MLLT3 is the commonest partner in therapy-induced KMT2A leukaemia: 49 of 142, and 47 of those are t-AML
also called t-AML MLL-AF9 · therapy-related t(9;11) · topoisomerase-II leukaemia
Among leukaemias caused by earlier cancer treatment, AF9 is the partner that turns up most often - a third of all such cases in the largest genotyped series. It is also the only partner whose break point inside KMT2A moves in adults specifically because of that cause.
The same thing, for a scientist
Of 142 therapy-induced leukaemias (TIL) in the KMT2A recombinome - 34 t-ALL, 101 t-AML and 7 other - the dominant partner genes were **MLLT3 (n=49), AFF1 (n=20), ELL (n=14) and MLLT1 (n=10)**; within the t-AML subset alone MLLT3 accounts for **47 cases**, the largest single partner [E1]. TILs were mostly paediatric (n=52) and adult (n=88), with only one infant case [E1]. The mechanistically loaded observation is the second one: the recombinome reports that the **shift of KMT2A breakpoints towards intron 11 in the adult MLLT3 group was clearly linked to therapy-induced leukaemia**, and that the same shift was *not* seen for ELL, EPS15 or SEPTIN6 [E1]. So for this partner, and apparently for this partner alone among those tested, **breakpoint position inside KMT2A carries information about aetiology**. Any analysis that stratifies adult t(9;11) patients by KMT2A intron is partly stratifying them by prior topoisomerase-II-targeting chemotherapy, and any analysis that finds a breakpoint-outcome association in adult MLLT3 disease has to exclude that confound before calling it biology.
The counts
142 therapy-induced leukaemias in a cohort of 3,401 1:
| n | |
|---|---|
| t-ALL | 34 |
| t-AML | 101 |
| other (MPAL 2, MDS 3, NHL 1, unclassified 1) | 7 |
Dominant partners overall: MLLT3 49, AFF1 20, ELL 14, MLLT1 10 1.
Within t-AML the partner list is MLLT3 (47), ELL (13), MLLT1 (9), KMT2A-PTD (5), AFDN (3), MLLT10 (3), ARHGEF12 (2), CEP170B (2), SEPTIN9 (2) and 14 partners seen once 1. Within t-ALL the dominant partner is AFF1 (20 of 34), with MLLT3 not listed at all - so the therapy-induced MLLT3 cases are essentially all myeloid.
Age: 52 paediatric, 88 adult, 1 infant, 1 unclassified. Gender 73 female, 69 male 1.
Why the intron-11 shift matters more than the count
The recombinome tested, partner by partner, whether the KMT2A breakpoint distribution differs by age group, and found several significant shifts. For MLLT3 in adults the shift towards KMT2A intron 11 was clearly linked to therapy-induced leukaemia, and the same was explicitly not true for ELL, EPS15 or SEPTIN6 1.
Three consequences.
- A confound for breakpoint-outcome studies. Adult t(9;11) patients with an intron-11 break are enriched for prior cytotoxic therapy, which independently predicts worse outcome. A breakpoint-outcome association in that group is not interpretable without an aetiology covariate, and this graph's
absence-of-testingrule applies: nobody has reported adjusting for it. - A confound for the reciprocal question. This graph records that breakpoints upstream of KMT2A exon 11 keep PHD1 structurally intact in the reciprocal fusion protein, while breaks within or downstream of exon 11 do not 1. So a therapy-induced adult t(9;11) is more likely than a de novo one to produce a reciprocal product with an altered PHD1 - a partner-by-aetiology-by-reciprocal interaction that nobody has examined.
- It is a mechanism claim about DNA, not about protein. Topoisomerase-II poisons cleave preferentially in particular chromatin contexts; the intron-11 preference in TIL is most simply read as where the drug cuts, not as selection on the resulting protein. That reading is available and is the null this observation should be tested against.
| metric | fusion-partner distribution in therapy-induced KMT2A leukaemia |
|---|---|
| value | 49 |
| unit | therapy-induced KMT2A::MLLT3 cases |
| population | therapy-induced acute leukaemia with a KMT2A rearrangement |
| region | international referral archive |
| year | 2023 |
| source | PMID 37019990, read from PMC10169636 on 2026-09-20 |
| til total | 142 |
| til tall | 34 |
| til taml | 101 |
| til other | 7 |
| mllt3 taml n | 47 |
| adult mllt3 intron11 shift | clearly linked to therapy-induced leukaemia; not seen for ELL, EPS15 or SEPTIN6 |
| baseline comparison | MLLT3 is 34.5% of TIL cases and 30.4% of all AML cases, so the count is close to baseline; the intron-11 shift is the part not explained by frequency |
| aetiology adjusted breakpoint outcome analysis exists | false |
til partners — the numbers behind the chart
[
{
"partner": "MLLT3",
"n": 49
},
{
"partner": "AFF1",
"n": 20
},
{
"partner": "ELL",
"n": 14
},
{
"partner": "MLLT1",
"n": 10
}
]til age — the numbers behind the chart
{
"infant": 1,
"paediatric": 52,
"adult": 88,
"unclassified": 1
}til gender
{
"female": 73,
"male": 69
}- A therapy-induced t(9;11) patient has already had cytotoxic therapy and often has a reduced marrow reserve, so the toxicity profile of any new agent is being read in a population selected for prior marrow injury - and this is the single largest partner group in that population.
- If breakpoint position in adult t(9;11) is partly an aetiology marker, a biomarker built on breakpoint position will carry prior-therapy status with it and will look prognostic for reasons that have nothing to do with the fusion protein.
- What is the KMT2A intron distribution for adult de novo t(9;11) against adult therapy-induced t(9;11), with numbers? The text states the linkage; the table behind it is Supplementary Table S2 and has not been read by this project.
- Does the reciprocal product differ between therapy-induced and de novo t(9;11) as the PHD1 argument predicts, and has anyone looked?
- Is the intron-11 preference in therapy-induced MLLT3 disease a property of where topoisomerase II cuts, or selection on the resulting protein? A cleavage-site map in treated cells would separate these.
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Among 142 therapy-induced leukaemias in the KMT2A recombinome (34 t-ALL, 101 t-AML, 7 other), the dominant partner genes were MLLT3 (n=49), AFF1 (n=20), ELL (n=14) and MLLT1 (n=10), with MLLT3 accounting for 47 of the t-AML cases; the cases were 52 paediatric, 88 adult and 1 infant; and the shift of KMT2A breakpoints towards intron 11 in the adult MLLT3 group was clearly linked to therapy-induced leukaemia, which was not the case for ELL, EPS15 or SEPTIN6.
The KMT2A recombinome of acute leukemias in 2023. · Meyer C, Larghero P, Almeida Lopes B, et al. · Leukemia 2023What this evidence cannot show
Read in full text from the PMC HTML page (PMC10169636) on 2026-09-20. Therapy-induced status is as reported by the submitting centre, not independently adjudicated, and prior-therapy details (agent, dose, latency) are not given. The intron-11 linkage is stated qualitatively in the text; the supporting statistics are in Supplementary Table S2, which this entry has not read. A referral archive over-represents cases sent for breakpoint characterisation, and therapy-induced cases may be preferentially characterised.
Evidence that argues against this (1)
- MLLT3 is also the commonest AML partner overall (30.4% of 1,116 AML cases), so its dominance among therapy-induced cases (49 of 142, 34.5%) is close to what its baseline AML share predicts. The count alone is therefore not evidence of any special affinity between MLLT3 and topoisomerase-II-induced breakage; the intron-11 shift is the part of this observation that is not explained by baseline frequency. Meyer C et al. 2023
- E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
The earlier instalment of the same archive, covering 2,345 acute leukaemia patients, reported an age-dependent breakpoint shift within MLL, with breakpoints in MLL intron 11 associated with acute lymphoblastic leukaemia and younger patients while breakpoints in intron 9 predominate in AML or older patients, and concluded that the molecular characterisation of MLL breakpoints suggests different aetiologies in the different age groups.
The MLL recombinome of acute leukemias in 2017. · Meyer C, Burmeister T, Gröger D, et al. · Leukemia 2018What this evidence cannot show
An association between breakpoint position, age and lineage is not a mechanism, and the direction of this general age shift (intron 11 with younger, lymphoid patients) runs opposite to the MLLT3-specific adult intron-11 shift reported later, so the two observations are not the same phenomenon and should not be merged.
Evidence that argues against this (1)
- Because the general age-related shift and the MLLT3 therapy-induced shift point in opposite directions with respect to intron 11, an adult t(9;11) intron-11 break is simultaneously unusual for its age band and typical for its aetiology - which means intron position cannot be read as a clean proxy for either. Meyer C et al. 2023
49 of 142 therapy-induced KMT2A leukaemias
the therapy-induced slice of the same archive
the KMT2A-side confound behind the partner-side distribution
a cleavage-geography model is also a statement about therapy-induced disease
the exposure-dependent breakpoint shift