Oncofusion

start / graph / epi-fusion-positive-newborns-versus-leukemia-risk

Epidemiologymoderate confidence 3 of 5: One strong study, or several smaller consistent ones.Q1b-genesis2

What fraction of fusion-positive newborns ever get leukaemia? For ETV6::RUNX1, very few. For KMT2A, nobody knows, and the biology says the answer may be most of them

also called covert preleukaemic clones · fusion prevalence at birth · necessary but not sufficient

For one common childhood-leukaemia fusion, far more babies are born carrying it than ever get ill, so the fusion alone is not enough. For the KMT2A fusion that this project is about, nobody has ever measured how many newborns carry it, and everything else we know suggests that a baby who carries it is in serious trouble.

The same thing, for a scientist

This entry exists to stop a widely repeated generalisation being applied where it does not belong. The claim that 'about 1% of newborns carry a leukaemia fusion, a hundred-fold more than ever develop leukaemia' comes from Mori et al., who screened cord bloods for ETV6::RUNX1 and RUNX1::RUNX1T1 [E1]. It was directly contradicted by Lausten-Thomsen et al., who screened 1,417 cord bloods with a highly sensitive real-time RT-PCR and, after flow-sorting and re-testing every initial positive, found no confirmed positives at all, concluding the frequency is markedly lower than previously suggested [E2]; a Danish National Birth Cohort attempt reached the same practical conclusion and put the estimated ETV6::RUNX1-positive cell frequency below 10^-5 [E3]. Crucially, **none of these studies tested for a KMT2A fusion.** There is no published, denominator-based estimate of KMT2A fusion prevalence in newborns. What is known points the opposite way from the ETV6::RUNX1 story: near-total monozygotic twin concordance and a latency of months rather than years imply the fusion is close to sufficient in the fetal target cell, so the fraction of KMT2A-fusion-positive newborns who develop leukaemia may be high rather than negligible.

Go deeper
The arithmetic, with explicit assumptions1 deeper

US infant (under 1 year) acute leukaemia incidence is about 41 per million infants per year, roughly 160 cases; KMT2A rearrangement is present in 70-80% of infant ALL and about 50% of infant AML (epi-infant-leukemia-incidence). That yields on the order of 100 KMT2A-rearranged infant leukaemias per year against roughly 3.6 million US births, i.e. a cumulative risk of about 1 in 30,000 to 1 in 40,000 births. If the fusion were nearly fully penetrant, the prevalence of KMT2A-fusion-positive newborns would be of the same order. If instead the ETV6::RUNX1 pattern applied and penetrance were about 1%, prevalence would be around 1 in 300. These two hypotheses differ by a hundred-fold, and the experiment that would distinguish them has been done once, at small n, with a negative result. Yamamoto et al. screened 65 normal cord bloods by nested PCR and detected no MLL/AF-4 chimeric mRNA at all [E_yamamoto].

That is a real measurement and this entry previously said none existed. But it does not settle the question: with 65 samples and zero events the rule of three puts the 95% upper bound near 4.5%, which excludes the 1-in-300 high-penetrance scenario only weakly and says nothing at all about the 1-in-30,000 range. The honest statement is measured once, at small n, and negative — not never measured, and not resolved.

The same study found something that complicates any future attempt: an in-frame fusion of AFF1 exon 11 to exon 4 or 5 in 20 of 65 (31%) healthy cord bloods, and in 4 of 6 t(4;11) leukaemia samples. That is a high-frequency, in-frame, partner-gene transcript in healthy newborns, so an RNA assay aimed at this fusion has a specific and common false-positive class to exclude — and it lives on AFF1's own transcript rather than on a trans-spliced chimera.

What the ETV6::RUNX1 controversy actually settled

Mori et al. reported TEL-AML1 or AML1-ETO in cord blood at a frequency ~100-fold greater than the risk of the corresponding leukaemia, with positive cell frequencies of 10^-4 to 10^-3 and lineage-restricted FISH confirmation 1. Lausten-Thomsen et al. then screened 1,417 cord bloods; 14 were positive in first-run PCR below 10^-5, 9 hybridised, and on re-analysis of stored mRNA and of flow-sorted CD19+, CD8+ and double-negative fractions (mean 18 million cells sorted) none confirmed 2. The honest reading is not that Mori was wrong about the direction of the asymmetry, which the 5% twin concordance and decade-long latencies independently require, but that the magnitude is unreliable and the assay is at the edge of what RT-PCR can do without generating artefacts.

The only KMT2A-specific prevalence data are at the DNA level and are hard to believe at face value

Brassesco et al. screened peripheral lymphocytes from 100 healthy adults by inverse PCR and reported KMT2A rearrangements in 49% of subjects, with sequences corresponding to t(4;11) in 66% and t(9;11) in 20% of those, but RT-PCR failed to detect any t(4;11) expression in the entire population 4. Taken at face value this says that KMT2A breaks-and-rejoins are a common, largely silent background event in normal blood and that the rate-limiting step is not the break but the productive, expressed, in-frame fusion in the right cell. Taken sceptically, inverse PCR at these frequencies is notoriously artefact-prone, the study is small, in a low-visibility journal, and has not been independently replicated. The graph records it because it is the only KMT2A-specific number that exists, and flags it as weak.

Necessary and sufficient, stated precisely

Three independent lines converge on 'necessary, and closer to sufficient than any other paediatric leukaemia fusion': near-total twin concordance with short latency (epi-twin-concordance-infant-kmt2a-leukemia); a mean of 1.3 non-silent somatic mutations in the dominant clone, with cooperating RAS-pathway lesions being subclonal and frequently lost at relapse (comut-low-mutation-burden-infant-kmt2a-r, mechanism-subclonal-cooperating-lesions-lost-at-relapse); and the fact that no second recurrent driver has emerged from whole-genome sequencing. Against that sits the hardest counter-evidence in the beat: enforced MLL-AF4 expression in human cord-blood CD34+ cells enhances engraftment and clonogenicity but does not transform, and adding FLT3 activation does not rescue it 5. Something is missing from the human experimental system, and the leading candidate is the fetal cell context itself rather than a mutation (see hyp-second-hit-is-developmental-window-not-mutation).

Facts and figures
metriccumulative risk of KMT2A-rearranged infant leukaemia per birth
value30000
unit1 in N live births (approximate, derived)
populationUS live births
regionUnited States
year2013
sourceDerived in-graph from epi-infant-leukemia-incidence (41 per million infants per year; ~160 infant leukaemias; KMT2A-r in 70-80% of infant ALL and ~50% of infant AML) against ~3.6M US births
derivation(~90 infant ALL x 0.75) + (~70 infant AML x 0.50) = ~102 KMT2A-r infant leukaemias per year; 3.6e6 births / 102 = ~1 in 35,000
prevalence of kmt2a fusion positive newbornsUNMEASURED - no published denominator-based estimate exists
comparator etv6 runx1claimed ~1% of newborns (PMID 12048236) but not reproduced in 1,417 cord bloods (PMID 20713965)
How this could bite us back
  • If the true penetrance in KMT2A-fusion-positive newborns is high, the ETV6::RUNX1-derived comfort that 'most fusion carriers are fine' does not apply, and any future decision to observe rather than treat a detected carrier would be wrong.
  • If instead penetrance is low, then a menin inhibitor given to a screen-detected carrier would be treating mostly healthy infants with a differentiation-syndrome-capable drug, and the harm calculation inverts.
  • Because the prevalence is unmeasured, a screening programme would be launched blind to its own positive predictive value. That is the classic failure mode of screening programmes, and it is entirely avoidable by doing the biobank study first.
Still open
  1. Yamamoto 1998 measured this once at n=65 and found zero. What would a modern re-measurement cost, using droplet digital PCR or DNA breakpoint capture on banked cord bloods, and what n is needed to distinguish 1-in-300 from 1-in-30,000?
  2. THE experiment this beat most wants: screen a large cord-blood biobank (>=10,000 samples) for KMT2A fusions at the genomic-DNA level with targeted capture across the breakpoint cluster region, not by RT-PCR. Genomic capture avoids the RNA-degradation problem that defeated the Danish cohort [E3] and detects non-expressing clones that RT-PCR misses. The answer, whatever it is, reprices every prevention and screening idea in this graph.
  3. Are KMT2A breaks common and productive fusions rare, as [E4] hints? A matched design measuring break frequency (inverse PCR or translocation-capture sequencing) and transcript frequency in the same samples would separate 'breaking is rare' from 'surviving the break in the right cell is rare', and those two have completely different prevention implications.
  4. If a KMT2A-fusion-positive newborn were identified prospectively, what is the actual conversion rate? Only a prospective cohort can answer this and the required n is prohibitive at 1 in 30,000; a targeted cohort of monochorionic twins of index cases is the only feasible enriched design.
  5. Why does MLL-AF4 fail to transform human cord-blood CD34+ cells [E5] when twin data say it is nearly sufficient in vivo? Resolving this contradiction is the central unsolved problem of infant leukaemia modelling.
Every source behind this entry (6)
  1. E1weak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.paper

    ETV6::RUNX1 and RUNX1::RUNX1T1 fusion genes were detected in cord bloods at a frequency about 100-fold greater than the risk of the corresponding leukaemia.

    Chromosome translocations and covert leukemic clones are generated during normal fetal development · Mori H, Colman SM, Xiao Z, Ford AM, Healy LE, Donaldson C, Hows JM, Navarrete C, Greaves M · Proceedings of the National Academy of Sciences of the USA 2002
    PMID 12048236DOI
    What this evidence cannot show

    Modest sample size, RT-PCR at the limit of detection, and a result that a much larger and more stringent study failed to reproduce. Did not test any KMT2A fusion. Confidence downgraded to weak because of E2.

    Evidence that argues against this (1)
    • 1,417 cord bloods screened with sensitive real-time RT-PCR plus flow-sorted re-testing yielded no confirmed ETV6::RUNX1 positives, indicating the frequency is markedly lower than Mori reported. Lausten-Thomsen U et al. 2011
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In 1,417 umbilical cord blood samples, no ETV6::RUNX1-positive cells could be confirmed after sorting and re-analysis, so the prevalence is markedly lower than earlier estimates.

    Prevalence of t(12;21)[ETV6-RUNX1]-positive cells in healthy neonates · Lausten-Thomsen U, Madsen HO, Vestergaard TR, Hjalgrim H, Nersting J, Schmiegelow K · Blood 2011
    PMID 20713965DOI
    What this evidence cannot show

    A negative result constrains the upper bound but cannot prove absence; cryopreservation and sorting lose cells, and a clone below the sorted-cell detection limit would be missed. Tests RNA, not genomic DNA, so a non-expressing rearranged clone would be invisible.

    Evidence that argues against this (2)
    • This paper was formally contested in print in the same journal issue, by the group whose 1% estimate it overturned. Greaves, Colman, Kearney and Ford published a correspondence on it arguing the negative result reflects the assay and the sorting strategy rather than the absence of the clone; a second correspondence from Zuna and colleagues argued the same from their own data, and the issue carried an editorial titled 'TEL-AML1 in cord blood: 1% or 0.01%?' framing the two-order-of-magnitude discrepancy as unresolved. The entry should not treat the low estimate as settled; it is one side of a live disagreement. Greaves M et al. 2011
    • The size of the disagreement is itself the finding: the accompanying editorial puts the two published prevalence estimates two orders of magnitude apart (1% versus 0.01%) and does not adjudicate between them. Any downstream calculation in this graph that divides leukaemia incidence by a cord-blood prevalence inherits a hundred-fold uncertainty in the denominator. Brown P 2011
  3. E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    In a Danish National Birth Cohort feasibility study the estimated frequency of ETV6::RUNX1-positive cells in newborns was below 10^-5, and RNA quality limits large-scale biobank screening.

    RT-PCR screening for ETV6-RUNX1-positive clones in cord blood from newborns in the Danish National Birth Cohort · Olsen M, Hjalgrim H, Melbye M, Madsen HO, Schmiegelow K · Journal of Pediatric Hematology/Oncology 2012
    PMID 22217495DOI
    What this evidence cannot show

    Primarily a methods/feasibility report; its main finding is that RNA must be prepared within 1-2 days of birth, which is itself a hard constraint on any population screening programme.

    Evidence that argues against this (1)
    • A DNA-based assay on healthy newborns returns a figure five orders of magnitude higher. Genomic inverse PCR for exploration of ligated breakpoints (GIPFEL), which detects the rearranged DNA rather than an expressed transcript, found an ETV6-RUNX1 fusion in five percent of healthy newborns. The RNA-based estimate of below 10^-5 and this DNA-based five percent are not measuring the same thing: a rearranged but transcriptionally silent clone is invisible to RT-PCR by construction. A prevalence figure in this field cannot be quoted without its assay. (Record and title fetched and checked; the full text is not open access, so the per-cohort numbers behind the headline were not read.) Schafer D et al. 2018
  4. E4weak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.paper

    KMT2A rearrangements were detectable by inverse PCR in peripheral lymphocytes of 49% of 100 healthy adults, yet no t(4;11) fusion transcript was expressed in any of them.

    MLL leukemia-associated rearrangements in peripheral blood lymphocytes from healthy individuals · Brassesco MS, Montaldi AP, Gras DE, de Paula Queiroz RG, Martinez-Rossi NM, Tone LG, Sakamoto-Hojo ET · Genetics and Molecular Biology 2009
    PMID 21637673DOI
    What this evidence cannot show

    Inverse PCR at these frequencies is highly artefact-prone; n=100; low-visibility journal; not independently replicated; adults rather than newborns. The result should be treated as a hypothesis (breaks are common, productive expressed fusions are rare) rather than as a prevalence estimate.

    Evidence that argues against this (1)
    • Contradicts the second half of the claim directly. Nested RT-PCR on a large number of peripheral blood samples from healthy individuals detected MLL-AF4 transcript, along with MLL partial tandem duplications, BCR-ABL p190 and p210, AML1-ETO, PML-RARA and CBFB-MYH11, all at measurable prevalence; cord blood from healthy neonates carried several of these too. So an expressed t(4;11) fusion transcript HAS been found in healthy people, and the tidy 'breaks are common, expressed fusions are rare' reading of the inverse-PCR result does not survive. Both studies use nested PCR at extreme sensitivity and both are therefore exposed to the same artefact risk, which is the honest summary: the assays disagree and neither is clean. Song J et al. 2011
  5. E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    MLL-AF4 expression in human cord-blood CD34+ cells enhances engraftment and clonogenicity but is not sufficient for leukaemogenesis, and adding FLT3 activation does not confer transformation.

    Ligand-independent FLT3 activation does not cooperate with MLL-AF4 to immortalize/transform cord blood CD34+ cells · Montes R, Ayllon V, Prieto C, Bursen A, Prelle C, Romero-Moya D, et al. · Leukemia 2014
    PMID 24240202DOI
    What this evidence cannot show

    Lentiviral overexpression from a heterologous promoter in postnatal cord-blood CD34+ cells is not the same as an endogenous-locus fusion in a first- or second-trimester fetal liver progenitor; a negative transformation result may reflect the wrong cell, the wrong dose or the wrong timing rather than genuine insufficiency. Xenograft follow-up was 16 weeks.

    Evidence that argues against this (1)
    • An endogenous-locus version of the same fusion DOES produce malignancy without an added second hit. Mll-AF4 knock-in mice made by homologous recombination in embryonic stem cells showed lymphoid and myeloid deregulation in young animals and went on to develop mixed lymphoid/myeloid hyperplasia and haematologic malignancies, most frequently B-cell lymphomas. That is a long latency and a different disease from infant B-ALL, so it does not rescue the human cord-blood experiment - but it shows the negative result is a property of the SYSTEM (lentiviral overexpression, postnatal human cord blood, 16-week follow-up) rather than a demonstration that the fusion is insufficient. Chen W et al. 2006
  6. E_yamamotomoderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    Nested PCR of 65 normal cord blood samples detected no MLL/AF-4 chimeric mRNA. The same study detected in-frame AFF1 exon 11 to exon 4 or 5 fusion transcripts in 20 of those 65 healthy cord bloods (31%), and in 4 of 6 leukaemic samples with t(4;11), while finding none in 21 peripheral blood samples or in fetal fibroblasts.

    High frequency of fusion transcripts of exon 11 and exon 4/5 in AF-4 gene is observed in cord blood, as well as leukemic cells from infant leukemia patients with t(4;11)(q21;q23) · Yamamoto S, Zaitsu M, Ishii E, Yatsuki H, Mizutani S, Eguchi M, Ihara K, Okamura T, Hara T, Miyazaki S · Leukemia 12:1398-403 1998
    PMID 9737688DOI
    What this evidence cannot show

    Read as abstract only. n=65 with zero events gives a 95% upper bound near 4.5% by the rule of three, so this constrains only the high-penetrance scenario and cannot address prevalence near the 1 in 30,000 birth incidence. 1998 nested PCR sensitivity is not stated in the abstract, and a negative result is only as good as the assay's limit of detection.

    Evidence that argues against this (1)
    • A single 65-sample negative is compatible with a wide range of true prevalences and should not be read as showing the fusion is absent from newborns. It is also 1998 methodology: modern droplet digital PCR or targeted DNA breakpoint capture would have a very different limit of detection, and the question deserves re-measurement rather than citation of this result as settled. Yamamoto S et al. 1998
Connected to