Cell identities considered and not written: what the search actually returned for basophils, mast cells, eosinophils, NK cells and the rest

The standing instruction that governs this ledger is the one about controlled intersections: a small intersection means nothing unless both halves are counted. A query returning 9 records could mean the association is absent or could mean the literature on either side is tiny. All counts below were run against NCBI eutils esearch on 2026-09-19 with [Title/Abstract] field restriction and no other filters, and every non-zero intersection under 20 records had its titles inspected before a disposition was assigned 1.

The disease half is (KMT2A[Title/Abstract] OR MLL-rearranged[Title/Abstract] OR MLL-AF4[Title/Abstract] OR MLL-AF9[Title/Abstract]) = 2,919 records. Using KMT2A alone returns 1,651 and misses the older literature, which matters for cell types whose association would have been described in the 1990s if it existed at all.

Basophil and mast cell are the strongest rejections in this ledger and are worth stating positively rather than as an absence of evidence: against a basophil literature of 20,779 records and a mast cell literature of 51,465, the intersection with a 2,919-record disease literature is exactly zero in both cases. That is not an under-studied association; it is the absence of one. Both are granulocytic/mast lineages downstream of the GMP or of a separate basophil-mast progenitor, and their exclusion sharpens the point made in cell-type-cmp that KMT2A fusions occupy a narrow slice of the myeloid branch — monocytic, not granulocytic, not basophil-mast, not erythroid, not megakaryocytic.

Plasma cells and Langerhans cells were checked because both appear in the differential of childhood histiocytic and lymphoid neoplasia; osteoblasts because they are a classical HSC niche cell.

The disposition here is narrower than the one above and the wording matters. These three are not rejected on the grounds that the co-occurrences are all accidental — two of the ten eosinophil records are real KMT2A biology, namely a KMT2A partial tandem duplication in hypereosinophilic syndrome and a YAP1::KMT2A sarcoma. They are rejected because none of the records describes a KMT2A-driven leukaemia of that lineage, which is what a cell-type entry in this beat would have to be about 3.

The lesson the ledger exists for still holds, and the severe-asthma sputum paper appearing in both the eosinophil and neutrophil intersections makes it: a raw intersection of 10-14 records looks like a thin but real literature, and a common reason for co-occurrence is that both terms appear in a paper about something else entirely. A count is not a finding, and reading the titles is what separates the two cases.

Note also what the eosinophil search surfaced: YAP1::KMT2A sarcoma. KMT2A fusions exist outside haematopoiesis. That is out of scope for a cells beat but is a genuine and probably under-represented fact in this graph.

1. The bone-marrow mesenchymal and neural niche — the highest-priority remaining gap. The intersection of the disease half with (mesenchymal OR niche OR stromal)[Title/Abstract] returns 86 records. It contains at least one substantial primary finding: neuropathy of the sympathetic nervous system promotes leukaemic bone-marrow infiltration in an MLL-AF9 AML model, with AML disrupting sympathetic nerves and the quiescence of Nestin-positive niche cells, expanding mesenchymal stem and progenitor cells primed for osteoblastic differentiation at the expense of HSC-maintaining NG2-positive periarteriolar niche cells, and with the leukaemogenic adrenergic signal transduced by the beta-2 but not beta-3 adrenergic receptor on stromal cells 2. That is a cell-type story (Nestin+ MSPC, NG2+ periarteriolar cells), it is KMT2A-specific, and the graph has no entry for any of it. It is deferred here only because a niche beat is a different beat, and it is flagged as the single best target for the next cells cycle.

2. The MPAL blast. Deliberately not written as a cell_type, because mechanism-kmt2a-mpal-mixed-phenotype already covers the WHO lineage-assignment criteria, the one-plastic-cell-versus-two-clones question, the roughly 10 percent KMT2A share of ambiguous-lineage leukaemia and the ALL-type-induction treatment finding, at cell-identity depth. Writing a parallel cell-type-mpal-blast would violate the one-entry-per-thing rule in the direction of duplication rather than granularity. The one piece of new material found this cycle — that ETP-ALL and T/myeloid MPAL have overlapping immunophenotypes and both carry KMT2A rearrangements, separated definitively only by cytoplasmic MPO — has been placed in cell-type-early-t-precursor instead 4.

3. The erythroid progenitor and proerythroblast. Intersection 14 records. The erythroid question that mattered to this project — whether erythroid character explains the haem-biosynthesis dependency — is answered in cell-type-mep, which records that none of the 11 KMT2A-rearranged DepMap lines is erythroid or megakaryocytic while eight such lines sit in the comparison group. A separate erythroid entry would add the normal biology without adding a KMT2A-specific fact.

4. Senescence as a distinct cell state. Folded into cell-state-quiescence rather than written separately, because the only KMT2A-adjacent paediatric-leukaemia report describing a 'senescence-associated stem-like' dormant subpopulation is in Ph-like ALL and uses 'senescent' and 'dormant' together, which conflicts with the standard distinction between reversible quiescence and irreversible senescence. The quiescence entry records the contradiction as an open question rather than asserting a state the literature does not cleanly define.

For completeness, the positive side of the ledger. Cell types: cell-type-monocyte, cell-type-monoblast, cell-type-gmp, cell-type-cmp, cell-type-mep, cell-type-megakaryocyte, cell-type-clp, cell-type-early-t-precursor, cell-type-plasmacytoid-dendritic-cell. Cell states: cell-state-quiescence, cell-state-gmp-like, cell-state-cell-cycle-g1-cdk6, cell-state-fetal-program-retention.

One of these, cell-type-plasmacytoid-dendritic-cell, sits close to the rejection threshold: its intersection is 9 records and the evidence is case-report-level, with one published KMT2A-rearranged 'BPDCN-like' case turning out to be acute monocytic leukaemia once MPO was assessed. It was written rather than rejected because the neonatal-leukaemia literature lists BPDCN among presentations of the population in which KMT2A-AFF1 is commonest, because CD123 is a live therapeutic target shared with AML blasts and the GMP, and because one case documents pDC and monocytic programmes in a single blast population. Its confidence_overall is set to weak and its caveats say so.