Mel Greaves (Centre for Evolution and Cancer, ICR London) — natural history, in-utero origin and the prevention question
also called Greaves Lab · Institute of Cancer Research London · Centre for Evolution and Cancer · delayed infection hypothesis
Greaves showed that childhood leukemia usually starts before birth and argues that the common form of it may be preventable by getting the infant immune system the microbial exposure it evolved to expect. His framework is why anyone thinks about prevention at all.
The same thing, for a scientist
Mel Greaves directs the Centre for Evolution and Cancer at the Institute of Cancer Research, London. His body of work established the two-step natural history of childhood ALL: an in-utero initiating event (fusion gene formation or hyperdiploidy) creating a covert preleukemic clone, followed in a small minority by postnatal secondary genetic change, largely RAG- and AID-driven copy-number alterations, that converts it to overt disease. The 'delayed infection' causal model, and the gut-microbiome-immaturity extension of it, frame childhood ALL as in principle preventable. Crucially for this project, Greaves is explicit that infant KMT2A-rearranged ALL is the exception to his own model.
What he owns that the Iqbal Lab does not2 deeper
The natural-history framework and the monozygotic twin and neonatal blood-spot evidence behind it. The claim that these fusions arise in utero is Greaves' territory, built on concordant twin studies and archived Guthrie cards 1. Any statement in this knowledge graph about prenatal origin ultimately traces to that work.
A prevention agenda with institutional backing. The 'Can we prevent childhood leukaemia?' programme is a distinct research direction with its own funding logic 2, and the microbiome-immaturity work is its current mechanistic arm 3. Nobody else has built a prevention narrative for a paediatric leukaemia.
Epidemiological reach. Greaves' arguments rest on incidence patterns, socioeconomic gradients, caesarean-section and breastfeeding associations, and cohort data. That is a different discipline from anything in this lab.
The critical caveat: his model is mostly not about our disease
This is the single most important thing to record honestly. The causal model Greaves proposes is built primarily around ETV6::RUNX1 and hyperdiploid childhood ALL, the peak-age-2-to-5 form, where the preleukemic clone is common (roughly 1% of newborns carry ETV6::RUNX1) and conversion is rare and infection-triggered 1.
Infant KMT2A-rearranged ALL does not fit. Its latency is months, not years. Its genome is nearly silent, so there is no obvious RAG/AID second-hit machinery to invoke. Its incidence peaks in the first year, before the delayed-infection mechanism could plausibly operate. Greaves' own review classifies the in-utero initiation step as general but treats the postnatal-infection conversion step as specific to the later-peak subtypes 1.
So the partnering angle is narrow and precise: Greaves is the right person to ask why infant KMT2A-r is the exception, not to ask for a prevention strategy for it. That question - what makes one fusion need a decade and a second hit while another needs neither - is genuinely open and is a good joint project.
Collaborator or competitor?
Collaborator, with essentially zero overlap. He works on causation and prevention; the lab works on druggable interfaces in established disease. There is no contested ground.
The practical value is different from data or samples: framing and credibility. A prevention or natural-history angle on a grant, endorsed by the person who defined the field, changes how a funder reads it. The reciprocal offer is mechanistic: the graph already contains detailed fusion-formation biology (topoisomerase II poisons, R-loops, 3D genome proximity, alt-EJ) that his epidemiological model needs but does not itself generate.
| group | Centre for Evolution and Cancer (Mel Greaves); Anthony M. Ford, Ioannis Peppas, Caroline Furness |
|---|---|
| institution | The Institute of Cancer Research, London, UK (affiliation as printed on PubMed-indexed papers through 2023) |
| focus | Natural history and causation of childhood acute lymphoblastic leukaemia; in-utero origin of fusion genes; monozygotic twin and neonatal blood-spot evidence; delayed-infection and gut-microbiome hypotheses; prevention |
| partnering angle | Not a data or sample partner. Approach with one precise question: why is infant KMT2A-rearranged ALL the exception to his own two-step model - short latency, silent genome, first-year peak, no plausible delayed-infection trigger? Offer the graph's mechanistic fusion-formation content (TOP2 poisons, R-loops, 3D proximity, alt-EJ) that his epidemiological framework needs but does not generate. The return is framing and credibility on the natural-history and prevention side of a grant, not reagents. |
| relationship | natural collaborator, zero overlap, limited to the aetiology question |
key papers — the numbers behind the chart
[
{
"title": "A causal mechanism for childhood acute lymphoblastic leukaemia",
"year": 2018,
"venue": "Nature Reviews Cancer",
"pmid": "29784935",
"doi": "10.1038/s41568-018-0015-6"
},
{
"title": "Can we prevent childhood Leukaemia?",
"year": 2021,
"venue": "Leukemia",
"pmid": "33833382",
"doi": "10.1038/s41375-021-01211-7"
},
{
"title": "Gut microbiome immaturity and childhood acute lymphoblastic leukaemia",
"year": 2023,
"venue": "Nature Reviews Cancer",
"pmid": "37280427",
"doi": "10.1038/s41568-023-00584-4"
},
{
"title": "Acute lymphoblastic leukaemia (Lancet seminar)",
"year": 2013,
"venue": "Lancet",
"pmid": "23523389",
"doi": "10.1016/S0140-6736(12)62187-4"
}
]leverage they hold
[ "the in-utero origin evidence base (twin studies, neonatal blood spots)", "the prevention research agenda and its funding narrative", "epidemiological cohort access and methods" ]
- If the delayed-infection and microbiome models do not apply to infant KMT2A-r disease, borrowing their rhetoric on a grant is a credibility risk with any reviewer who knows the field.
- A prevention narrative can crowd out a treatment narrative in funder attention while being untestable on any realistic timescale.
- The prenatal-origin result cuts against the lab in one direction: if the transforming event is complete before birth, interventions aimed at the fusion-formation step have no clinical window in an already-diagnosed infant.
- Why does infant KMT2A-rearranged ALL need no apparent second hit and no postnatal latency when ETV6::RUNX1 needs both? Answering this would either extend Greaves' model or show it has a hard boundary, and it is the most interesting question the two groups could ask jointly.
- What is the true population prevalence of KMT2A rearrangements in newborns? If it is as high relative to disease incidence as for ETV6::RUNX1, then most KMT2A breaks are silent and a second event must exist after all.
- Is the prevention framing usable for infant leukaemia at all, given the exposure window is entirely prenatal and the only identified modifiable exposures (topoisomerase-II-active dietary flavonoids, some drugs) are weakly evidenced?
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.review
Greaves' causal model holds that childhood ALL evolves in two steps: in-utero initiation by fusion gene formation or hyperdiploidy generating a covert preleukemic clone, then in a small fraction postnatal secondary genetic change (primarily RAG- and AID-driven copy-number alterations in ETV6::RUNX1 ALL) driving conversion to overt leukaemia, with delayed microbial exposure in early life as the modifier.
A causal mechanism for childhood acute lymphoblastic leukaemia · Greaves M · Nature Reviews Cancer 2018What this evidence cannot show
A single-author review presenting a synthesis the author has advocated for three decades; the in-utero initiation step is strongly supported by twin and blood-spot data, but the delayed-infection conversion step remains epidemiological and correlative with no interventional test. The model is built on ETV6::RUNX1 and hyperdiploid ALL and does not straightforwardly apply to infant KMT2A-rearranged disease, which has a much shorter latency and a near-silent genome.
Evidence that argues against this (1)
- A rival infective model of the same epidemiology has been defended for three decades and makes opposite predictions. Kinlen's population-mixing hypothesis holds that childhood leukaemia excesses follow the introduction of a specific, as-yet-unidentified infectious agent into susceptible (previously unexposed, rural or isolated) populations, rather than following a DEFICIT of early microbial exposure; his 2012 meta-analysis of population-mixing studies reports consistent excesses in the mixed populations. Greaves' delayed-infection model predicts protection from early exposure; Kinlen's predicts harm from a novel exposure at any age. Both fit the day-care and population-mixing data, and neither is established. Kinlen LJ 2012
- E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.review
Greaves and colleagues have set out an explicit research agenda arguing childhood leukaemia may be a preventable cancer.
Can we prevent childhood Leukaemia? · Greaves M, Cazzaniga V, Ford A · Leukemia 2021What this evidence cannot show
An advocacy review, not a trial. No prevention intervention for childhood ALL has been tested prospectively, and the feasibility of such a trial (very large n, long follow-up, rare endpoint) is the main unaddressed obstacle.
Evidence that argues against this (1)
- "Potentially preventable" is a research agenda, which is a statement of intent rather than a finding. The delayed-infection hypothesis it rests on is supported by epidemiological association and has no interventional test, and it was developed for childhood B-ALL generally — a disease with a postnatal second hit — whereas infant KMT2A-rearranged leukaemia is largely complete before birth. Blood Cells, Molecules and Diseases 2010
- E3weak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.review
The group's current mechanistic arm proposes that children with ALL have a delayed-maturation gut microbiome at diagnosis, with a consistent deficit in short-chain-fatty-acid-producing taxa that could promote dysregulated immune responses and increase the risk of preleukemic clone transformation on infectious challenge.
Gut microbiome immaturity and childhood acute lymphoblastic leukaemia · Peppas I, Ford AM, Furness CL, Greaves MF · Nature Reviews Cancer 2023What this evidence cannot show
A review mapping emerging evidence, and the authors say so. Microbiome composition at diagnosis is heavily confounded by illness, antibiotics and diet in the weeks before presentation, so reverse causation is not excluded. No causal test exists. This applies least of all to infant leukaemia, where the disease presents before the gut microbiome has matured in any child.
Evidence that argues against this (1)
- A microbiome measured AT DIAGNOSIS is measured after weeks of symptoms, often after antibiotics, and in children whose feeding and activity have already changed. Short-chain-fatty-acid-producing taxa are among the first to fall under antibiotic exposure, so the deficit has a mundane candidate explanation that a cross-sectional comparison cannot exclude. Statistics in Medicine 2006
- E4established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.review
Greaves co-authored the standard Lancet seminar on acute lymphoblastic leukaemia with Hiroto Inaba and Charles Mullighan, which notes that prognosis remains poor in infants and adults despite roughly 90% paediatric survival overall.
Acute lymphoblastic leukaemia · Inaba H, Greaves M, Mullighan CG · Lancet 2013What this evidence cannot show
A 2013 seminar; survival figures and the genomic subtype landscape have both moved substantially since, so use it for framing and for the co-authorship link to St. Jude (Inaba, Mullighan) rather than for current numbers.
Evidence that argues against this (1)
- A seminar is a teaching synthesis and its numbers are the field's headline ones. "Roughly 90% paediatric survival overall" is exactly the figure that hides this graph's population: infant KMT2A-rearranged survival sits far below it, and quoting the aggregate alongside "prognosis remains poor in infants" understates how far apart the two are. Leukemia 2021
natural history and the in-utero origin of childhood leukemia
in-utero origin framework; explicitly notes infant KMT2A-r as the exception to his two-step causal model
shared prenatal-origin and preleukemic-clone research agenda, approached epidemiologically versus experimentally
- supports →Open ground: the questions in this graph that nobody in the landscape appears to be working on
his own model excludes infant KMT2A-r, which is itself a piece of open ground: why does this fusion need neither latency nor a second hit?