brings myeloid antigens expressed on normal progenitors into the crosshairs, risking prolonged marrow aplasia in an infant.
Still open
  1. What assay reaches 10-4 sensitivity for a transcriptional state rather than for a sequence? Molecular MRD methods are sensitive but sequence-based; single-cell methods read state but sample shallowly.
  2. Does the precursor population exist in patients who never receive CD19-directed therapy, and does it expand under menin inhibition too?
  3. Is 'myeloid-primed' the right description, or is the precursor lineage-ambiguous in the way MPAL is, which would connect this to the existing MPAL mechanism entry?
Every source behind this entry (7)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In a 70-case cohort of post-immunotherapy lineage switch, 64.3% carried a KMT2A rearrangement, median time to switch was 1.5 months (1 month after blinatumomab, 2.8 months after CAR T cells), 81.4% occurred within six months, and 54.3% of patients had achieved MRD-negative complete remission of their B-ALL beforehand.

    PMID 40193715DOI
    What this evidence cannot show

    Cases without a denominator; the speed of switch is consistent with expansion of a pre-existing population but is not proof of it.

    Evidence that argues against this (1)
    • A single well-documented case argues against selection of a pre-existing clone and for reversible transdifferentiation of the same cell. In KMT2A-rearranged infant ALL, a blinatumomab-induced ALL-to-AML lineage switch SPONTANEOUSLY REVERTED to the lymphoid phenotype, which a competitive-release model cannot easily produce: once a distinct myeloid subclone has outgrown its competitors there is nothing to revert to. The authors describe a leukaemia rapidly adapting to immunological stimulus. If the switch is a state transition rather than a selection event, counting myeloid cells at diagnosis will not be a biomarker, because the cells that switch need not be phenotypically myeloid beforehand. Wölfl M et al. 2018
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Whole-genome phylogenetic reconstruction in a child developing AML after anti-CD19 cellular and antibody therapy showed the AML derived from the initial ALL and carried distinct driver mutations detectable before the AML emerged, interpreted as extensive prior clonal diversification followed by selection.

    PMID 37474833DOI
    What this evidence cannot show

    A single case and explicitly KMT2A wild-type. It supports pre-existing diversification in principle and not in the genotype this hypothesis is about.

    Evidence that argues against this (1)
    • A single child, and the reconstruction's strength — distinct driver mutations detectable before the AML emerged — is also its limit: detectable at what sensitivity, in which compartment, and with what chance of the same being true in patients who never switched. Pre-existing subclones are ubiquitous; most do not take over. Journal of Clinical Oncology 2025
  3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Myeloid relapses after lineage switch in MLL/AF4 leukemia shared oncogene fusion breakpoints with matched lymphoid presentations and originated from a range of differentiation stages; switching involved chromatin-accessibility change and transcriptional rewiring recurrently associated with altered expression, splicing or mutation of chromatin modifiers notably CHD4, and perturbing CHD4 alone induced myeloid gene expression in MLL/AF4 models.

    Epigenetic regulator genes direct lineage switching in MLL/AF4 leukemia · Tirtakusuma R, Szoltysek K, Milne P, et al. · Blood 2022
    PMID 35839448DOI
    What this evidence cannot show

    The principal competing account. The range of originating differentiation stages is nevertheless consistent with a pre-existing substrate, so this paper supports both readings in different parts.

    Evidence that argues against this (1)
    • Shared breakpoints establish ancestry, not direction, which is precisely what separates competitive release from reprogramming. Chromatin-accessibility change accompanies both. Journal of Clinical Oncology 2025
  4. E4weak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.paper

    In a 6-month-old with KMT2A::EPS15 B-ALL who switched to AML after blinatumomab, ClonoSEQ found three dominant immunoglobulin sequences shared across the original B-ALL, the switched AML and the post-transplant relapse, with three of four additional late sequences already present at low abundance in the diagnostic sample.

    Lineage Switch in an Infant B-Lymphoblastic Leukemia With t(1;11)(p32;q23); KMT2A/EPS15, Following Blinatumomab Therapy. · Du J, Chisholm KM, Tsuchiya K, et al. · Pediatric and Developmental Pathology 2021
    PMID 33749383DOI
    What this evidence cannot show

    One case, and immunoglobulin clonality tracks the clone rather than its lineage program. The authors interpret their own data as reprogramming.

    Evidence that argues against this (1)
    • Immunoglobulin rearrangement is inherited by every descendant of the cell that made it, so shared IGH/IGL sequences across three phases are equally consistent with both models. One patient. Blood Cells, Molecules and Diseases 2010
  5. E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    With median follow-up 4.2 years, 30 KMT2A-rearranged infants receiving Interfant-06 chemotherapy plus one post-induction blinatumomab course had 4-year disease-free survival of 83.3% versus 44.0% and overall survival 93.3% versus 60.2% against selected historical controls.

    Sustained Benefit of Blinatumomab in Infants With KMT2A-Rearranged ALL: Long-Term Outcomes, Toxicity, and Pharmacokinetics · Vieira Martins M, de Lorenzo P, Kotecha RS, et al. · Journal of Clinical Oncology 2026
    PMID 41505662DOI
    What this evidence cannot show

    n=30 with historical controls. Cited here to bound what any switch-risk biomarker may ethically be used for.

    Evidence that argues against this (1)
    • Thirty infants against a historical comparator, and the comparator was SELECTED from Interfant-06 rather than randomised. Every survival figure this beat draws from that study — 81.6% two-year DFS, 83.3% four-year DFS, 93.3% four-year OS — rests on the same 30 children, so quoting several of them across entries gives one small single-arm study the appearance of several. With 30 patients the confidence interval on any of those figures spans a range that overlaps the control estimate. It is also the wrong evidence for this hypothesis: a survival improvement with added blinatumomab says nothing about whether the switches that do occur come from pre-existing clones. Leukemia 2021
  6. E6strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Among 420 children and young adults treated with CD19 CAR T cells, 166 relapsed of which 12 (7.2%) were lineage switch, and KMT2A rearrangement was the only pre-infusion factor associated with lineage-switch relapse.

    Preinfusion factors impacting relapse immunophenotype following CD19 CAR T cells · Lamble AJ, Myers RM, Taraseviciute A, et al. · Blood Advances 2023
    PMID 35482927DOI
    What this evidence cannot show

    That genotype is currently the only pre-infusion predictor is the gap this hypothesis proposes to fill; it also means any new biomarker must beat genotype alone to be worth measuring.

    Evidence that argues against this (1)
    • Twelve lineage-switch events among 166 relapses, and KMT2A rearrangement as the only associated pre-infusion factor is a univariate result on twelve outcomes. Statistics in Medicine 2006
  7. E7strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Single-cell multiomics of KMT2A-rearranged ALL across patient ages revealed preexisting lymphomyeloid primed progenitors and myeloid blasts already present at initial diagnosis of B-ALL, and clonal tracking in two patients whose leukaemia underwent a lineage switch documented the evolution of those same clones into frank acute myeloid leukaemia. The study also found leukaemia cells from patients under six months have significantly increased lineage plasticity, and identified a hematopoietic stem and progenitor-like population in the blood of younger patients containing leukemic blasts.

    Single-cell multiomics reveals increased plasticity, resistant populations, and stem-cell-like blasts in KMT2A-rearranged leukemia · Chen C, Yu W, Alikarami F, Qiu Q, Chen CH, Flournoy J, Gao P, Uzun Y, Fang L, Davenport JW, Hu Y, Zhu Q, Wang K, Libbrecht C, Felmeister A, Rozich I, Ding YY, Hunger SP, Felix CA, Wu H, Brown PA, Guest EM, Barrett DM, Bernt KM, Tan K · Blood 139(14):2198-2211 2022
    PMID 34864916DOI
    What this evidence cannot show

    THE CLONAL-TRACKING RESULT IS n=2. Two patients who switched is a demonstration that the route exists, NOT an estimate of how often it is taken, and this hypothesis's quantitative claim - that diagnostic abundance of the myeloid-primed subpopulation predicts switch risk - is not tested by two cases. The study is also not designed as a test of this hypothesis: preexisting lymphomyeloid primed progenitors are reported as an observation alongside its main age-plasticity finding, with no control cohort of non-switching KMT2A-r patients matched on that abundance, so the key comparison - do switchers have MORE of these cells at diagnosis than non-switchers - remains unmade. Detection of a myeloid-primed progenitor does not establish that it is fusion-positive in every case.

    Evidence that argues against this (2)
    • This evidence was added to correct an error, which is a circumstance that encourages reading it generously. The finding that cuts the other way is in the same paper: leukaemias from patients under six months show markedly increased lineage PLASTICITY, which is the CHD4 reprogramming account's central claim, not this hypothesis's. A paper reporting both a pre-existing subpopulation and heightened plasticity supports the mixed model - pre-existing substrate PLUS induced transition - over the competitive-release hypothesis as stated.
    • That this paper sat in the corpus, cited by five entries, while this entry asserted its non-existence, is evidence about the project's instruments rather than about the biology. The same cycle found the citation index used to check exactly this kind of claim was missing 73.7% of the corpus's PMIDs. Any other absence claim in this graph written before 2026-09-23 carries the same risk and has not been re-checked.
Connected to