Oncofusion

start / graph / gene-cd7

Genemoderate confidence 3 of 5: One strong study, or several smaller consistent ones.Q45-immunophenotype

CD7 - the T/NK antigen on a B blast that flags KMT2A rearrangement and worse event-free survival, and the fourth antigen on the lineage-switch list

also called CD7 · GP40 · T-cell antigen CD7 · T-cell leukemia antigen · TP41 · P09564

CD7 belongs on T cells and natural killer cells, so finding it on a B-lineage leukaemia is a sign that the cell is not respecting lineage boundaries. In a prospective series of 152 B-ALL patients, the handful with CD7 on their blasts were significantly more likely to carry a KMT2A rearrangement and did worse.

The same thing, for a scientist

CD7 (UniProt **P09564**, CD7_HUMAN, **240 aa**, single reviewed human record, no gene synonyms) is the T-cell/NK antigen also called GP40. It is the **shortest protein in this beat** and the least written: named in 8 entries of this graph, the bottom of the queue the coordinator handed over, and it earns an entry for a reason that has nothing to do with frequency. **CD7 on a B blast is a measured KMT2A signal.** In a prospective study of 152 consecutive precursor-B ALL patients screened for aberrant T/NK antigens (CD1a, CD5, CD4, CD7, CD8, CD56), CD5 was found in 1, **CD7 in 6** and CD56 in 9. The two aberrant-antigen groups behaved oppositely: **CD56-positive patients were 89% children, standard clinical risk (P = .010), with frequent ETV6::RUNX1 (P = .021)**, whereas **CD7-positive patients were adolescent/young-adult/adult-skewed (83%) with an adverse cytogenetic profile (P = .001), frequent BCR::ABL1 (P = .004) and frequent KMT2A rearrangement (P = .045)**, and inferior event-free survival (P = .040) with no significant overall survival difference (P = .317) [E1]. **CD7 is also one of the four antigens after which lineage switch has been collected.** Project EVOLVE's antigen list is **CD19, CD22, CD38 and CD7**, and it explicitly includes 5 rare presentations such as **T-ALL transforming to AML** [E2]. So CD7 sits on both sides: it is a marker whose presence predicts the genotype that switches, and a target whose pressure has preceded switches. **And CD7 marks the strangest KMT2A-rearranged entity in this beat.** Wuchter's 533-sample series included 8 **CD7+CD56+ myeloid/natural killer cell precursor acute leukaemias**; 2 of them were NG2/7.1-positive, and one adult 7.1-positive case of this entity was **MLL-germline** by Southern blot - one of the specificity failures that made 7.1 'sensitive but not entirely specific' [E3]. A published case describes an aggressive **MLL-rearranged CD7+CD56+ myeloid/NK precursor acute leukaemia** with extramedullary relapse [E4].

Go deeper
Why six patients justify an entry

The CD7 cohort in 1 is six patients, and this entry says so in every claim. It is written anyway for three reasons that are about structure rather than statistics:

  1. The CD56 arm is the internal control. The same study, same laboratory, same panel, same period found CD56 aberrancy going the opposite way on age, risk and genotype. A finding with a matched internal contrast of opposite sign is much harder to explain as a fishing expedition than the same finding alone.
  2. It is the only marker in this beat that is both a lineage-infidelity signal at diagnosis and a targeted antigen with switches behind it. CD33 and CD117 are myeloid markers on a B blast; CD7 is a T/NK marker on a B blast, a bigger jump, and this graph's mechanism-lineage-switch-kmt2a has almost no T-axis evidence in it.
  3. CD7-directed therapy is arriving. CD7 CARs are in development for T-ALL, and EVOLVE already records switches after CD7 targeting. The graph should have a CD7 entry before someone proposes CD7 in this disease, not after.

What it does not justify. Six patients cannot give a rate, and the association is with 'KMT2A rearrangement' in an adult-skewed cohort, not with KMT2A::AFF1 in infants. The honest claim is that CD7 aberrancy in B-ALL is rare, is enriched for adverse cytogenetics including KMT2A rearrangement, and has never been measured in the infant population this graph is about.

Cell type, cell state, localisation, epigenetics
  • Cell type. T cells from the earliest thymic stages, NK cells, and a small fraction of normal myeloid progenitors. On blasts: aberrantly in a small percentage of B-ALL 1; characteristically in early T-cell precursor ALL and in myeloid/NK precursor acute leukaemia 34.
  • Cell state. An early-lymphoid/NK state marker. Its co-expression with CD117 and cytoplasmic CD79a was read in 1999 as suggesting a stem cell disorder, which is the same intuition the modern plasticity literature formalises.
  • Localisation. Plasma membrane, type I transmembrane Ig-superfamily member, 240 residues 5.
  • Epigenetics. Nothing in this beat places the KMT2A fusion at the CD7 locus. CD7 here is a lineage-infidelity read-out, not a fusion target gene - and that distinction is the whole point of hyp-ng2-tracks-the-transcriptional-programme-not-the-rearrangement.
Facts and figures
uniprotP09564
gene symbolCD7
length AA240
cd numberCD7
lineage switch antigenone of four antigens (CD19, CD22, CD38, CD7) named in Project EVOLVE; per-antigen breakdown not published (PMID 40193715)
accession resolutiongene_exact:CD7 AND organism_id:9606 AND reviewed:true, fields include gene_primary and length; one row; primary name read back as CD7; length 240 asserted for verify-accessions.py.
aberrant expression in b all
{
  "cohort_n": 152,
  "cd7_positive_n": 6,
  "cd5_positive_n": 1,
  "cd56_positive_n": 9,
  "cd7_kmt2a_r_p": 0.045,
  "cd7_bcr_abl1_p": 0.004,
  "cd7_adverse_cytogenetics_p": 0.001,
  "cd7_efs_vs_cd56_p": 0.04,
  "pmid": "33526373",
  "title": "Precursor B-lineage acute lymphoblastic leukemia patients with aberrant natural killer cell and T cell - lineage antigen expression: experience from a tertiary cancer care center.",
  "authors": "Bommannan K, Arumugam JR, Radhakrishnan V, et al.",
  "venue": "Hematol Transfus Cell Ther",
  "year": 2022,
  "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."
}
depmap kmt2a selectivity
{
  "chronos_mean_kmt2a_rearranged": -0.1251,
  "chronos_mean_other_leukaemia": -0.1558,
  "difference": 0.0307,
  "pan_essentiality_all_lines": -0.1252,
  "selectivity_rank": 12932,
  "rank_denominator": 17728,
  "provenance": "Joined 2026-09-21 from analyses/depmap-kmt2a-selective-scan-full (generated 2026-09-19T10:12:06Z), the project's own genome-wide scan over 11 KMT2A-rearranged and 43 other leukaemia DepMap lines. NOT re-fetched and NOT independent of that scan: every caveat it carries applies here, including that the KMT2A-rearranged arm is a small number of lines drawn from nine individuals, that the rank denominator is 17,728 genes, and that a difference is not a dependency. Added because only 26 of 308 gene entries carried this number, which made any KMT2A-selectivity question answerable for 8% of the corpus.",
  "how_to_read": "A negative difference means MORE required in KMT2A-rearranged lines than in other leukaemia lines. A rank near 17,728 is ANTI-selective. Neither is evidence of a fusion-driven dependency on its own: check pan_essentiality_all_lines before calling anything selective, and note that a paralogue-redundant gene can score near zero for reasons that have nothing to do with this disease - see ana-the-paralogue-redundancy-excuse-is-not-tested-and-my-proxy-decided-the-answer."
}
How this could bite us back
  • CD7 is on normal T and NK cells, so a CD7-directed CAR made from T cells fratricides unless CD7 is genetically removed from the product, and a CD7-directed antibody depletes the patient's T and NK compartment. In an infant with a developing immune system and a likely transplant ahead, that is a different risk than in an adult.
  • Project EVOLVE collected lineage switches after CD7-directed therapy among its four named antigens. Adding CD7 to a targeting strategy in KMT2A-rearranged disease adds a fourth axis of antigen pressure to a blast whose answer to pressure is to change lineage, with median survival after switch of 4.8 months.
  • The entire KMT2A association for CD7 in B-ALL rests on six patients at P = .045 in an adult-skewed cohort. Anyone acting on it should treat it as a reason to measure CD7 prospectively in infants, not as an established feature of the disease.
  • CD7 positivity is also a diagnostic trap in the other direction: the CD7+CD56+ entity in which NG2 gave a false positive was one of the cases that made NG2 'not entirely specific'. So CD7 co-expression degrades the reliability of the marker this beat's headline rests on.
Still open
  1. What is the frequency of aberrant CD7 expression in infant KMT2A::AFF1 B-ALL? The only measurement found is six patients in an adult-skewed cohort, and the infant number is unknown.
  2. Does aberrant CD7 at diagnosis predict lineage switch, as opposed to predicting adverse cytogenetics? The two are confounded in the only available series because KMT2A-r and BCR::ABL1 both cluster there.
  3. How many of the Project EVOLVE switches followed CD7-directed therapy? The abstract names CD7 among four antigens without a breakdown.
  4. Does CD7 co-expression predict NG2 false positivity? The one MLL-germline 7.1-positive case in the 533-sample series was a CD7+CD56+ leukaemia, which would be an actionable caveat on the graph's headline marker if it generalised.
Every source behind this entry (5)
  1. E1moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    In a prospective study of 152 consecutive precursor-B ALL patients analysed by flow cytometry for aberrant T/NK-cell antigens (CD1a, CD5, CD4, CD7, CD8, CD56), CD5, CD7 and CD56 expression were observed in one, six and nine patients respectively. CD56-expressing patients were predominantly children (89%), presented as standard clinical risk (P = .010) and frequently carried ETV6::RUNX1 (P = .021). CD7-expressing patients were adolescent-young-adult/adult-age skewed (83%) and had an adverse cytogenetic profile (P = .001), especially frequent BCR::ABL1 (P = .004) and KMT2A rearrangement (P = .045). CD7-expressing patients had inferior event-free survival (P = .040) than their CD56-expressing counterparts, with no significant difference in overall survival (P = .317).

    PMID 33526373DOI
    What this evidence cannot show

    Abstract read in full from raw efetch XML on 2026-09-20; full text not read. **The CD7 group is SIX patients.** Every P value in that arm is computed on six cases, including the KMT2A association at P = .045, which is the weakest of them and would not survive any multiplicity correction across the six antigens and multiple outcomes tested. **The survival comparison is CD7-positive against CD56-positive, not against conventional B-ALL**, so 'inferior event-free survival' is relative to the favourable aberrancy group. Single tertiary centre; **adult-skewed cohort, so this says nothing directly about infants**, the population this graph is about.

    Evidence that argues against this (1)
    • The CD56 arm of the same study found the opposite pattern - paediatric, standard risk, ETV6::RUNX1-enriched - which is what makes the CD7 result credible and also shows how strongly aberrant-antigen findings depend on which antigen is examined. A panel that had tested only CD56 would have concluded that lineage infidelity in B-ALL is a favourable feature. Bommannan K et al. 2022
  2. E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    Project EVOLVE identified cases of lineage switch after a host of antigen-targeted therapies, naming CD19, CD22, CD38 and CD7, and evaluated 75 cases: 53 (70.7%) B-ALL to AML, 17 (22.7%) B-ALL to MPAL or acute leukaemia of ambiguous lineage, and 5 (6.7%) rare presentations such as T-cell ALL to AML. Ten further cases with incomplete immunophenotypic change were described as lineage drift. Among the 70 B-ALL-derived cases, switch emerged at a median of 1.5 months after immunotherapy with 81.4% within 6 months, 45 of 70 (64.3%) involved KMT2A rearrangements, remission rates were under 40% and median overall survival after switch was 4.8 months.

    PMID 40193715DOI
    What this evidence cannot show

    Abstract read in full from raw efetch XML on 2026-09-20; full text not read in this session. **A retrospective case collection with no denominator**, so no incidence of switch after CD7-directed therapy can be derived; the abstract does not break the cases down by preceding antigen, so the number of CD7-preceded switches is **not established here**. The 5 rare presentations include T-ALL to AML, which is the closest this dataset comes to a CD7-axis event and is described without a count by antigen.

    Evidence that argues against this (1)
    • Because the series has no denominator and does not break cases down by preceding antigen, it establishes that CD7-directed therapy was among the exposures in the collected cases and nothing about relative risk. Reading the four-antigen list as four equally implicated targets is an over-reading. Kovach AE et al. 2026
  3. E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    Among 533 acute leukaemia samples screened with monoclonal antibody 7.1, eight were CD7+CD56+ myeloid/natural killer cell precursor acute leukaemias, of which two reacted with 7.1. One adult with 7.1-positive CD7+CD56+ myeloid/natural killer cell precursor acute leukaemia was MLL-germline by Southern blot, contributing with 11 adult AML cases to the conclusion that 7.1 is a sensitive but not entirely specific marker. Separately, 7.1-positive AML cases showed a myelomonocytic immunophenotype with CD56 co-expression in 33 of 38 cases.

    PMID 10914547DOI
    What this evidence cannot show

    Abstract read in full from raw efetch XML on 2026-09-20; full text not read. **Eight cases of an entity that has since been reclassified.** 'Myeloid/natural killer cell precursor acute leukaemia' was a proposed entity in 2000 and is not a current WHO or ICC category, so these eight patients would be distributed across other diagnoses today, and MLL status was determined by Southern blot, which misses cryptic rearrangements. The CD7 content of this evidence item is a small subgroup of a study about a different marker.

    Evidence that argues against this (1)
    • A single MLL-germline 7.1-positive CD7+CD56+ case is one patient from a reclassified entity assessed by Southern blot; if that rearrangement was cryptic, the 'specificity failure' it contributes to is not one. This matters because the same sentence is the basis for the widely quoted claim that NG2 is not entirely specific. Soon G et al. 2015
  4. E4weak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.paper

    A published case describes an aggressive MLL gene rearranged CD7-positive CD56-positive myeloid/natural killer cell precursor acute leukaemia with extramedullary relapse in the uterus and breast.

    PMID 26308138DOI
    What this evidence cannot show

    **Title only: the PubMed record carries no abstract**, and the bibliographic fields here were fetched from efetch on 2026-09-20 rather than recalled. The five-route full-text test was not run on this record. **A single case report in an adult with extramedullary relapse in uterus and breast** - it establishes that a CD7+CD56+ KMT2A-rearranged entity occurs and nothing about frequency, infants, or mechanism. Carried under this graph's known-but-unread convention at weak confidence.

    Evidence that argues against this (1)
    • A case report selected for an unusual relapse site cannot establish anything about the CD7+CD56+ KMT2A-rearranged phenotype in general, and the entity it names has since been reclassified. It corroborates only that the combination exists, which the 533-sample series [E3] already showed. Wuchter C et al. 2000
  5. E5established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    CD7 is UniProt P09564 (CD7_HUMAN), T-cell antigen CD7, also known as GP40, T-cell leukemia antigen and T-cell surface antigen Leu-9, 240 amino acids, a single reviewed human record with primary gene name CD7 and no gene synonyms.

    What this evidence cannot show

    Fetched 2026-09-20 with `gene_exact:CD7 AND organism_id:9606 AND reviewed:true`, requesting `gene_primary` and `length` per rules 62 and 65; primary name read back as CD7; exactly one reviewed row; length 240 asserted for `verify-accessions.py`. Gene symbol and CD number coincide. **At 240 aa this is the shortest protein in this beat**, a useful discriminator against any swap with the other short markers here (CD38 at 300, CD33 at 364, CD34 at 385).

    Evidence that argues against this (1)
    • CD7 is expressed on normal T and NK cells, which is the central obstacle to CD7-directed cell therapy: a CD7 CAR made from T cells fratricides unless CD7 is edited out of the product. Nothing in the accession conveys that, and it is the dominant practical fact about this target. UniProt Knowledgebase, reviewed (Swiss-Prot) 2026
Connected to