t(9;11) is the reciprocal-POSITIVE arm: 2 of 612 patients are spliced, against 241 of 460 for t(11;19)
also called MLL-AF9 reciprocal status · AF9-MLL · spliced KMT2A::MLLT3
For the ENL partner, half of all patients have a version of the translocation that makes only one fusion protein; for the AF9 partner, almost nobody does. So the two commonest partners also differ in whether a mirror-image second fusion exists, and nobody has used that.
The same thing, for a scientist
The KMT2A recombinome reports, for 3,401 patients, how often a rearrangement is a **spliced fusion** - a truncated KMT2A recombined 5-prime of an intact partner gene, from which only the direct transcript and **no reciprocal one** is expressed. The partner-by-partner split is extreme: **MLLT1, 241 of 460 (about 50%); MLLT3, 2 of 612 (0.33%)**; AFF1, 1 of 1,262; ELL, 10 of 119; EPS15, 12 of 51 [E1]. This is the exact mirror of the YEATS-retention asymmetry and it runs the same way round: the partner whose fusion keeps the reader is also the partner whose patients are half reciprocal-negative, and the partner whose fusion discards the reader is almost uniformly reciprocal-positive. **t(9;11) is therefore the closest thing this disease has to a pure reciprocal-positive arm**, and this graph's standing threat to the menin-interface hypothesis - that no study has ever stratified by reciprocal status - has a cleaner comparator than it has been using. The catch is that the comparison across partners is confounded by everything else that differs between AF9 and ENL, which is why the within-partner MLLT1 comparison (153 reciprocal-negative against 101 reciprocal-positive, both arms making an identical fusion protein) remains the better-controlled test and this entry is its denominator, not its replacement.
The numbers, and what a spliced fusion is
A spliced fusion in the recombinome nomenclature arises when the recombination puts a truncated KMT2A gene upstream of an intact partner gene rather than inside it. Because there is no transcription termination between them, transcription reads through and a chimeric transcript is made using the partner's own exon 1 splice acceptor. The der(9) - or whichever derivative would carry the reciprocal - does not encode a reciprocal fusion, so only one fusion protein exists in that patient 1.
| partner | spliced cases | total cases | spliced % |
|---|---|---|---|
| MLLT1 | 241 | 460 | ~52% |
| EPS15 | 12 | 51 | 24% |
| ELL | 10 | 119 | 8.4% |
| AFDN | 2 | 138 | 1.4% |
| MLLT3 | 2 | 612 | 0.33% |
| AFF1 | 1 | 1,262 | 0.08% |
275 cases across the whole cohort use this mechanism 1.
Why this is the reciprocal question's denominator
This graph's largest standing threat to the menin-interface hypothesis is that reciprocal status is stratified by nobody, and its sharpest proposed test is the MLLT1 within-partner comparison: 153 patients breaking 5-prime of MLLT1 but 3-prime of ACER1 against 101 breaking within MLLT1 intron 1, both arms producing a fusion protein that begins at MLLT1 exon 2 and therefore an identical direct product 2.
What the MLLT3 numbers add is the other half of the design. A cross-partner comparison of t(9;11) against spliced t(11;19) is not the same experiment - AF9 and ENL differ in YEATS retention, in wild-type requirement, in lineage association and in age distribution - but t(9;11) is the natural reciprocal-positive reference population against which any reciprocal-negative arm should be read, and it is enormous: 610 of 612.
The practical statement: if reciprocal status matters, KMT2A::MLLT3 is the disease where it always matters, and KMT2A::MLLT1 is the disease where it matters in half of patients. Any menin-inhibitor analysis that pools these two partners is pooling across the variable of interest.
What is not established here
Three honest limits.
- Spliced status is inferred from genomic breakpoint position, not measured as an absent transcript in that patient 1. A patient counted as reciprocal-capable has not been shown to express a reciprocal transcript.
- This graph's
concept-reciprocal-product-of-t-9-11-mllt3-kmt2arecords that the reciprocal expression frequency in t(9;11) patients is not reported anywhere. So 'reciprocal-positive' here means architecturally capable, not demonstrated. - The archive is a diagnostic referral series. Partner-specific spliced-fusion frequencies could in principle be affected by which cases get breakpoint-characterised, though the size of the MLLT1/MLLT3 gap (52% against 0.33%) makes ascertainment an implausible sole explanation.
| partner | MLLT3 |
|---|---|
| fusion | fusion-kmt2a-mllt3 |
| spliced cases | 2 |
| total cases | 612 |
| spliced PCT | 0.33 |
| reciprocal capable cases | 610 |
| comparator partner | MLLT1 |
| comparator spliced cases | 241 |
| comparator total cases | 460 |
| comparator spliced PCT | 52.4 |
| total spliced in cohort | 275 |
| cohort n | 3401 |
| reciprocal expression measured in patients | false |
| status basis | genomic breakpoint architecture, not transcript assay |
all partners — the numbers behind the chart
[
{
"partner": "MLLT1",
"spliced": 241,
"total": 460
},
{
"partner": "EPS15",
"spliced": 12,
"total": 51
},
{
"partner": "ELL",
"spliced": 10,
"total": 119
},
{
"partner": "AFDN",
"spliced": 2,
"total": 138
},
{
"partner": "MLLT3",
"spliced": 2,
"total": 612
},
{
"partner": "AFF1",
"spliced": 1,
"total": 1262
}
]- If reciprocal products contribute to menin-inhibitor response, a trial population dominated by t(9;11) (as AUGMENT-101's was, at 22.3% of the safety population and the largest single named partner) is a population in which the reciprocal is almost always present, and the inhibitor's apparent partner-agnostic efficacy is being measured largely in one reciprocal stratum.
- A therapy directed at the direct fusion alone would, in t(9;11), leave a reciprocal allele untouched in essentially every patient; in spliced t(11;19) there is nothing left to leave.
- Are the 2 spliced KMT2A::MLLT3 cases the same 2 patients who retain the YEATS domain? Both counts come from the same archive and nobody has stated whether they are the same cases.
- Does any t(9;11) patient actually express an AF9-MLL reciprocal transcript at protein level? The architecture says 610 of 612 could; nothing says any does.
- Would a menin-inhibitor response analysis that used t(9;11) as the reciprocal-positive reference and spliced t(11;19) as the reciprocal-negative reference be interpretable, or is the partner difference too large to cross?
- E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
In 3,401 KMT2A-rearranged patients, spliced fusions - in which a truncated KMT2A is recombined upstream of an intact partner gene so that only the direct fusion transcript and no reciprocal one is expressed - occur in about 50% of MLLT1 cases (241 of 460) but in only 2 of 612 MLLT3 cases, 1 of 1,262 AFF1 cases, 10 of 119 ELL cases, 12 of 51 EPS15 cases and 2 of 138 AFDN cases, for 275 cases in total.
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. Spliced status is inferred from the genomic breakpoint position relative to the partner's exon 1, not from an assay demonstrating absence of a reciprocal transcript in that patient. The cohort is a diagnostic referral archive accumulated 2003-2022, not a population sample.
Evidence that argues against this (1)
- Because the assignment is architectural rather than assayed, a patient counted as reciprocal-capable may still not express a reciprocal product - out-of-frame reciprocal fusions are common in this archive, and the recombinome itself notes that most out-of-frame reciprocal KMT2A fusions express only the 3-prime KMT2A portion (KMT2A*) from an internal promoter rather than a reciprocal fusion protein. Meyer C et al. 2023
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
In 302 t(11;19) patients, 50.7% (n=153) break 5-prime of MLLT1 but 3-prime of ACER1 and 33.4% (n=101) break within MLLT1 intron 1, and in combination 84.1% (n=254) retain the YEATS domain; in 442 t(9;11) patients 98.4% break downstream of MLLT3 exon 4 and only two (0.5%) retain the YEATS domain.
The ENL YEATS epigenetic reader domain critically links MLL-ENL to leukemic stem cell frequency in t(11;19) Leukemia. · Hu H, Saha N, Yang Y, et al. · Leukemia 2023What this evidence cannot show
Read verbatim from the PMC HTML body (PMC11246743) on 2026-09-20 after `fullTextXML` returned HTTP 500. Genomic breakpoints re-analysed from a previously published archive; no protein measured. The 153/101 split is a breakpoint classification, not a direct assay of reciprocal transcript expression.
Evidence that argues against this (1)
- This graph's `concept-the-two-mllt1-arms-do-not-carry-identical-direct-products` argues that the two MLLT1 arms are not as cleanly matched as the identical-protein framing implies, which if correct weakens the within-partner test that this entry defers to. Hu H et al. 2023
reciprocal status of the partner population
- analogous to →Half of all KMT2A::MLLT1 patients make no reciprocal product at all - a ready-made control arm of about 240 people
the mirror arm: 0.33% against 52%
610 of 612 t(9;11) patients are architecturally reciprocal-capable
supplies the reciprocal-positive reference population
cross-partner denominator for the within-partner test
same two patients, possibly