Seven KMT2A-rearranged sarcomas retain the menin-binding motif, three of them carry the complete leukaemic N-terminal module, and all sixteen partner names were in one unopened supplementary table
also called the 16 non-recurrent sarcoma partners · sarcoma KMT2A fusions that keep the MBM · KTM2A typo
Last cycle this project concluded that the KMT2A sarcomas all throw away the piece of the protein that menin drugs grip, based on the three tumour types anyone had examined. The full list of sixteen was sitting in a free spreadsheet attached to the original paper, and seven of them keep that piece, three of them keep the entire machinery that drives the leukaemia.
The same thing, for a scientist
**The gap that was named and is now closed.** `ana-kmt2a-sarcoma-partner-census-and-what-was-read-and-dropped` recorded that *the individual identities of the 16 non-recurrent partners are in a supplementary table that was not opened for this entry*, and cycle 51's log concluded that **exon 1 is absent from every arm with a published breakpoint**. The table is Supplemental Table 2 of PMID 32461620, a free 16 KB xlsx at media.springernature.com, and it gives exon-level breakpoints for all 18 non-YAP1 cases [E1]. **Seven of the sixteen retain the menin-binding motif, and the cycle-51 conclusion was an artefact of which arms had been examined.** The 16 non-recurrent partners split **9 with KMT2A 3-prime and 7 with KMT2A 5-prime** (the census had recorded 8 and 7, summing to 15 of 16). The seven 5-prime cases break at KMT2A exons **3, 8, 8, 9, 15, 20 and 22** — every one downstream of exon 1 — so all seven retain **exon 1 and the MBM at aa 6-25**. Exon 1 is absent from the three arms that had been looked at (YAP1, VIM, LDB1) and present in seven that had not. The partners are **CPT1A, DOCK10, C2CD3, ELP4, AHDC1, TMPRSS13 and RALGAPA2**, in a malignant peripheral nerve sheath tumour, an undifferentiated pleomorphic sarcoma, a second MPNST, an angiosarcoma, a leiomyosarcoma, a high-grade endometrial stromal sarcoma and a conventional osteosarcoma. **Three of them are structurally leukaemia-type fusions.** `KMT2A exon 8 :: DOCK10`, `KMT2A exon 8 :: C2CD3` and `KMT2A exon 9 :: ELP4` break **inside the leukaemic breakpoint cluster region (exons 8-14)**, so they retain the complete leukaemic N-terminal module: MBM, IBM1, IBM2, all three AT-hooks and the CXXC domain. Nothing about the retained KMT2A portion distinguishes these from an infant-leukaemia fusion; only the partner and the tissue differ [E1][E2]. **And the inverse case this project has been looking for exists.** `KMT2A exon 3 :: CPT1A` retains the MBM, both LEDGF IBMs and all three AT-hooks but breaks **before** the CXXC domain in exons 5-6. It is **tether without address**. YAP1::KMT2A::YAP1 is **address without tether**. Every sarcoma arm examined until now confounded the two, and the pair separates them in the same disease and the same cohort. **What this does to the menin framework.** These seven are lesions where the polarity hypothesis and `hyp-menin-dependency-tracks-available-mbm` **both predict menin sensitivity** — in a lineage where menin is a tumour suppressor, where the only non-haematologic menin trial was terminated for lack of efficacy, and where no menin measurement of any kind has ever been made. **If any of the seven is menin-insensitive, the 'if and only if' in the MBM statement fails.** These are not a rare entity needing a bespoke model: they are ordinary leiomyosarcoma, MPNST, angiosarcoma and osteosarcoma diagnoses. **A retrieval trap worth the registry.** The source table spells the index gene **KTM2A** in two rows — S5 (PHLDB1) and S17 (TMPRSS13). A reader grepping the table for `KMT2A` silently loses two of eighteen cases, one of them an MBM-retaining 5-prime arm.
| script | analyses/sarcoma-nonrecurrent-partners/run.py |
|---|---|
| results | analyses/sarcoma-nonrecurrent-partners/results.json |
| source table | Supplemental Table 2 of PMID 32461620 (41379_2020_582_MOESM2_ESM.xlsx, 16 KB, free) |
| cases parsed | 18 |
| kmt2a 5prime | 7 |
| kmt2a 3prime | 11 |
| non recurrent split corrected | 9 with KMT2A 3-prime and 7 with KMT2A 5-prime, summing to 16. The census had recorded 8 and 7, which sums to 15. |
| leukaemic bcr cases | KMT2A exon 8 :: DOCK10, KMT2A exon 8 :: C2CD3, KMT2A exon 9 :: ELP4 -- breakpoints inside the leukaemic BCR (exons 8-14), retaining MBM + IBM1 + IBM2 + AT-hooks + CXXC |
| tether without address | KMT2A exon 3 :: CPT1A -- MBM, IBM1, IBM2 and all three AT-hooks retained, CXXC (exons 5-6) lost. The inverse of YAP1::KMT2A::YAP1. |
| what it corrects | Cycle 51's log: 'Exon 1 is absent from every arm with a published breakpoint.' True of YAP1, VIM and LDB1; false for 7 of the 16 non-recurrent arms. Also corrects the census's 8/7 split to 9/7, and the census's record that these partners have no published breakpoints. |
| built in positive control | The parser's independent domain call for VIM::KMT2A (S1, S2) reproduces this graph's hand-curated fusion-vim-kmt2a record exactly: KMT2A exon 2 onward, MBM and IBM1 lost, IBM2 and CXXC retained. The classifier was not tuned to that entry. |
| retrieval trap | The source spells the index gene KTM2A in rows S5 and S17. A grep for KMT2A loses two of eighteen cases, one of them an MBM-retaining 5-prime arm. |
| the falsifier this creates | A menin-inhibitor or MEN1-knockdown measurement on KMT2A exon 8::DOCK10 or KMT2A exon 9::ELP4. Both hypotheses predict sensitivity; a negative result breaks the MBM hypothesis's 'if and only if' and would be the first natural counter-example to it. NARROWED 2026-09-22 from 'any of the three leukaemic-BCR sarcomas': the third, KMT2A exon 8::C2CD3, is predicted OUT OF FRAME and so is the weakest of the three for this test -- see data.frame_predictions_2026_09_22. |
| shown for fusion | false |
| absence class | unclassified-not-checked |
| recombinome cross check | None of the 7 MBM-retaining sarcoma partners is a known leukaemia KMT2A partner. Query (GENE[tiab]) AND (KMT2A[tiab] OR MLL[tiab]) AND (leukemia[tiab] OR leukaemia[tiab] OR fusion[tiab]) returns 0 records for DOCK10, C2CD3, ELP4, AHDC1, TMPRSS13 and RALGAPA2, and 1 for CPT1A -- PMID 39703843, an ARMH1 paediatric-AML fatty-acid paper in which CPT1A is a metabolic gene and not a fusion partner. Controls on the same query shape: AFF1 128, MLLT3 162, MLLT1 55, so the shape fires. The sarcoma and leukaemia partner sets remain disjoint, which is the same result cycle 51 got for YAP1, VIM, CBX6, PYGO1 and LDB1: the retained KMT2A module can be leukaemia-like while the partner never is. |
| recall caveat on that zero | The zeros are bounded but not airtight, and the control that showed why is worth keeping. SEPTIN9 was used as a fourth positive control and returned only 1 record despite being a genuine KMT2A partner with an entry in this graph -- because the literature spells it SEPT9, which returns 19. So this query shape under-recalls any partner whose literature predates an HGNC rename, and a 0 for a gene with a deprecated alias would be a false zero. None of the seven was renamed by HGNC, so the zeros stand; the hazard is now registry row 109. |
mbm retaining partners
[ "CPT1A (KMT2A exons 1-3, MPNST)", "DOCK10 (KMT2A exons 1-8, undifferentiated pleomorphic sarcoma)", "C2CD3 (KMT2A exons 1-8, MPNST)", "ELP4 (KMT2A exons 1-9, angiosarcoma)", "AHDC1 (KMT2A exons 1-15, leiomyosarcoma)", "TMPRSS13 (KMT2A exons 1-20, high grade endometrial stromal sarcoma)", "RALGAPA2 (KMT2A exons 1-22, conventional osteosarcoma)" ]
frame predictions 2026 09 22
{
"what_changed": "Q96 computed the reading frame of all seven junctions from RefSeq exon coordinates, which this entry had not done. Four are predicted IN FRAME and TWO are predicted OUT OF FRAME, so only four of the seven are predicted to make a chimeric protein at all.",
"in_frame": {
"CPT1A": "exon 15 starts 1,740 nt after the ATG, 1740 mod 3 = 0",
"DOCK10": "exon 2 starts 123 nt after the ATG, 123 mod 3 = 0",
"ELP4": "exon 10 starts 1,143 nt after the ATG, 1143 mod 3 = 0",
"TMPRSS13": "exon 2 starts 21 nt after the ATG, 21 mod 3 = 0"
},
"out_of_frame": {
"C2CD3": "exon 24 begins 4,603 nt after the ATG, 4603 mod 3 = 1",
"RALGAPA2": "exon 2 begins 106 nt after the ATG, 106 mod 3 = 1"
},
"not_applicable": {
"AHDC1": "no partner coding sequence for the frame question to apply to"
},
"all_six_recomputed_by_the_king": "2026-09-22, and all six agree with Q96's calls.",
"what_it_does_NOT_overturn": "The MBM claim survives in full, and this is the part worth being precise about. An out-of-frame junction does not disturb the KMT2A portion, which is translated in its own frame up to the junction -- so the predicted product of KMT2A exon 8::C2CD3 is a KMT2A 1-1362 fragment that STILL CONTAINS MBM 6-25, IBM1, IBM2, the AT-hooks and CXXC, followed by a short frameshifted tail. All seven arms retain the menin-binding motif in their predicted product. What two of them do not make is a CHIMERA.",
"and_it_creates_a_different_species": "A KMT2A N-terminal fragment carrying the MBM and the CpG address but no partner is not a fusion oncoprotein; it is closer to a truncated KMT2A. This graph already holds protein-kmt2a-star-truncated-product-of-the-internal-promoter as a comparator, and that is the right analogy for the two out-of-frame arms rather than the leukaemic-fusion one.",
"status": "A DERIVATION, not an observation, and graded as such by Q96 at moderate. It is settled for either case by one RT-PCR across the junction, or by the sequencing panel's own transcript model, which was never published."
}- This entry is a coordinate derivation from a supplementary table and it corrects a claim this project published one cycle ago. It should not be read as evidence that menin inhibitors have a role in sarcoma. Menin is a tumour suppressor in mesenchyme, five MEN1-syndrome sarcomas show MEN1 LOH with loss of menin expression, and the only non-haematologic menin trial with a result was terminated by the sponsor after phase 1b for not meeting pre-defined efficacy signals. The existence of an MBM-retaining KMT2A fusion in an osteosarcoma is a mechanistic fact that makes a prediction testable; it is not a treatment rationale, and this graph already carries a risk entry about a menin inhibitor being prescribed off a KMT2A gene name in a solid tumour.
- The seven 5-prime partners are singleton cases the source authors call of unclear significance. Naming them makes them auditable; it does not promote them to drivers, and any use of these names should carry that status.
- Is any of the three leukaemic-BCR sarcomas menin-inhibitor sensitive? Both hypotheses predict yes, and a negative result would be the first natural counter-example to the MBM hypothesis's 'if and only if'.
- Are the seven KMT2A 5-prime fusions expressed as proteins at all? Everything here is transcript-level, and the source authors decline to call them drivers.
- Does KMT2A exon 3 :: CPT1A activate HOXA? It is the tether-without-address case, and paired with the YAP1 arm it separates the menin tether from the CpG address in one cohort.
- Is the S17 endometrial stromal sarcoma a YWHAE-driven tumour with an incidental KMT2A event? The source's own footnote says the panel could not have detected YWHAE or FAM22.
- Do the 9 KMT2A 3-prime non-recurrent arms show the same HOXA-low phenotype as the YAP1 arm, and does the one with the most retained KMT2A (MRGPRF, exon 2 onward) behave like VIM::KMT2A?
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Supplemental Table 2 of PMID 32461620 gives exon-level breakpoints for 18 KMT2A-fusion sarcomas. Eleven have KMT2A 3-prime (including the two VIM cases) and seven have KMT2A 5-prime, with breakpoints at KMT2A exons 3, 8, 8, 9, 15, 20 and 22 to CPT1A, DOCK10, C2CD3, ELP4, AHDC1, TMPRSS13 and RALGAPA2 respectively; all seven therefore retain exon 1 and the menin-binding motif at aa 6-25, and the three breaking at exons 8, 8 and 9 fall inside the leukaemic breakpoint cluster region and retain MBM, IBM1, IBM2, all three AT-hooks and CXXC.
Pan-sarcoma genomic analysis of KMT2A rearrangements reveals distinct subtypes defined by YAP1-KMT2A-YAP1 and VIM-KMT2A fusions, Supplemental Table 2 (41379_2020_582_MOESM2_ESM.xlsx), parsed 2026-09-21 by analyses/sarcoma-nonrecurrent-partners/run.py · Massoth LR, et al. · Modern Pathology 2020What this evidence cannot show
These are exon-level breakpoints from a commercial targeted RNA panel in a cohort enriched for advanced disease, and the source authors describe the non-recurrent partners as of unclear significance, so several may be passengers rather than drivers. No fusion protein was observed for any case and no reciprocal transcripts were reported. The domain calls are coordinate derivations from the exon numbers against Ensembl ENST00000534358, not measurements. A panel that targets specific KMT2A exons can also bias which breakpoints are detectable at all, so the exon distribution is not a population estimate. The S17 row carries the source's own footnote that the panel lacks coverage for YWHAE or FAM22 rearrangements, which is the defining lesion class for high-grade endometrial stromal sarcoma, so that case's KMT2A event may be incidental to an undetected driver. PMID and DOI verified live by esummary on 2026-09-21.
Evidence that argues against this (2)
- The strongest objection is that retaining the MBM in a transcript is not the same as making a protein that engages menin, and the KMT2A 5-prime cases may not be expressed as stable fusion proteins at all. Only 2 of 34 cases in this screen had a recurrent partner with any functional follow-up; the non-recurrent group has none, and the authors themselves decline to call them drivers. So 'seven sarcomas retain the MBM' is a statement about seven transcripts.
- This cuts against the sarcoma polarity story that cycle 51 built, but it also weakens the cleanest version of the counter-argument to the menin-interface hypothesis. Cycle 51 said a KMT2A rearrangement transforms without menin, without LEDGF and without AT-hooks; that remains true for the YAP1 and VIM arms, which are the recurrent, entity-defining ones with morphological and methylation support. The seven 5-prime cases are singletons of unclear significance, so the entity-level claim survives even though the blanket claim about 'every arm' does not.
- E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database
The leukaemic KMT2A breakpoint cluster region spans exons 8-14, and KMT2A exon 1 encodes residues 1-144 including the menin-binding motif at 6-25 and LEDGF IBM1 at 123-134, exon 2 encodes 145-168 including IBM2 at 147-152, exon 3 encodes 168-1052 including all three AT-hooks, and exons 5-6 encode 1112-1212 including the CXXC domain at 1147-1195.
What this evidence cannot show
Exon-to-residue assignment uses the canonical ENST00000534358 numbering, and published sarcoma breakpoints are reported against various RefSeq transcripts, so an exon number quoted by the source may not correspond to the same boundary in the canonical transcript. For the cases here the discrepancy cannot move the MBM call, because every 5-prime breakpoint is at exon 3 or later and exon 1 is retained under any KMT2A transcript numbering. The CXXC and AT-hook calls are more sensitive to transcript choice than the MBM call.
Evidence that argues against this (1)
- Treating 'breakpoint inside exons 8-14' as equivalent to a leukaemic fusion overstates the parallel. Leukaemic KMT2A fusions are defined by their partners as much as by the breakpoint -- partners supplying dimerisation or elongation function -- and DOCK10, C2CD3 and ELP4 are not members of any recognised KMT2A leukaemia partner class. The retained KMT2A module is leukaemia-like; the fusion as a whole is not.
one of the seven MBM-retaining 5-prime sarcoma partners; one of the three inside the leukaemic BCR
one of the seven MBM-retaining 5-prime sarcoma partners (case S16, leiomyosarcoma)
one of the seven MBM-retaining 5-prime partners and one of the three inside the leukaemic BCR
the tether-without-address case: the only one of the seven that loses CXXC while keeping the MBM
one of the seven MBM-retaining 5-prime partners and one of the three inside the leukaemic BCR
one of the seven MBM-retaining 5-prime sarcoma partners; the row the source misspells as KTM2A
one of the seven MBM-retaining 5-prime sarcoma partners, and the case where allele disruption is the better hypothesis than a chimera
CLOSES ITS NAMED GAP AND CORRECTS TWO FIGURES: that entry's own open_questions asked what the 16 non-recurrent partners are and recorded that the supplementary table was not opened. It was reachable in one request. The split is 9 with KMT2A 3-prime and 7 with 5-prime, not the 8 and 7 recorded, which summed to 15 of 16; and the group is not functionally opaque -- it holds the only KMT2A-rearranged sarcomas that retain the menin-binding motif.
CONFIRMS THE COUPLING ON SEVEN NEW CASES WHILE BREAKING THE BLANKET CLAIM: all 7 KMT2A 5-prime sarcoma arms retain exon 1 and the MBM, exactly the polarity-to-tether coupling this hypothesis asserts, so the confounding it shares with the MBM hypothesis now holds over 12 lesions with 0 disagreements. What fails is cycle 51's broader statement that exon 1 is absent from every sarcoma arm -- true of YAP1, VIM and LDB1 only.
CREATES THE FIRST NATURAL FALSIFIER FOR ITS 'IF AND ONLY IF': three sarcomas -- KMT2A exon 8::DOCK10, exon 8::C2CD3 and exon 9::ELP4 -- break inside the leukaemic breakpoint cluster region and retain MBM, both IBMs, all three AT-hooks and CXXC. This hypothesis predicts they are menin-inhibitor sensitive. They are an undifferentiated pleomorphic sarcoma, an MPNST and an angiosarcoma, in a lineage where menin is a tumour suppressor and menin inhibitors have never worked. A negative result breaks the iff.
SUPERSEDES ONE OF ITS FIELDS AND STRENGTHENS ITS CONCLUSION: that entry recorded, earlier the same cycle, that the 7 KMT2A 5-prime non-recurrent arms had no published breakpoints and were the last place a natural dissociation between polarity and MBM availability could hide. They have breakpoints, and all 7 are 5-prime WITH the MBM, so the dissociation is not there either -- 12 cases, 0 disagreements.
SUPPLIES ITS STRUCTURAL INVERSE FROM THE SAME COHORT: YKY retains CXXC and discards the menin and LEDGF tethers -- address without tether. KMT2A exon 3::CPT1A retains the MBM, both IBMs and all three AT-hooks but breaks before CXXC in exons 5-6 -- tether without address. Every sarcoma arm examined until now confounded the two functions; this pair separates them in one disease and one cohort.
USES IT AS THE INSTRUMENT: every domain call here is the source table's exon number resolved against this map (exon 1 = aa 1-144 with MBM 6-25 and IBM1 123-134; exon 2 = 145-168 with IBM2 147-152; exon 3 = 168-1052 with all three AT-hooks; exons 5-6 = 1112-1212 with CXXC 1147-1195). The MBM call is robust to transcript choice because every 5-prime breakpoint is at exon 3 or later; the CXXC and AT-hook calls are not.
RAISES THE VALUE OF CLOSING IT AND NAMES A BETTER TARGET: the absence was scoped to an entity with no MBM-retaining arm, where a negative menin result would have been unsurprising. Three sarcomas retain the complete leukaemic N-terminal module, so the same missing measurement now has a hypothesis-breaking outcome available, and the tumours are ordinary MPNST, undifferentiated pleomorphic sarcoma and angiosarcoma rather than a rare entity needing a bespoke model.
SHARPENS THE RISK RATHER THAN EASING IT: the risk entry warns against prescribing off a gene name. These seven cases are exactly the rows most likely to be misread as a rationale, because they retain the drug-relevant motif and three look leukaemia-like on paper. Nothing here is a treatment rationale -- menin is a tumour suppressor in mesenchyme, and the only non-haematologic menin trial with a result was terminated after phase 1b for not meeting pre-defined efficacy signals.
INDEPENDENTLY REPRODUCES ITS HAND-CURATED DOMAIN CALL, WHICH IS THIS ANALYSIS'S POSITIVE CONTROL: parsing S1 and S2 from the source table and resolving the exon numbers without reference to this entry returns KMT2A exon 2 onward with MBM and IBM1 lost and IBM2 and CXXC retained -- exactly what this entry records. The classifier was not tuned to it.