A KMT2A fusion is necessary but not sufficient: the cell it lands in decides whether there is a disease and which one
also called fusion insufficiency hypothesis · permissive cell hypothesis · cell-of-origin determines disease for KMT2A fusions
The same KMT2A fusion can produce childhood leukaemia, an adult thymus tumour, or nothing at all for thirteen years. If one genetic change gives three different answers, the change cannot be the whole story - the cell it happens in is doing much of the work.
The same thing, for a scientist
KMT2A::MAML2 provides an unusually clean three-armed natural experiment in humans, with the oncogene held constant. Arm A: in a thymic epithelial cell it produces type B2/B3 thymoma in middle-aged adults, at 4% of thymomas, with infrequent co-occurring alterations [E1]. Arm B: in a haematopoietic cell after cytotoxic therapy it produces acute leukaemia - secondary AML, MDS, T-ALL, T-lymphoblastic lymphoma, B-ALL - across at least eight reported patients [E2][E3]. Arm C: in one child it arose during TOP2-poison chemotherapy for neuroblastoma, was expressed at low level, became undetectable and produced no leukaemia over 12.8 years of prospective follow-up [E4]. Same fusion, same junction architecture, three outcomes. The hypothesis is that the fusion is **necessary and not sufficient**, and that what it meets - cell identity, chromatin state, self-renewal capacity, developmental window - determines whether a disease appears and what kind. This is not a new idea in the field; what is new here is that KMT2A::MAML2 supplies all three arms of the comparison in human tissue without any model system, which no other KMT2A partner does.
The three arms, and why they are a real comparison2 deeper
| arm A: thymoma | arm B: leukaemia | arm C: no disease | |
|---|---|---|---|
| cell | thymic epithelial | haematopoietic progenitor | unknown, haematopoietic |
| fusion | KMT2A ex 8-11 :: MAML2 ex 2 | KMT2A ex 7-9 :: MAML2 ex 2 | inv(11), transcript expressed |
| context | none reported | prior cytotoxic therapy in most | TOP2-poison therapy |
| n | 11 | at least 8 | 1 |
| outcome | epithelial neoplasm, adults | acute leukaemia | clone regressed, 12.8 y |
| source | 1 | 23 | 4 |
The junctions differ only in which major-BCR KMT2A exon is the donor, and every reported case uses the MAML2 exon 2 acceptor, so the protein is essentially the same molecule across all three arms. What varies is the cell.
What this is not
It is not evidence that the fusion is unimportant. In arms A and B it is the only recurrent lesion identified, and in thymoma concurrent alterations are explicitly infrequent 1. The hypothesis is about sufficiency, not necessity. Nor is it a claim that arm C proves clearance: 'undetectable' is bounded by assay sensitivity, and the clone may have persisted below the floor 4.
Where it agrees with what the graph already holds
This graph already carries the de novo version of the same argument: KMT2A fusions detectable in neonatal blood spots at rates far exceeding the incidence of infant leukaemia, and twin-concordance data placing the fusion in utero with variable latency and penetrance. This entry adds the therapy-related arm and the cross-tissue arm. Three independent lines converging on the same conclusion is worth more than any one of them, and the convergence is across settings that share no methodology.
What the hypothesis predicts that could fail
The point of writing it down is the falsifiers, so they are stated sharply.
- If the transcriptional programme of KMT2A::MAML2 thymoma turns out to be the same as the leukaemia programme - HOXA and MEIS1 induced in an epithelial cell - then the fusion is imposing its own programme regardless of cell, and the cell's role shrinks to a modifier of tempo rather than a determinant of disease.
- If the arm C clone is shown to have sat in a committed progenitor with no self-renewal, its regression is unremarkable and arm C stops being evidence about sufficiency at all.
- If co-operating lesions are found in arm B patients and not in arm A, the difference is genetic rather than cellular and the hypothesis is answering the wrong question.
The first of these is directly testable on existing archival thymoma tissue and is the cheapest disconfirming experiment available.
Why it matters for a therapeutic programme
If the fusion is necessary and not sufficient, then the dependency a drug should attack may not be the fusion itself but whatever the permissive cell supplies. Two concrete consequences:
- A fusion-directed agent that works in one cell context may do nothing in another, which is exactly the risk in treating all 'KMT2A-rearranged' disease as one entity - the standing concern in this graph about menin inhibitors being prescribed on rearrangement status alone.
- The permissive factor is itself a target, and it would be a target shared across partners rather than partner-specific. That is a better commercial and scientific proposition than a 30-patient fusion, and nothing in the current literature identifies what it is.
| statement | A KMT2A fusion is necessary but not sufficient for disease. What determines whether a disease appears, and which disease, is the identity and state of the cell the fusion arises in. KMT2A::MAML2 demonstrates this in humans with the oncogene held constant: thymic epithelial cell to thymoma, haematopoietic progenitor to acute leukaemia, and in one documented case to nothing at all over 12.8 years. |
|---|---|
| origin | data-pattern |
| prior probability | 0.75 |
predictions
[ "KMT2A::MAML2 thymoma will NOT show the canonical HOXA9/MEIS1 leukaemia programme; if it does, the fusion imposes its programme regardless of cell and this hypothesis weakens sharply.", "The regressed clone of arm C, if the archived material allows it to be localised, will be found in a committed progenitor rather than an HSC, or else in an HSC that lacked some second requirement.", "Other KMT2A partners with solid-tumour counterparts will show the same pattern - the same fusion, different tissue, different disease - and the thymoma/leukaemia split will not turn out to be unique to MAML2.", "Screening survivors of TOP2-poison therapy will find KMT2A-rearranged clones at a rate substantially higher than the rate of therapy-related leukaemia.", "Co-operating lesions will NOT distinguish arm A from arm B; if they do, the difference is genetic rather than cellular." ]
experiments
[ "exp-menin-dependence-kmt2a-maml2-thymoma" ]
analyses
[ "ana-enumeration-usp2-maml2-second-bcr" ]
falsifiers
[ "KMT2A::MAML2 thymoma expresses the same HOXA/MEIS1 programme as KMT2A-rearranged leukaemia, showing the fusion dictates its output irrespective of cell type.", "The arm C clone is shown to have resided in a committed progenitor with no self-renewal capacity, removing its force as evidence about sufficiency.", "Systematic sequencing finds co-operating driver lesions in the leukaemia cases and not in the thymomas, relocating the explanation from cell identity to genetics.", "A KMT2A::MAML2 knock-in transforms every cell type tested with equal efficiency, showing no permissive-cell requirement.", "The thymoma cases turn out to arise from a thymic haematopoietic rather than epithelial cell, collapsing arms A and B into one." ]
- If the fusion is necessary but not sufficient, a fusion-directed drug attacks only part of the requirement, and clinical failure would be misread as target failure when it is really context failure. That is a specific way a good target can look like a bad one.
- The permissive-cell reading argues for treating KMT2A-rearranged disease as several diseases, which cuts against enrolling patients on rearrangement status alone - the current basis for menin-inhibitor eligibility.
- If covert KMT2A clones are common after TOP2-poison therapy and mostly harmless, screening for them in survivors would generate overdiagnosis in children who have already been treated for cancer, with no established action to offer.
- What is the permissive factor? If cell identity decides, something concrete supplies it - an open HOXA cluster, a particular CpG methylation landscape, LEDGF/PSIP1 abundance, a self-renewal programme - and none of the candidates has been tested against this fusion in either tissue.
- Do KMT2A::MAML2 thymomas express HOXA genes? One RNA-seq comparison against fusion-negative thymoma would materially move this hypothesis in either direction, on tissue that already exists.
- Is the thymoma arm truly epithelial in origin, or could these tumours arise from a thymic cell closer to the T lineage? The T-lineage bias of the leukaemia arm makes this worth checking before the comparison is treated as cross-lineage.
- How many other KMT2A partners have a solid-tumour counterpart? If MAML2 is the only one, it is a curiosity; if several do, the insufficiency argument generalises and the graph should look for them systematically.
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
KMT2A-MAML2 occurs in 11 patients with type B2/B3 thymoma, at 4% of all thymomas and 6% of B2/B3 histology, in adults with a median age of 48, with concurrent genomic alterations infrequent - i.e. the fusion in a thymic epithelial cell produces an epithelial neoplasm, not leukaemia.
Pan-Cancer Landscape Analysis Reveals Recurrent KMT2A-MAML2 Gene Fusion in Aggressive Histologic Subtypes of Thymoma. · Massoth LR, Hung YP, Dias-Santagata D, Onozato M, Shah N, Severson E, Duncan D, Gillespie BJ, Williams NF, Ross JS, Vergilio JA, Harkins SK, Glomski K, Nardi V, Zukerberg LR, Hasserjian RP, Louissaint A, Williams EA · JCO Precision Oncology 2020What this evidence cannot show
A commercial advanced-cancer profiling cohort, so the thymoma denominator is enriched for refractory disease. 'Concurrent alterations infrequent' is reported without a quantitative comparison against fusion-negative thymomas. Crucially, the cell of origin is inferred from the tumour's histology, not demonstrated - thymomas are epithelial neoplasms, but no lineage-tracing has placed the fusion in a specific thymic epithelial progenitor. Abstract read in full.
Evidence that argues against this (1)
- Thymus is a haematopoietic-adjacent organ full of developing T cells, and the reported KMT2A-MAML2 leukaemias are strikingly T-lineage. It is possible that both diseases arise in the thymus from different resident cells, in which case the anatomical site is shared and the comparison is less clean than 'a completely different organ' suggests. Mariani RA et al. 2020
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
The same fusion in haematopoietic cells produces acute leukaemia: secondary AML and MDS in the founding report, with MLL exon 7 fused to MAML2 exon 2 - the same acceptor exon as the thymoma cases.
Identification of a novel fusion gene MLL-MAML2 in secondary acute myelogenous leukemia and myelodysplastic syndrome with inv(11)(q21q23). · Nemoto N, Suzukawa K, Shimizu S, Shinagawa A, Takei N, Taki T, Hayashi Y, Kojima H, Kawakami Y, Nagasawa T · Genes, Chromosomes & Cancer 2007What this evidence cannot show
Two patients. Both had prior cytotoxic therapy, so the haematopoietic arm is confounded by a mutagenic exposure that the thymoma arm does not report - the cells differ, but so does the context. Abstract read in full.
Evidence that argues against this (1)
- Almost every leukaemia case of this fusion follows cytotoxic therapy while no thymoma case does, so the comparison confounds cell type with prior genotoxic exposure. A therapy-exposed marrow may carry additional unmeasured lesions that a thymus does not, and the studies did not look. Massoth LR et al. 2020
- E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
KMT2A-MAML2 leukaemias span multiple lineages and ages - secondary T-cell ALL in children including the youngest reported patient, B-ALL in a 47-year-old after sarcoma, T-lymphoblastic lymphoma after RUNX1::RUNX1T1 AML - so within the haematopoietic arm alone the fusion does not specify a single disease.
Inv(11)(q21q23); KMT2A-MAML2, a Recurrent Genetic Abnormality in T-Cell Therapy-related Acute Lymphoblastic Leukemia. · Mariani RA, Silva M, Caparelli E, Jennings LJ, Yap KL, Leuer KM, Weinstein J, Gong S · Journal of Pediatric Hematology/Oncology 2020What this evidence cannot show
The lineage spread is assembled from separate single-case reports rather than from one cohort, so it reflects what was published as much as what occurs. No case has a documented cell of origin. Abstract read in full.
Evidence that argues against this (1)
- Lineage heterogeneity is characteristic of KMT2A-rearranged leukaemia in general, independent of partner, so multi-lineage output may be a property of the KMT2A moiety rather than evidence that the cell of origin is doing the deciding. Nemoto N et al. 2007
- E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
In one child, a KMT2A-MAML2 clone arising during TOP2-poison chemotherapy for neuroblastoma expressed a low-level transcript, became undetectable and produced no leukaemia over 12.8 years, leading the authors to conclude that not all KMT2A rearrangements herald a therapy-related leukaemia diagnosis.
KMT2A-MAML2 rearrangement emerged and regressed during neuroblastoma therapy without leukemia after 12.8-year follow-up. · Felix CA, Slater DJ, Davenport JW, Yu X, Gregory BD, Li MM, Rappaport EF, Cheung NV · Pediatric Blood & Cancer 2021What this evidence cannot show
One patient out of twelve studied, and the twelve were selected for not having developed leukaemia. 'Undetectable' is bounded by assay sensitivity. The cell type carrying the clone is not reported, which is the fact that would most change the interpretation. Abstract read in full.
Evidence that argues against this (1)
- If the clone sat in a committed progenitor rather than a self-renewing stem cell, its disappearance is the expected behaviour of any short-lived cell and says nothing about the fusion's transforming power. The study does not exclude this and it is the most deflationary reading available. Mariani RA et al. 2020
the therapy-related arm of the sufficiency argument: expressed fusion, prospective follow-up, no disease
- related to →KMT2A::MAML2 type B2/B3 thymoma - the same fusion, a different organ, a different disease
arm A of the three-arm natural experiment: same fusion, epithelial cell, different disease
- related to →The differentiation block — what it actually is, operationally, in KMT2A::AFF1 leukaemia
if the cell decides the outcome, the differentiation block is a property of the permissive cell state rather than of the fusion alone
the fusion keeps Evi1 on where the founding cell had it on and cannot switch it on otherwise
The cleanest natural experiment for the permissive-cell premise: one oncoprotein held constant across three unrelated tissues - congenital fibrosarcoma, cellular mesoblastic nephroma and secretory breast carcinoma (12 of 13 cases) - so the disease is set by the cell and not by the fusion. LIMIT: a signalling kinase with generic RAS-MAPK/PI3K/STAT3 output has more room to be cell-agnostic than a chromatin oncoprotein does.
Same 5-prime gene as the index fusion and the opposite answer on sufficiency: a KMT2A::MAML2 clone emerged during neuroblastoma therapy, was expressed at low level, became undetectable and produced no leukaemia over 12.8 years.