A child who made a KMT2A fusion during chemotherapy, lost it again, and never got leukaemia in 12.8 years
also called Felix 2021 neuroblastoma KMT2A-MAML2 clone · the regressed KMT2A-MAML2 clone · covert KMT2A-R clone
A child being treated for high-risk neuroblastoma developed a cell clone carrying a KMT2A fusion - the kind that normally means leukaemia. It was there, at low level, and then it was gone, and after nearly thirteen years of follow-up the child never developed leukaemia.
The same thing, for a scientist
Twelve patients who completed TOP2-poison chemotherapy in a high-risk neuroblastoma regimen and who did **not** develop therapy-related leukaemia were studied. One harboured an inv(11) that was a KMT2A rearrangement, expressing a KMT2A-MAML2 transcript at low level. The patient was followed prospectively; the inv(11) was undetectable in ensuing samples; no leukaemia developed over 12.8 years. Enriched etoposide-induced TOP2A cleavage in the relevant MAML2 genomic region supports a TOP2A DNA-damage mechanism for how the rearrangement formed. The authors conclude that covert KMT2A-rearranged clones may occur in a small minority of patients after TOP2-poison chemotherapy, and that **not all KMT2A rearrangements herald a therapy-related leukaemia diagnosis** [E1]. This is the cleanest single observation in this graph that a KMT2A fusion is not by itself sufficient to cause leukaemia, and it is a prospective, longitudinal human observation rather than a model.
Why this one case is worth an entry1 deeper
This graph already holds the neonatal-blood-spot and twin-concordance evidence that KMT2A fusions can exist in people who never develop leukaemia - epi-fusion-positive-newborns-versus-leukemia-risk, pt-story-concordant-leukaemia-infant-twins-in-utero-origin. Those establish the same point for the prenatal, de novo setting.
This case establishes it for the therapy-related setting, which is different in three ways that matter:
- The clone arose in a child whose marrow had just been through cytotoxic therapy - a selective environment, not a quiet one.
- The fusion was expressed as a transcript, not merely present as DNA. A silent rearrangement and a transcribed one are different objects, and the commonest deflationary reading of blood-spot data (the fusion is there but does nothing) is harder to apply here.
- It was followed prospectively for 12.8 years, so this is not a retrospective absence of a diagnosis; it is a documented appearance and disappearance.
The detection floor, stated honestly
'Undetectable in ensuing samples' means below the assay's floor 1. A clone at one cell in a million is undetectable and is not gone. So the strongest defensible statement is that the clone fell below detection and never expanded, not that it was eliminated. The distinction matters because the two imply different things: elimination suggests immune or intrinsic clearance, while persistence below threshold suggests the clone was simply not competitive - and the second is the more interesting for the sufficiency question.
What is missing that would make it decisive
The study does not report:
- Which cell carried the fusion. A clone in a committed progenitor with limited self-renewal would be expected to disappear; one in an HSC would not. This is the single fact that would most change the interpretation and it is not available.
- The genomic breakpoint in KMT2A, so the clone cannot be placed relative to the major BCR.
- Transcript quantification beyond 'low level'.
- What was different about this child compared with the several reported patients in whom exactly this fusion did produce leukaemia after cytotoxic therapy.
The last is the comparison the field needs and nobody has it: one child's KMT2A-MAML2 clone regressed and several others' did not, and no feature distinguishes them.
The TOP2A finding riding alongside
The study mapped enriched etoposide-induced TOP2A cleavage in the relevant MAML2 genomic region 1. That is a partner-specific fragility measurement, which is rare - this graph's fragility mapping (breakpoint-kmt2a-bcr-architecture) is almost entirely about the KMT2A side. It supports a TOP2A DNA-damage mechanism for how this particular fusion forms, and it is consistent with the observation that most reported KMT2A-MAML2 leukaemias follow cytotoxic therapy.
It also creates a tension with the strong exon 2 acceptor preference recorded in breakpoint-maml2-exon-2-preferred-acceptor: if TOP2A cleavage sites dictate where MAML2 breaks, that preference may be chemistry rather than functional selection. The study did not analyse its cleave-ome data against junction positions, and doing so would be a reanalysis rather than a new experiment.
| age at dx | paediatric; exact age not reported in the abstract |
|---|---|
| fusion | fusion-kmt2a-maml2 |
| course | Prospectively followed. The inv(11) was undetectable in subsequent samples. No leukaemia developed. |
| outcome | No therapy-related leukaemia after 12.8 years of follow-up. |
| cohort context | One of 12 patients studied after TOP2-poison chemotherapy who had not developed therapy-related leukaemia |
| source type | case report within a small prospective cohort study |
| consent note | public source only, no PHI; no identifying detail beyond what the publication reports |
| why it matters | A prospective human observation that a KMT2A fusion can arise, be expressed, and regress without causing leukaemia - the therapy-related counterpart to the neonatal-blood-spot evidence for de novo fusions. |
presentation
[ "high-risk neuroblastoma treated with a TOP2-poison-containing regimen", "inv(11) KMT2A rearrangement detected after completion of therapy", "KMT2A-MAML2 transcript expressed at low level" ]
not reported
[ "which cell type carried the clone (HSC versus committed progenitor)", "the KMT2A genomic breakpoint, so BCR assignment is impossible", "quantitative transcript level", "any feature distinguishing this clone from those that did progress" ]
- If KMT2A fusions can arise and regress after TOP2-poison therapy, then any screening programme that looks for them in survivors will find clones that would never have caused disease. Detecting them creates a monitoring and anxiety burden with no established action, which is the classic overdiagnosis failure mode and would fall on children already treated for cancer.
- The result should not be used to reassure. One child's clone regressed; several other patients with the same fusion developed acute leukaemia. Nothing published lets a clinician tell which case they are looking at.
- The insufficiency reading cuts both ways for therapeutics. If the fusion needs a second event or a permissive cell state, then a drug against the fusion alone may be attacking a necessary but not sufficient lesion, and the co-operating factor is where the real dependency lives.
- Which cell carried the clone? If it was a committed progenitor, its disappearance is unremarkable and the case proves much less than it appears to; if it was an HSC, the fusion demonstrably failed to transform a self-renewing cell over more than a decade.
- What distinguished this child from the several patients in whom the same fusion did cause leukaemia? Neither host genetics, cumulative epipodophyllotoxin dose, nor co-occurring lesions are reported for any of them.
- How often do covert KMT2A-rearranged clones arise after TOP2-poison therapy? One in twelve in a cohort selected for not developing leukaemia is the only figure available, and the design cannot turn it into an incidence.
- Does the study's TOP2A cleave-ome data, reanalysed against the published MAML2 junction positions, explain the strong exon 2 acceptor preference as chemistry rather than selection? The data exist and the analysis has not been done.
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Among 12 patients without therapy-related leukaemia studied after completing TOP2-poison chemotherapy in a high-risk neuroblastoma regimen, one harboured an inv(11) KMT2A rearrangement expressing a low-level KMT2A-MAML2 transcript; the inv(11) was undetectable in ensuing samples and no leukaemia developed over 12.8 years of prospective follow-up; enriched etoposide-induced TOP2A cleavage in the relevant MAML2 genomic region supports a TOP2A DNA-damage mechanism; the authors 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 with a rearrangement out of twelve studied, and the twelve were selected for NOT having developed therapy-related leukaemia, so the design cannot estimate how often such clones arise or how often they progress. 'Undetectable' is bounded by assay sensitivity: a clone below the floor is not a clone that is gone. The cell type carrying the clone, the KMT2A genomic breakpoint and the transcript level are not reported. The TOP2A cleave-ome mapping is in a model system, not in the patient's cells. Abstract read in full; the body was not read.
Evidence that argues against this (1)
- Several independent patients with this exact fusion did develop leukaemia after cytotoxic therapy - secondary T-ALL after Ph-positive B-ALL treatment, secondary T-ALL after hepatoblastoma, T-lymphoblastic lymphoma after AML, secondary AML and MDS. Regression in one child does not generalise, and the study offers nothing that distinguishes this clone from those. Mariani RA et al. 2020
- E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
The same fusion in other children and adults did progress to acute leukaemia after cytotoxic therapy, including secondary T-cell ALL after chemotherapy for Philadelphia-chromosome-positive B-ALL, making regression the exception rather than the expectation.
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
Case reports, so the denominator is unknown: how many children develop a KMT2A-MAML2 clone after cytotoxic therapy is not measurable from the published literature, and therefore neither is the fraction that progress. The comparison between progressing and regressing clones is therefore qualitative. Abstract read in full.
Evidence that argues against this (1)
- Case reports are published because something happened; a clone that regresses silently generates no report, so the literature is structurally biased toward progression. The true ratio of regressing to progressing clones is unknown and is probably far more favourable than the case literature suggests - which, if so, strengthens rather than weakens the insufficiency reading. Felix CA et al. 2021
- related to →KMT2A::MAML2, inv(11)(q21q23) - a therapy-related fusion that also causes a solid tumour
the fusion arose during TOP2-poison therapy, was expressed, became undetectable, and caused no leukaemia in 12.8 years
the therapy-related arm of the sufficiency argument: expressed fusion, prospective follow-up, no disease
same conclusion from the therapy-related setting that the blood-spot data give for the prenatal setting