KMT2A::MLLT3 / t(9;11): the best-known partner prognosis claim, and the three cohorts that did not reproduce it
also called t(9;11) favourable prognosis · MLL-AF9 prognosis · Rubnitz 2002 and its non-replication
For twenty years the textbook said that one partner, AF9, carried a better outlook in myeloid leukemia. When much larger studies looked, the advantage was not there.
The same thing, for a scientist
The favourable prognosis of t(9;11)(p22;q23)/KMT2A::MLLT3 in AML originated in a single-institution series of 298 children in which the 15 patients with t(9;11) had EFS 66% (+/-15) against 25.9% (+/-11.2) for the other 27 MLL-rearranged cases (P=.004), with t(9;11) the only independent favourable genetic factor besides low presenting WBC (relative risk of relapse 0.32, 95% CI 0.16-0.64, P=.002). It did not survive scale. An international retrospective study of 756 children with central karyotype review explicitly could not confirm it. The HARMONY adult AML database (n=205) found OS similar across translocations including t(9;11), P=0.756. In the I-BFM childhood classification MLLT3 is placed in the **non**-high-risk group, which is the weaker claim that the data support: not favourable, merely not one of the five adverse partners. In non-infant childhood ALL, t(9;11) T-ALL was the **worst** reported cell (5-year EFS 41.7% +/-17.3, n=9). MLLT3 is also the fusion from which most of this graph's mechanistic evidence is imported.
Why this entry belongs in a menin beat
A partner-provenance audit of this graph found that a majority of the load-bearing mechanistic experiments behind the menin-interface bet were performed in MLL-AF9 systems rather than in KMT2A::AFF1. If the graph is going to import mechanism from MLLT3, it should know what MLLT3 does clinically. The answer is: the one thing everyone believes about it clinically is not supported 1, 2, 3. That is a caution about importing anything else from the same fusion without checking.
What survives
The defensible residue is that MLLT3 is not among the adverse partners. It sits in the I-BFM non-high-risk group 4 and the HARMONY null 2 is consistent with that. What does not survive is 'favourable'. The distinction matters for trial design: a non-adverse partner is a legitimate control arm; a favourable partner is a group you should hesitate to expose to a new agent.
The morphology confound in the original claim
In the founding series, t(9;11) was essentially coextensive with M5 morphology - 21 of 23 cases - and the favourable-outcome analysis was repeated within M5 patients, where 21 t(9;11) cases had 71.1% (+/-11) against 25.8% (+/-7.9) for 42 others 1. That within-M5 comparison is the strongest form of the original argument and it still rests on 21 patients at one institution. It is a good example of a subgroup claim that was internally careful and externally wrong.
| metric | event-free survival for t(9;11)/KMT2A::MLLT3 across cohorts |
|---|---|
| value | 66 |
| unit | % (founding single-institution claim, +/-15, n=15) |
| population | children and adults with KMT2A::MLLT3 leukaemia |
| region | multiple |
| year | 2024 |
| prespecified | no - all comparisons retrospective subgroups |
| source | PMIDs 11981001, 19528532, 38965370, 36256911, 36996387 |
replication record — the numbers behind the chart
[
{
"cohort": "single institution, 298 children with AML",
"n_t911": 15,
"result": "EFS 66% vs 25.9%, P=.004",
"pmid": "11981001",
"verdict": "origin",
"title": "Favorable impact of the t(9;11) in childhood acute myeloid leukemia.",
"authors": "Rubnitz JE, Raimondi SC, Tong X, et al.",
"venue": "J Clin Oncol",
"year": 2002,
"title_provenance": "Title/authors/venue backfilled from PubMed esummary on 2026-09-20 because this citation stored a bare PMID with no title, which made it unscoreable by verify-citations.py. BACKFILLING IS NOT VERIFICATION: it confirms the PMID resolves to a real paper, not that this is the right paper for the claim. That still requires reading."
},
{
"cohort": "international retrospective, 756 children with 11q23 AML",
"result": "favourable prognosis not confirmed",
"pmid": "19528532",
"verdict": "non-replication",
"title": "Novel prognostic subgroups in childhood 11q23/MLL-rearranged acute myeloid leukemia: results of an international retrospective study.",
"authors": "Balgobind BV, Raimondi SC, Harbott J, et al.",
"venue": "Blood",
"year": 2009,
"title_provenance": "Title/authors/venue backfilled from PubMed esummary on 2026-09-20 because this citation stored a bare PMID with no title, which made it unscoreable by verify-citations.py. BACKFILLING IS NOT VERIFICATION: it confirms the PMID resolves to a real paper, not that this is the right paper for the claim. That still requires reading."
},
{
"cohort": "HARMONY adult AML, 205 KMT2A-r",
"result": "OS similar across translocations, t(9;11) P=0.756",
"pmid": "38965370",
"verdict": "non-replication",
"title": "Rearrangements involving 11q23.3/KMT2A in adult AML: mutational landscape and prognostic implications - a HARMONY study.",
"authors": "Hernández-Sánchez A, González T, Sobas M, et al.",
"venue": "Leukemia",
"year": 2024,
"title_provenance": "Title/authors/venue backfilled from PubMed esummary on 2026-09-20 because this citation stored a bare PMID with no title, which made it unscoreable by verify-citations.py. BACKFILLING IS NOT VERIFICATION: it confirms the PMID resolves to a real paper, not that this is the right paper for the claim. That still requires reading."
},
{
"cohort": "Ponte-di-Legno non-infant ALL, 629 children",
"n_t911": 76,
"result": "t(9;11) T-ALL 5-year EFS 41.7% (n=9), the lowest reported cell",
"pmid": "36256911",
"verdict": "opposite direction in T-lineage",
"title": "Outcomes of Childhood Noninfant Acute Lymphoblastic Leukemia With 11q23/KMT2A Rearrangements in a Modern Therapy Era: A Retrospective International Study.",
"authors": "Attarbaschi A, Möricke A, Harrison CJ, et al.",
"venue": "J Clin Oncol",
"year": 2023,
"title_provenance": "Title/authors/venue backfilled from PubMed esummary on 2026-09-20 because this citation stored a bare PMID with no title, which made it unscoreable by verify-citations.py. BACKFILLING IS NOT VERIFICATION: it confirms the PMID resolves to a real paper, not that this is the right paper for the claim. That still requires reading."
},
{
"cohort": "I-BFM childhood AML, 1,130",
"result": "MLLT3 assigned non-high-risk, not favourable",
"pmid": "36996387",
"verdict": "downgraded",
"title": "Measurable Residual Disease and Fusion Partner Independently Predict Survival and Relapse Risk in Childhood KMT2A-Rearranged Acute Myeloid Leukemia: A Study by the International Berlin-Frankfurt-Münster Study Group.",
"authors": "van Weelderen RE, Klein K, Harrison CJ, et al.",
"venue": "J Clin Oncol",
"year": 2023,
"title_provenance": "Title/authors/venue backfilled from PubMed esummary on 2026-09-20 because this citation stored a bare PMID with no title, which made it unscoreable by verify-citations.py. BACKFILLING IS NOT VERIFICATION: it confirms the PMID resolves to a real paper, not that this is the right paper for the claim. That still requires reading."
}
]- This graph imports most of its mechanistic evidence for the menin interface from MLL-AF9 systems. The clinical claim most associated with that same fusion did not replicate, which is a direct reason to distrust unexamined transfer from MLLT3 to AFF1.
- If MLLT3 is not favourable, then trials that excluded or de-intensified t(9;11) patients on the strength of the original claim may have undertreated them for two decades.
- Is there any modern cohort in which t(9;11) is favourable after adjustment for morphology, age and comutation? If not, the textbook statement should be retired explicitly rather than quietly downgraded.
- Does the graph's reliance on MLL-AF9 model systems import a fusion whose clinical behaviour is the least well characterised of the common partners?
- Was the original favourable signal a treatment-era effect - t(9;11)/M5 disease responding particularly well to the cytarabine-heavy regimens of that period?
- E1moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
The t(9;11) favourable-prognosis claim originates in 298 children with primary AML at one institution: of 42 with MLL rearrangements, the 15 with t(9;11) had EFS 66% (+/-15) versus 25.9% (+/-11.2) for the other 27 (P=.004), and within M5 morphology 21 t(9;11) cases had 71.1% (+/-11) versus 25.8% (+/-7.9) for 42 others (P=.0004); t(9;11) was an independent favourable factor (relative risk of relapse 0.32, 95% CI 0.16-0.64, P=.002).
Favorable impact of the t(9;11) in childhood acute myeloid leukemia. · Rubnitz JE, Raimondi SC, et al. · J Clin Oncol 2002What this evidence cannot show
Single institution, four consecutive trials, molecular studies performed on 152 of 298 cases so ascertainment is incomplete. Fifteen t(9;11) patients drive the primary comparison. Karyotype-based partner assignment. t(9;11) is nearly coextensive with M5 morphology in this cohort, so partner and morphology cannot be separated. Era predates modern risk-adapted therapy.
Evidence that argues against this (1)
- An international retrospective study of 756 children with 11q23/MLL-rearranged AML and central karyotype review explicitly states that it could not confirm the favourable prognosis of the t(9;11)(p22;q23) subgroup. Balgobind BV et al. 2009
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
In 205 adults with KMT2A-rearranged AML from the HARMONY database, overall survival was similar across the different translocations, including t(9;11)(p21.3;q23.3)/KMT2A::MLLT3, P=0.756.
Rearrangements involving 11q23.3/KMT2A in adult AML: mutational landscape and prognostic implications - a HARMONY study. · Hernández-Sánchez A, González T, et al. · Leukemia 2024What this evidence cannot show
Adult cohort, so a paediatric favourable effect is not strictly refuted by it. Retrospective and assembled across studies with heterogeneous treatment; per-translocation n not given in the abstract, so the null has unstated power.
Evidence that argues against this (1)
- Two smaller adult cohorts do place MLLT3 on the favourable side: a 125-patient study grouped MLLT3/ELL together in a favourable tier with 3-year OS 78.1% versus 34.9% for other KMT2A-r (P=0.044), though MLLT3 and ELL are not separated there. Chen L et al. 2025
- E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
In non-infant childhood KMT2A-r ALL, t(9;11)-positive T-ALL was the worst reported partner-by-lineage cell, 5-year EFS 41.7% (+/-17.3) in 9 patients, against a whole-cohort 69.1%.
Outcomes of Childhood Noninfant Acute Lymphoblastic Leukemia With 11q23/KMT2A Rearrangements in a Modern Therapy Era: A Retrospective International Study. · Attarbaschi A, Möricke A, et al. · J Clin Oncol 2023What this evidence cannot show
Nine patients with a standard error of 17.3 percentage points; a subgroup of a subgroup in a retrospective cohort, cited to show that MLLT3 is not favourable everywhere rather than to assert that it is adverse in T-ALL. Primary abstract re-read by the coordinator on 2026-09-19; the trial specifies t(11;19)(q23;p13.3), confirming MLLT1 rather than ELL at p13.1, so the partner-by-lineage sign flip is not a band-assignment artefact.
Evidence that argues against this (2)
- In the same cohort t(9;11) overall accounted for 76 patients (14.3%), and no adverse whole-partner figure is reported for it; the 41.7% applies only to the 9 T-lineage cases. Attarbaschi A et al. 2023
- **The symmetry matters and this entry should not trade one underpowered claim for another.** The favourable claim it refutes rests on **15 patients** at one institution; this T-ALL cell rests on **9**, with a standard error of ±17.3 percentage points — a 95% interval running roughly from 8% to 76%. Verified against the primary abstract by the coordinator on 2026-09-19: Ponte-di-Legno, 629 patients, whole-cohort 5-year EFS 69.1 ± 1.9, t(4;11) B-ALL 64.8 ± 3.0 (n=266), t(11;19) T-ALL 91.2 ± 4.9 (n=34), t(9;11) T-ALL 41.7 ± 17.3 (n=9). The two large cells in that range are precise; the t(9;11) T-ALL cell is not, and quoting it as evidence against a 15-patient claim is quoting a 9-patient one. Journal of Clinical Oncology 2023
- E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
In the I-BFM classification for childhood KMT2A-r AML, 9p22/KMT2A::MLLT3 is assigned to the non-high-risk group rather than to the high-risk group, which is the weaker claim the data support.
Measurable Residual Disease and Fusion Partner Independently Predict Survival and Relapse Risk in Childhood KMT2A-Rearranged Acute Myeloid Leukemia: A Study by the International Berlin-Frankfurt-Münster Study Group. · van Weelderen RE, Klein K, et al. · J Clin Oncol 2023What this evidence cannot show
Non-high-risk is a residual category containing every partner not on the five-member adverse list, so membership is not a positive statement about MLLT3.
Evidence that argues against this (1)
- The MD Anderson adult series concluded that all KMT2A-r should be classed as adverse regardless of partner, which would remove even the non-high-risk designation in adults. Issa GC et al. 2021
structural counterpart to the contested t(9;11) favourability claim
verified the 756-patient non-confirmation from the primary record
- related to →KMT2A::MLLT3 (MLL-AF9), t(9;11)(p21.3;q23.3)
the favourable-prognosis claim and its non-replication in three larger cohorts
- related to →MLLT3 / AF9 — a famous fusion partner that is not a dependency, and a hub domain that is
clinical profile of the partner from which this graph imports most mechanism
the best-known clinical claim about MLL-AF9 did not replicate, a reason to distrust transfer from it
worked example of an MLLT3-derived claim failing outside MLLT3