AnalysisstrongQ-usp2-maml2-second-bcr

Enumeration for the USP2 / MAML2 / second-BCR beat: what was searched, what was counted, what was rejected

also called USP2 MAML2 enumeration · second-BCR beat search ledger

Before writing anything, every relevant search was run and every hit was checked by reading its title and abstract. This records the numbers, including the searches that found nothing and the papers that looked relevant and were not.

The same thing, for a scientist

A systematic PubMed enumeration over USP2, USP8, MAML2 and the KMT2A second-breakpoint-cluster-region question, run 2026-09-19, with every hit in the two primary intersections title- and abstract-checked. **USP2: 11 raw records, 11 on-topic - an unusual case where raw equals checked.** MAML2: 21 raw records, 15 on-topic after rejecting 6. USP8: the symbol query returns only 1 record and **under-retrieves**, because the founding letter that named USP8 as a KMT2A partner has no indexed abstract. Three counted-absence results matter more than any positive: **(MLL OR KMT2A) AND USP2 AND menin returns 0**, **(MLL OR KMT2A) AND MAML2 AND menin returns 0**, and adding revumenib or MEN1 to the USP2 query also returns 0. Two retrieval traps were confirmed rather than assumed: the feared USP2/USP22 collision **does not occur** in retrieval (USP22 intersection = 1 record, no overlap), while the MAML2 salivary-gland noise is real and large (544 records for MAML2 alone, 345 of them salivary or mucoepidermoid). The retired-symbol lesson recurred twice: SEPT3 returns 1 record where SEPTIN3 returns 0, and MECT1 adds 49 MAML2 records that a CRTC1-only search misses.

Go deeper

Primary intersections, title-checked1 deeper

USP2 - (MLL OR KMT2A) AND USP2, 11 raw, 11 on-topic.

PMID	what it is	on-topic?
30899083	founding second/minor BCR letter	yes - the central paper of this beat
30869817	first KMT2A-USP2 infant AML, by WTS	yes
32196814	cryptic KMT2A-USP2 and -USP8 by mate-pair sequencing	yes
35087776	GATA2 outlier activation; 2/2 KMT2A-USP2 cases	yes
35734412	ML classifier; LAMP5 estimated all KMT2A-USP2 fusions	yes, verified by reading the abstract text
36476892	Chinese paediatric B-ALL RNA-seq series; 1 case of KMT2A-USP2	yes, verified by reading the abstract text
37803862	two refractory B-ALL KMT2A-USP2 cases, blinatumomab	yes
40628245	venetoclax R/R paediatric ALL; one KMT2A::USP2 patient	yes
41189082	dual KMT2A::USP2 and CBL::GRIK4 infant B-ALL	yes
41402159	Australian targeted RNA-seq exchange; KMT2A::USP2 reference fusion	yes, verified by reading the abstract text
42389523	blinatumomab + DLI post-HSCT in KMT2A::USP2 B-ALL	yes

Three of these (35734412, 36476892, 41402159) had gene symbols stripped from the MCP-delivered abstract and were individually re-fetched as raw text to confirm they genuinely name KMT2A-USP2 rather than matching on a reference list. All three do 1.

MAML2 - 21 raw, 15 on-topic, 6 rejected

Rejected, with the reason:

PMID	title subject	why rejected
38084644	NR1D1-rearranged soft tissue tumour	MAML2 appears as an NR1D1 partner; no KMT2A involvement
34364112	primary pulmonary salivary gland-type tumours, targeted NGS	CRTC1-MAML2 territory; no KMT2A-MAML2 finding
23035786	MAML2 gene split in mucoepidermoid carcinoma	CRTC1-MAML2; the canonical salivary literature
34219124	non-coding RNAs in thymoma	thymoma context but not a KMT2A-MAML2 finding; retained only as background
28597942	SNP-array copy-number changes indicative of gene fusions in paediatric cancer	methodological; no KMT2A-MAML2 case identifiable from the abstract
25016062	functional characterisation of MLL fusion proteins with Sleeping Beauty vectors	MLL fusions generally; MAML2 not identifiable as one of the constructs from the abstract

Accepted and used: 17551948 (founding), 18337764 (secondary T-ALL), 31343482 (recurrent in T-t-ALL), 31599492 (post-hepatoblastoma T-ALL), 28535805 (first B-ALL), 34550633 (regressed clone), 32923872 (thymoma), 40879670 (MAML3 thymoma), 25031740 (biallelic inv(11)), 38604799 (inv(11) ALL progressing to AML), 41834483 (secondary T-LBL), 37517063 (KMT2A-rearranged metachronous mucoepidermoid carcinoma, partner unstated), 38067300 and 33674910 (thymic-tumour fusion reviews, used as maps only), 36966725 (mitotic recombinatory evolution in acute leukaemia, retained as background) 1.

Two of the accepted set are used only as maps (the two thymoma reviews) and one only as background, so the count of papers actually carrying a claim in this beat's entries is 12, not 15. That distinction is recorded because 'on-topic' and 'load-bearing' are different numbers and conflating them inflates apparent evidence.

Counted absences, and their limits2 deeper

query	count	what it means
`(MLL OR KMT2A) AND USP2 AND menin`	0	nobody has tested menin dependence of KMT2A::USP2
`(MLL OR KMT2A) AND USP2 AND (revumenib OR menin OR MEN1)`	0	same, with the drug and the gene symbol added
`(MLL OR KMT2A) AND MAML2 AND menin`	0	nobody has tested menin dependence of KMT2A::MAML2, in leukaemia or thymoma
`(MLL OR KMT2A) AND SEPTIN3`	0	the current symbol finds nothing
`(MLL OR KMT2A) AND UBPY`	0	the USP8 alias adds nothing
`(MLL OR KMT2A) AND (UBP41 OR USP2a OR USP2-69)`	0	the USP2 aliases add nothing

The standing warning applies to all of these. A counted absence is only as good as its symbol set, and these were run over gene symbols plus 'menin' and 'revumenib', not over every menin-inhibitor code or trade name. More importantly, menin-inhibitor trials enrol on KMT2A rearrangement rather than on partner identity, so patients with these fusions have very likely received revumenib without any paper being indexed under the partner's name. The absence is in the literature, not necessarily in the clinic.

The trap that did NOT materialise

The dispatch warned that USP2 would collide with USP22 and with the general ubiquitin-protease literature. Checked rather than assumed:

(MLL OR KMT2A) AND USP22 returns 1 record, and it does not overlap the 11 USP2 records.
USP2 AND (deubiquitinase OR deubiquitinating) returns 231 records - a large unrelated literature, but it does not intrude once MLL/KMT2A is required.

So the USP2/USP22 hazard in this project is a writing hazard, not a retrieval hazard. That changes where the guard belongs: in the entry body, as a discriminating table (see gene-usp2), not in the search strategy. Reporting the feared collision as though it had occurred would have been a fabricated methodological finding.

The traps that did materialise

MAML2 salivary noise. MAML2 alone 544; MAML2 AND (mucoepidermoid OR salivary) 345; CRTC1 AND MAML2 151. A bare MAML2 search is roughly 1-in-26 useful for this project.
Retired symbols, twice. MECT1 AND MAML2 returns 49 records that a CRTC1-only search misses - MECT1 is the retired symbol for CRTC1. And (MLL OR KMT2A) AND SEPT3 returns 1 record where SEPTIN3 returns 0, for the septin partner newly named in the 2019 letter.
A new failure mode: the un-indexed Letter. (MLL OR KMT2A) AND USP8 returns exactly 1 record - and it is not the letter that named USP8 as a KMT2A partner, because that letter (PMID 30899083) has no indexed abstract and so its gene symbols are not searchable. An enumeration keyed to symbols would report USP8 as having a one-paper literature and would miss the discovery paper. This is the retired-symbol lesson in a new form: the defect is in what PubMed indexes, not in the symbol.

What this beat deliberately did not enumerate

Stated so the gaps are visible rather than implied:

SNX9 and SEPT3, the other two novel partners named in the 2019 letter (1 record each in intersection). They are real KMT2A partners and got no entry; the beat prioritised USP2, USP8 and MAML2.
AFDN in the minor BCR (9 patients, the second commonest BCR2 partner). AFDN already has graph coverage as a major-BCR partner, and its BCR2 slice - which would test whether the UCH-rescue hypothesis is necessary - was not pursued. This is the biggest deliberate gap in the beat.
MAML1 and MAML3 beyond citing the single MAML3 thymoma case (1 and 3 records respectively).
CRTC1::MAML2 salivary-gland biology, a 151-record literature that shares the exact MAML2 fragment with the leukaemia fusion. It is the largest untapped resource bearing on what the MAML2 module does, and reading it was out of scope.

Facts and figures

status	run
run date	2026-09-19
result summary	USP2 11/11 on-topic; MAML2 15/21 on-topic with 6 itemised rejections; USP8 under-retrieved by the symbol query; three menin intersections empty; the second-BCR paper is the founding report of the existing minor BCR, not a third cluster.

counts

rejections

inputs

[
  "PubMed E-utilities esearch and efetch",
  "PubMed MCP get_article_metadata and get_full_text_article",
  "PMC full text for PMC6756029 and PMC10169636"
]

counts — the numbers behind the chart

{
  "(MLL OR KMT2A) AND USP2": {
    "raw": 11,
    "title_checked_on_topic": 11,
    "rejected": 0
  },
  "(MLL OR KMT2A) AND MAML2": {
    "raw": 21,
    "title_checked_on_topic": 15,
    "rejected": 6,
    "load_bearing": 12
  },
  "(MLL OR KMT2A) AND USP8": {
    "raw": 1,
    "actually_relevant": 3,
    "note": "under-retrieval: the founding letter has no indexed abstract"
  },
  "(MLL OR KMT2A) AND USP22": {
    "raw": 1,
    "overlap_with_usp2_set": 0
  },
  "(MLL OR KMT2A) AND MAML3": {
    "raw": 3
  },
  "(MLL OR KMT2A) AND MAML1": {
    "raw": 1
  },
  "(MLL OR KMT2A) AND SEPT3": {
    "raw": 1
  },
  "(MLL OR KMT2A) AND SEPTIN3": {
    "raw": 0
  },
  "(MLL OR KMT2A) AND SNX9": {
    "raw": 1
  },
  "(MLL OR KMT2A) AND UBPY": {
    "raw": 0
  },
  "(MLL OR KMT2A) AND (UBP41 OR USP2a OR USP2-69)": {
    "raw": 0
  },
  "(MLL OR KMT2A) AND USP2 AND menin": {
    "raw": 0
  },
  "(MLL OR KMT2A) AND USP2 AND (revumenib OR menin OR MEN1)": {
    "raw": 0
  },
  "(MLL OR KMT2A) AND MAML2 AND menin": {
    "raw": 0
  },
  "MAML2": {
    "raw": 544
  },
  "MAML2 AND (mucoepidermoid OR salivary)": {
    "raw": 345
  },
  "CRTC1 AND MAML2": {
    "raw": 151
  },
  "MECT1 AND MAML2": {
    "raw": 49
  },
  "USP2 AND (deubiquitinase OR deubiquitinating)": {
    "raw": 231
  },
  "USP2 AND MDM2": {
    "raw": 17
  },
  "thymoma AND KMT2A": {
    "raw": 6
  }
}

rejections — the numbers behind the chart

[
  {
    "pmid": "38084644",
    "reason": "NR1D1-rearranged soft tissue tumour; MAML2 as an NR1D1 partner, no KMT2A",
    "title": "NR1D1-rearranged soft tissue tumour: A clinicopathological and molecular analysis of four additional cases.",
    "authors": "Yin T, Zhu P, Lao IW, et al.",
    "venue": "Histopathology",
    "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."
  },
  {
    "pmid": "34364112",
    "reason": "pulmonary salivary gland-type tumours; CRTC1-MAML2 territory",
    "title": "Mutational landscape of primary pulmonary salivary gland-type tumors through targeted next-generation sequencing.",
    "authors": "Wang F, Xi SY, Hao WW, et al.",
    "venue": "Lung Cancer",
    "year": 2021,
    "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."
  },
  {
    "pmid": "23035786",
    "reason": "MAML2 split in mucoepidermoid carcinoma; canonical CRTC1-MAML2 literature",
    "title": "Clinicopathological significance of MAML2 gene split in mucoepidermoid carcinoma.",
    "authors": "Noda H, Okumura Y, Nakayama T, et al.",
    "venue": "Cancer Sci",
    "year": 2013,
    "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."
  },
  {
    "pmid": "34219124",
    "reason": "non-coding RNAs in thymoma; no KMT2A-MAML2 finding retrievable from the abstract",
    "title": "Identification and Characterization of Non-Coding RNAs in Thymoma.",
    "authors": "Ji G, Ren R, Fang X",
    "venue": "Med Sci Monit",
    "year": 2021,
    "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."
  },
  {
    "pmid": "28597942",
    "reason": "SNP-array methodology paper; no identifiable KMT2A-MAML2 case in the abstract",
    "title": "Copy number alterations determined by single nucleotide polymorphism array testing in the clinical laboratory are indicative of gene fusions in pediatric cancer patients.",
    "authors": "Busse TM, Roth JJ, Wilmoth D, et al.",
    "venue": "Genes Chromosomes Cancer",
    "year": 2017,
    "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."
  },
  {
    "pmid": "25016062",
    "reason": "Sleeping Beauty characterisation of MLL fusions; MAML2 not identifiable among the constructs from the abstract",
    "title": "Functional characterisation of different MLL fusion proteins by using inducible Sleeping Beauty vectors.",
    "authors": "Wächter K, Kowarz E, Marschalek R",
    "venue": "Cancer Lett",
    "year": 2014,
    "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."
  }
]

traps confirmed

[
  "MAML2 salivary-gland noise is real: 345 of 544 records",
  "MECT1 is the retired symbol for CRTC1 and adds 49 records a CRTC1-only search misses",
  "SEPT3 returns 1 record where SEPTIN3 returns 0",
  "an un-indexed Letter makes its own gene symbols unsearchable, which under-retrieves USP8 threefold"
]

traps that did not materialise

[
  "USP2/USP22 retrieval collision: USP22 intersection is 1 record with zero overlap, so the hazard is transcription, not search"
]

deliberate gaps

[
  "SNX9 and SEPT3, the other two novel partners from the 2019 letter (1 record each)",
  "AFDN's minor-BCR slice (9 patients) - the natural control for the UCH-rescue hypothesis and the biggest gap left",
  "MAML1 and MAML3 beyond one cited case",
  "the 151-record CRTC1::MAML2 salivary literature, which shares the exact MAML2 fragment"
]

How this could bite us back

Every count here will be wrong within a year, and the three empty menin intersections are the ones most likely to change - they are exactly the experiments this beat argues should be done, so success would silently invalidate the entry's headline claim.
Recording 11/11 on-topic for USP2 risks giving the impression of a healthy literature. Eleven papers for the ninth commonest partner of a well-studied oncogene is not a literature; it is an absence with citations attached.
The one trap that did not materialise is the most important methodological result here, because reporting a feared collision as though it had occurred would have manufactured a finding. Checking the negative was worth the two queries it cost.

Still open

Would a full-text search (Europe PMC, Google Scholar) recover KMT2A-USP8 or KMT2A-MAML2 papers that the abstract-indexed PubMed query misses? The un-indexed-Letter failure mode suggests yes, and this beat did not test it.
How many KMT2A-rearranged patients on menin-inhibitor trials carry USP2 or MAML2 fusions? Trial enrolment is by rearrangement, not partner, so the data may exist in sponsor databases while the literature is empty.
Does the CRTC1::MAML2 salivary literature contain functional work on the shared MAML2 fragment that would transfer directly to KMT2A::MAML2? 151 records is a large untapped resource for a fusion with almost no functional characterisation of its own.

Every source behind this entry (2)

E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database
PubMed E-utilities esearch and efetch, run 2026-09-19, returned the counts and records enumerated in this entry, with every hit in the USP2 and MAML2 intersections checked against its title and abstract.
PubMed E-utilities esearch and efetch queries for USP2, USP8, MAML2, MAML3, SEPT3/SEPTIN3, SNX9 and menin intersections with MLL/KMT2A, run 2026-09-19 · National Center for Biotechnology Information (NCBI), PubMed · PubMed (NCBI E-utilities) 2026
What this evidence cannot show
A single day's snapshot of the PubMed index; every count will drift. Queries use [All Fields], which matches reference lists and keyword blocks as well as titles and abstracts. 'Title-checked' means title plus indexed abstract where one exists; for Letters without indexed abstracts (30899083, 18337764, 31599492, 41834483, 41189082) only the title and MeSH terms were available, so their classification rests on less. Non-English records (37803862, 36476892, 38604799) were classified from their English titles and, where present, English abstracts.
Evidence that argues against this (1)
- PubMed is not the whole literature. Conference abstracts, cooperative-group data supplements and the DCAL group's own unpublished per-patient breakpoint database all contain information about these fusions that no PubMed query reaches, and the recombinome papers show the DCAL database is substantially larger than what is published case by case. Meyer C et al. 2023
E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
The enumeration's central negative finding is that PMID 30899083's 'second breakpoint cluster region' is the minor BCR already held in this graph as breakpoint-kmt2a-minor-bcr, not a third, previously unrecorded cluster - the paper's own closing text calls it 'a minor BCR' and 'a second, minor BCR'.
Human MLL/KMT2A gene exhibits a second breakpoint cluster region for recurrent MLL-USP2 fusions. · Meyer C, Lopes BA, Caye-Eude A, Cavé H, Arfeuille C, et al., Marschalek R · Leukemia 2019
PMID 30899083 DOI
What this evidence cannot show
This is a reading of a Letter's full text, and the Letter never states in so many words 'this is the same region other authors call BCR2' - the identification rests on the coordinates (intron 19/20 to exon 24), the partner spectrum (USP2-dominated) and the authors' own interchangeable use of 'second' and 'minor'. The 5' boundary differs by one intron between this paper and the 2023 recombinome, which is recorded rather than resolved. Full text read (PMC6756029).
Evidence that argues against this (1)
- The one-intron boundary discrepancy is the strongest reason to hesitate: if intron 19 breakpoints are real and the 2023 intron 20 boundary is a consensus that excludes them, then the two descriptions are not quite the same interval. Nothing in either paper settles it, and the per-patient coordinates that would are unpublished. Meyer C et al. 2023

Connected to

supports →The "second breakpoint cluster region" of KMT2A (Meyer 2019): the founding report of BCR2, not a third cluster
records the reading that the 2019 'second BCR' is the existing minor BCR, with the counts behind it
tests →USP2 (ubiquitin carboxyl-terminal hydrolase 2) - the ninth commonest KMT2A partner and the only one that is a deubiquitinase
11 raw records, 11 title-checked on-topic; the feared USP2/USP22 retrieval collision does not occur
tests →MAML2 (mastermind-like 2) - the Notch co-activator that is a KMT2A partner in leukaemia and in thymoma
21 raw, 15 on-topic, 6 itemised rejections; 345 of 544 bare MAML2 records are salivary-gland work
related to →The deferred KMT2A partners: what the counts actually are, what was written, what was rejected, and three numbers the previous enumeration got wrong
continues the deferred-partner enumeration into USP2, USP8 and MAML2, the partners that beat left