Oncofusion

start / graph / fusion-etv6-ntrk3

Fusionestablished confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.Q11-comparative

ETV6::NTRK3 (ETV6-TRKC), t(12;15)(p13;q25)

also called ETV6-NTRK3 · TEL-TRKC · ETV6-TRKC · EN fusion · ETV6::TRKC · t(12;15)(p13;q25)

A gene called ETV6, whose job includes sticking to copies of itself, is fused to the enzyme half of a nerve-growth-factor receptor called TRKC, so the enzyme is forced into a long chain of active copies and never switches off. Because that enzyme has a real ATP pocket, a pill can shut it down - and in babies with congenital fibrosarcoma this has replaced disfiguring surgery.

The same thing, for a scientist

ETV6::NTRK3 arises from t(12;15)(p13;q25) and was cloned from congenital (infantile) fibrosarcoma, where it fuses the ETV6 self-association domain to the NTRK3/TRKC protein tyrosine kinase domain and was present in all three tumours first analysed [E1]. The 5' module is the ETV6 **SAM/PNT domain** - originally described as a helix-loop-helix dimerisation domain [E1] but now understood to form an **open-ended head-to-tail helical polymer** through a mid-loop/end-helix interface, a mechanistically stronger form of forced self-association than the closed oligomers of NPM1 [E2]. The retained NTRK3 module is a classical ATP-site receptor tyrosine kinase domain stripped of its extracellular NT-3-binding region, so the kinase is constitutively active, ligand-independent and signals through RAS-MAPK, PI3K-AKT and STAT3. The same fusion recurs across unrelated tissues - congenital fibrosarcoma, cellular mesoblastic nephroma, and **secretory breast carcinoma, where it was found in 12 of 13 cases (92%)** [E3] - making it one of the clearest examples of a single fusion whose tissue outcome is set by the cell it lands in rather than by the oncoprotein. Therapeutically it is the most successfully drugged fusion in this comparison set: larotrectinib gave an overall response rate of **75% by independent review (80% by investigator) across 55 patients and 17 TRK-fusion tumour types**, with 71% of responses ongoing at one year [E4]; entrectinib gave objective responses in **31 of 54 patients (57%)** [E5]; and in children, larotrectinib has been used to shrink locally advanced infantile fibrosarcoma and avoid disfiguring surgery [E6].

What it looks like
loading 1JI7…
PDB 1JI7 ↗
shape

This is real experimental data — atoms placed by crystallography or cryo-EM, deposited by the people who did the experiment.

I have never read one of these pictures before

A protein is one long chain of amino acids that folds up. The ribbon traces that chain from one end to the other; it is not a wire, it is a simplification of where the backbone runs. Drag to rotate, scroll to zoom.

Domains are the stretches of chain that do a particular job, and they are what the coloured blocks above the picture mark out. Click one and the view flies to it. In this project the important question about any domain is usually whether the fusion still has it — the labels say lost where it does not.

Anything in yellow sticks is a bound small molecule, which is usually the drug.

Go deeper
What is kept and what is lost on each side1 deeper

Kept from ETV6 (the 5' partner): the N-terminal SAM/PNT domain (sterile alpha motif / pointed domain, residues ~40-120), the self-association module. Knezevich and colleagues described it as the 'helix-loop-helix (HLH) protein dimerization domain' in 1998 - the historical name, worth recognising because the older literature uses it throughout 1.

Lost from ETV6: the central internal region and, critically, the C-terminal ETS DNA-binding domain. ETV6 is normally a transcriptional repressor; the fusion loses all of its DNA-binding and repressive function. So the fusion is not a corrupted transcription factor - it is a kinase carrying a polymerisation tag.

Kept from NTRK3/TRKC: the protein tyrosine kinase domain with its bilobal fold and ATP-binding cleft, plus the juxtamembrane region and C-terminal tail including the docking site that binds the phosphotyrosine-binding domain of IRS-1 1.

Lost from NTRK3: the signal peptide, the leucine-rich and immunoglobulin-like extracellular NT-3-binding domain, and the transmembrane segment. The kinase therefore cannot be regulated by neurotrophin-3 and is cytoplasmic.

The comparative shape. This is architecturally the same object as NPM1::ALK: self-association module + naked kinase domain. Both are the diametric opposite of KMT2A::AFF1, where the 5' partner supplies chromatin targeting and the 3' partner supplies a disordered elongation scaffold with no pocket.

Polymer, not dimer - and why that distinction matters

The ETV6 SAM domain does not simply dimerise. It polymerises head-to-tail into an open-ended helical filament, using a mid-loop surface on one subunit and an end-helix surface on the next; the polymer has been crystallised (PDB 1JI7) and mutation of either interface abolishes polymerisation and the associated transcriptional repression 2.

That has two consequences worth recording in a comparative graph:

  1. Potency. An open-ended polymer concentrates many kinase domains in one assembly, which drives trans-autophosphorylation far more effectively than a dimer. This is the strongest form of the 'forced oligomerisation' mechanism in the set - stronger than NPM1's closed pentamer and stronger than AFDN's RA1-mediated self-association.
  2. Drug-target logic. The SAM polymer interface is structurally defined and obligate, which makes it superficially attractive as a target. It is nonetheless not pursued, for the same reason NPM's oligomerisation domain is not: the ATP pocket downstream is an easier and better target, and the polymerisation function is generic enough that mutational escape or partner substitution is plausible. This is a direct, informative counterexample to the proposition that obligate partner interfaces are the right place to intervene.
Mechanism: constitutive kinase driven by polymerisation

Class: constitutive-kinase.

  1. ETV6 SAM domains polymerise head-to-tail, assembling many fusion molecules into a filament 2.
  2. The clustered NTRK3 kinase domains trans-autophosphorylate their activation loops, locking into the active conformation with no neurotrophin required.
  3. Active TRKC phosphorylates docking tyrosines and fires RAS-MAPK (the dominant proliferative arm), PI3K-AKT (survival) and STAT3. A highly conserved NTRK3 C-terminal sequence binds the phosphotyrosine-binding domain of IRS-1, and that interaction is required for transformation - an interesting secondary interface unique to this fusion.
  4. Retroviral transfer of ETV6-NTRK3 into murine mammary epithelial cells produced transformed cells that formed tumours in nude mice while still producing glands and expressing epithelial antigens, showing that the fusion's transforming activity is compatible with maintained differentiation 3.

That last point is a striking contrast with the KMT2A fusions, whose defining consequence is a differentiation block. ETV6::NTRK3 drives proliferation and survival while leaving the differentiation program largely intact - which is part of why the diseases it causes are indolent and why removing the signal is curative rather than merely cytostatic.

Clinical phenotypes across three tissues, and their ages

One fusion, several diseases, each in a different cell of origin.

Congenital / infantile fibrosarcoma (CFS). A malignant fibroblastic tumour of children aged two years or younger, and the setting in which the fusion was cloned. CFS is, as Knezevich and colleagues emphasised, 'unique among human sarcomas in that it has an excellent prognosis and very low metastatic rate' despite being histologically identical to adult-type fibrosarcoma, which has a poor prognosis - the fusion is what distinguishes them 1. Cell of origin: a primitive mesenchymal fibroblast.

Cellular mesoblastic nephroma. The renal counterpart, in infants; ETV6-NTRK3 plus trisomy 11 establishes a histogenetic link between the two tumours. Cell of origin: renal mesenchyme.

Secretory breast carcinoma (SBC). A rare, indolent subtype of infiltrating ductal carcinoma. ETV6-NTRK3 was confirmed in 12 of 13 SBC cases (92%) and in no other ductal carcinomas, making it a primary driving event in a human breast cancer 3. SBC occurs across a wide age range including children and adolescents. Cell of origin: a breast secretory epithelial cell.

The comparative lesson. The identical oncoprotein makes an indolent infant sarcoma, an infant renal tumour and an indolent adult/paediatric breast carcinoma. Tissue identity - not the fusion - sets the disease. That is exactly the argument the KMT2A nodes make from the other direction (one 5' half, six partners, opposite lineages and prognoses), and the two converge on the same conclusion: cell of origin is a first-class determinant, not a footnote.

Druggability and therapeutic status

Clinically validated, on the fusion protein itself, and tumour-agnostic.

Larotrectinib (selective TRK inhibitor). In a combined analysis of three protocols, 55 patients aged 4 months to 76 years with 17 unique TRK-fusion tumour types had an overall response rate of 75% by independent review and 80% by investigator assessment; at one year 71% of responses were ongoing and 55% of patients remained progression-free; adverse events were predominantly grade 1 and no patient discontinued for drug-related toxicity 4. This basket result led to the first genuinely histology-agnostic oncology approvals.

Entrectinib (TRK/ROS1/ALK inhibitor with CNS penetration). Objective responses in 31 of 54 patients (57%, 95% CI 43.2-70.8) with advanced NTRK-fusion solid tumours 5.

Paediatric practice change. The phase 1 paediatric study was amended specifically to enrol children with locally advanced infantile fibrosarcoma who would otherwise require disfiguring surgery for complete resection - i.e. the drug was used to convert a mutilating operation into a conservative one 6. That is about as strong a demonstration of targeted-therapy value in a child as exists.

Second-generation agents (selitrectinib, repotrectinib) address solvent-front and gatekeeper resistance mutations in the NTRK kinase domain.

The contrast that defines this graph.

ETV6::NTRK3KMT2A::AFF1
Retained effectorfolded kinase with ATP cleftdisordered elongation scaffold
Drug bindsthe fusion proteina partner protein (menin)
Best responseORR 75-80%, durable 4CR+CRh 22.8%
Effect on differentiationproliferation without blocking differentiation 3differentiation block is the disease
Partner interface as targetobligate, structurally defined (1JI7), and not worth pursuingthe only validated node
A note on how this fusion draws

The retained NTRK3 portion is the tyrosine protein kinase domain, and the element this entry previously named — the conserved C-terminal IRS-1 PTB-docking sequence — lies outside and C-terminal to it and is not separately annotated in the NTRK3 domain record. The lane now draws to the kinase domain. That is the right landmark for the fusion's catalytic activity and it does not depict the docking sequence, which is where the signalling specificity of ETV6::NTRK3 is thought to come from.

Analogy: the same class as NPM1::ALK, and why it is the strongest cell-of-origin argument in the beat

Closest analogue, and the axis. fusion-npm1-alk, on the axis of shared mechanism class - a self-associating 5-prime module bolted to a retained receptor tyrosine kinase domain, with the ligand-binding ectodomain gone. Both are the class mechanism-constitutive-kinase-activation names, and both are drugged: larotrectinib 75% overall response by independent review across 55 patients and 17 TRK-fusion tumour types 4, entrectinib 31 of 54 5, crizotinib 83-90% in relapsed paediatric ALK-positive ALCL.

What the analogy predicts, and the test that has been run. If the 5-prime partner contributes only self-association, it must be replaceable. On the NPM1::ALK side it demonstrably is - NPM was functionally replaced by the unrelated TPR leucine zipper with almost equal transforming efficiency, and three NPM deletion mutants that failed to multimerise each lost kinase activity and transforming ability. The prediction generalises to this fusion and is the reason nobody attempts to drug the ETV6 SAM interface: the partner is a switch, and the kinase is the target.

Where the two differ inside the class. The ETV6 SAM/PNT module forms an open-ended head-to-tail helical polymer through a mid-loop/end-helix interface 2, a mechanistically stronger form of forced self-association than NPM1's closed oligomer. The consequence, if any, is unmeasured.

Where the analogy breaks, and it is the most useful thing this entry contributes to the beat. NPM1::ALK is essentially one disease - ALK-positive ALCL, ~90% of paediatric cases. ETV6::NTRK3 recurs across three unrelated tissues: congenital fibrosarcoma, where it was cloned 1; cellular mesoblastic nephroma; and secretory breast carcinoma, in 12 of 13 cases 3. Same mechanism, same class of partner, same kind of drug - and one is confined to a lymphoid entity while the other crosses mesenchyme and epithelium. Mechanism class predicts druggability and predicts the disease not at all. The variable doing the work is the cell of origin, and nothing in the oncoprotein encodes it.

Why that matters for the index question. This graph's central worry, restated in the cycle-39 log, is whether KMT2A::AFF1's phenotype is set by the fusion or by the fetal progenitor it lands in. ETV6::NTRK3 is the cleanest available demonstration that one oncoprotein plus several cells gives several diseases - a fully controlled natural experiment with the driver held constant, which is the experiment the KMT2A field cannot run because the fusion is restricted to haematopoiesis. It does not settle the KMT2A question; it establishes that the lineage explanation is not a weak default but the observed behaviour of fusion oncoproteins that get to try more than one tissue.

And the limit on transferring it. ETV6::NTRK3's effector is a signalling kinase whose output (RAS-MAPK, PI3K-AKT, STAT3) is deliberately generic, so a permissive relationship with cell identity is the expected result. A chromatin oncoprotein that binds a defined set of loci has far less room to be cell-agnostic, and the fact that a kinase fusion is promiscuous across tissues says less about KMT2A::AFF1 than the reverse comparison would.

Facts and figures
gene5ETV6
gene3NTRK3
cytobandt(12;15)(p13;q25)
prognosisExcellent for the paediatric mesenchymal tumours: congenital fibrosarcoma is described as unique among human sarcomas for its excellent prognosis and very low metastatic rate despite being histologically identical to poor-prognosis adult-type fibrosarcoma. Secretory breast carcinoma is likewise indolent. In advanced/metastatic NTRK-fusion disease, larotrectinib gave ORR 75-80% with 71% of responses ongoing at 1 year, and entrectinib 57%.
uniprot5P41212
uniprot3Q16288
uniprot5 verifiedP41212 = ETV6_HUMAN, gene ETV6, transcription factor ETV6, 452 aa, Homo sapiens (UniProt REST, checked 2026-09-18)
uniprot3 verifiedQ16288 = NTRK3_HUMAN, gene NTRK3, NT-3 growth factor receptor, 839 aa, Homo sapiens (UniProt REST, checked 2026-09-18)
PDB noteNo structure of the intact ETV6::NTRK3 fusion protein or of a polymerised fusion filament exists. 1JI7 = crystal structure of the TEL (ETV6) SAM polymer, the retained 5' self-association module in its polymeric state - the mechanism, visualised; 3V5Q, 6KZC and 4YMJ = the NTRK3/TRKC kinase domain bound to selective or pan-TRK inhibitors, i.e. the retained effector module with drugs in its ATP cleft. Deliberately EXCLUDED: 1WWC, the NT-3-binding domain of human TRKC - it resolves and is genuine NTRK3, but that domain is lost from the fusion. All entries verified against the RCSB entry API on 2026-09-18.
druggability classclinically-validated
druggability noteValidated ON THE FUSION PROTEIN ITSELF at the NTRK3 ATP cleft, and validated tumour-agnostically. The partner interface (the SAM polymer) is obligate and has a crystal structure, yet is not a drug target - because the downstream ATP pocket is easier and because a generic self-association function invites substitution or escape. A direct counterexample to the idea that obligate partner interfaces are the natural place to intervene.
stage of hematopoiesisNot haematopoietic. Tissue cell of origin varies with the disease: a primitive mesenchymal fibroblast in congenital/infantile fibrosarcoma, renal mesenchyme in cellular mesoblastic nephroma, and a breast secretory epithelial cell in secretory breast carcinoma. One oncoprotein, three lineages - the clearest evidence in this set that tissue identity rather than the fusion determines the disease.
age peak<2 y for congenital/infantile fibrosarcoma and cellular mesoblastic nephroma; secretory breast carcinoma spans childhood to adulthood
developmental order noteNull because this is a non-haematopoietic comparator, so it has no position on the haematopoietic developmental axis. If an analogous axis were drawn for mesenchymal development, congenital fibrosarcoma would sit at the primitive end (a fetal/neonatal mesenchymal progenitor, tumour present at or near birth), and in that loose sense it is the mesenchymal analogue of KMT2A::AFF1's fetal origin.
drugs noteLarotrectinib, entrectinib, selitrectinib and repotrectinib are the relevant agents; none has a node in this graph yet, so the array is left empty rather than populated with dangling ids.
domain coordinate noteDomain-coordinate note (coordinator, 2026-09-19): the 'Protein kinase' is a short motif, not a UniProt-annotated domain of NTRK3 (Q16288), whose annotated features are the LRRs, LRRCT, two Ig-like C2-type domains and the protein kinase domain (538-839). It therefore has no residue range here and the 3D viewer cannot draw it. Naming it in prose is correct; giving it a bar would require coordinates from primary work.
diseases
[
  "congenital/infantile fibrosarcoma",
  "cellular mesoblastic nephroma",
  "secretory breast carcinoma",
  "secretory carcinoma of salivary gland (mammary analogue)",
  "rare NTRK-fusion AML and thyroid carcinoma"
]
frequency
{
  "secretory_breast_carcinoma_pct": 92,
  "secretory_breast_carcinoma_n_over_total": "12/13",
  "congenital_fibrosarcoma_pct": null,
  "congenital_fibrosarcoma_pct_note": "The founding report detected ETV6-NTRK3 transcripts in all three congenital fibrosarcomas analysed, too few to support a percentage; later series report the fusion in the large majority of infantile fibrosarcomas, but we did not verify a specific figure and therefore record null rather than invent one."
}
breakpoints
[
  {
    "id": "bp-etv6-e5-ntrk3-e15",
    "gene5_exon_last": 5,
    "gene3_exon_first": 15,
    "frequency_pct": 100,
    "population": "ETV6::NTRK3-positive tumours across all three tissue contexts; the fusion is essentially monomorphic at the transcript level, which is why one RT-PCR assay serves as a diagnostic marker across congenital fibrosarcoma, mesoblastic nephroma and secretory breast carcinoma",
    "domains_retained_5": [
      "SAM/PNT self-association domain (historically the 'HLH dimerisation domain')"
    ],
    "domains_lost_5": [
      "ETV6 central internal/regulatory region",
      "ETS DNA-binding domain",
      "C-terminal repression region"
    ],
    "domains_retained_3": [
      "protein tyrosine kinase domain with ATP-binding cleft",
      "juxtamembrane cytoplasmic region",
      "Protein kinase"
    ],
    "domains_lost_3": [
      "signal peptide",
      "extracellular leucine-rich and immunoglobulin-like NT-3-binding region",
      "transmembrane domain"
    ],
    "phenotype_notes": "Exon numbers follow the conventional ETV6 exon-5 / NTRK3 kinase-domain junction description and should be treated as indicative rather than exact, since annotation differs between transcript sets. Functionally the junction is unambiguous: everything 5' contributes only self-association, everything 3' contributes only the kinase. Loss of the NT-3-binding region makes the kinase ligand-independent; retention of the SAM domain makes it constitutively polymeric."
  }
]
PDB IDS
[
  "1JI7",
  "3V5Q",
  "6KZC",
  "4YMJ"
]
mechanism class
[
  "constitutive-kinase"
]
drugs
[]
key interactors
[
  "IRS-1 (via a conserved NTRK3 C-terminal PTB-docking sequence)",
  "RAS-MAPK",
  "PI3K-AKT",
  "STAT3",
  "wild-type ETV6 (co-polymerisation partner)"
]
trials
[
  "NCT02122913",
  "NCT02637687",
  "NCT02576431"
]
How this could bite us back
  • Resistance: solvent-front (e.g. NTRK3 G623R-equivalent) and gatekeeper mutations in the kinase domain arise under first-generation TRK inhibitors, which is why selitrectinib and repotrectinib exist. Even an excellent pocket is not a permanent solution.
  • Compensation: bypass activation through MET, KRAS, BRAF or downstream MAPK reactivation has been documented as resistance to TRK inhibition, so the pathway is escapable without touching the kinase.
  • Essential-function toxicity: TRK receptors (NTRK1/2/3) are required in sensory, proprioceptive and central neural development, and the characteristic class toxicities - dizziness, weight gain, withdrawal pain on stopping - reflect on-target effects on normal neurotrophin signalling. Treating neonates with a pan-TRK inhibitor during nervous-system development is a real, incompletely characterised risk.
  • Clinical-equipoise hazard: congenital fibrosarcoma has an excellent prognosis with surgery alone. Introducing indefinite systemic kinase inhibition into a curable infant disease trades a surgical morbidity for a pharmacological one, and the duration of therapy needed is unknown.
  • Conceptual bite-back for the lab's hypothesis: like NPM1::ALK, this fusion shows that an obligate partner self-association interface can be necessary and yet be the wrong target. Obligate does not imply druggable, and druggable does not imply the best node.
Still open
  1. The ETV6 SAM polymer interface is obligate, structurally solved (1JI7) and never pursued as a drug target, because a better one sits downstream. For KMT2A::AFF1 the situation is reversed - the interface is all there is. Is there any general rule that predicts when an obligate interface is worth drugging, and can it be stated in advance rather than after the fact?
  2. ETV6::NTRK3 drives proliferation while permitting differentiation, whereas every KMT2A fusion enforces a differentiation block. Is that difference intrinsic to signalling versus chromatin mechanisms, or a consequence of the cell of origin? If a kinase fusion were introduced into a fetal lympho-myeloid progenitor, would it block differentiation too?
  3. The same oncoprotein produces an indolent infant sarcoma, an infant renal tumour and an indolent breast carcinoma. What makes each of those three tissues permissive and everything else non-permissive - partner-locus expression, chromatin accessibility at the NTRK3 promoter, or a tissue-specific requirement for a co-operating lesion?
  4. What are the long-term developmental consequences of TRK inhibition started in infancy? NTRK signalling is required in neural development, and the surgery-sparing use of larotrectinib in neonates is now routine on limited long-term data.
  5. Can the cell-of-origin experiment ETV6::NTRK3 runs naturally be run deliberately for KMT2A::AFF1? The fusion is confined to haematopoiesis in patients, but nothing forbids expressing it in a mesenchymal or epithelial progenitor and asking whether it transforms, and if so into what. KMT2A::MAML2 already provides a hint that a KMT2A fusion can sit in an epithelial tumour - it is the first recurrent fusion described in type B thymoma - so the restriction of KMT2A::AFF1 to B-lineage infant disease may be about which cells acquire the translocation rather than which cells it can transform.
Every source behind this entry (6)
  1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    A recurrent t(12;15)(p13;q25) in congenital fibrosarcoma fuses ETV6 to NTRK3, producing transcripts that encode the ETV6 self-association (originally called helix-loop-helix dimerisation) domain joined to the NTRK3 protein tyrosine kinase domain, and was present in all three congenital fibrosarcomas analysed but not in adult-type fibrosarcoma or infantile fibromatosis of the same age group.

    A novel ETV6-NTRK3 gene fusion in congenital fibrosarcoma · Knezevich SR, McFadden DE, Tao W, Lim JF, Sorensen PH · Nature Genetics 1998
    PMID 9462753DOI
    What this evidence cannot show

    Only three CFS tumours were analysed for transcripts in this initial report, so the frequency estimate came later; the paper describes the ETV6 module with the now-superseded 'HLH dimerisation domain' nomenclature, which is the SAM/PNT domain. It proposes but does not demonstrate the transforming mechanism.

    Evidence that argues against this (2)
    • The tumour-type specificity asserted here broke within the same year, and in a different organ. Knezevich et al. detected t(12;15)-associated ETV6-NTRK3 chimeric transcripts in 8 of 9 CELLULAR congenital mesoblastic nephromas and in 2 of 2 mixed CMNs - a renal spindle-cell tumour of infancy, not a fibrosarcoma - while all 4 classical CMNs, plus the Wilms tumours and clear cell sarcomas of the kidney tested, were negative; trisomy 11 was found only in the fusion-positive cellular and mixed CMNs. Absence of the fusion from adult-type fibrosarcoma and infantile fibromatosis therefore does not make it a marker of congenital fibrosarcoma: it marks a cell type that presents in at least two organs, and the list has since grown further. Knezevich SR et al. 1998
    • Infantile-fibrosarcoma morphology is also produced by drivers that are not ETV6-NTRK3 at all, so the fusion is neither necessary for the histology nor a safe proxy for it. Penning et al. assembled 14 BRAF-altered spindle-cell neoplasms with IFS-like morphology - 5 with BRAF point mutations and 10 with BRAF fusions, including seven previously unreported fusion partners, one tumour carrying two BRAF fusions and another three alternative splice variants of EPB41L2-BRAF - in patients from birth to 32 years (median 6 months), twelve soft-tissue and two visceral including one cellular congenital mesoblastic nephroma. They note the same morphology also arises with NTRK1/2/3, MET, RET and RAF1 fusions. A 3-of-3 hit rate in the 1998 index series is a statement about three tumours, not about the entity. Penning AJ et al. 2021
  2. E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    The ETV6/TEL SAM domain polymerises head-to-tail into an open-ended helical polymer through complementary mid-loop and end-helix surfaces, and this polymerisation is required for its function in transcriptional repression and leukemogenesis.

    Polymerization of the SAM domain of TEL in leukemogenesis and transcriptional repression · Kim CA, Phillips ML, Kim W, et al. · EMBO Journal 2001
    PMID 11483520DOI
    What this evidence cannot show

    The polymerisation requirement was demonstrated in the context of ETV6-mediated repression and ETV6::PDGFRB/ETV6-driven leukemogenesis rather than specifically for ETV6::NTRK3, so its role in the fibrosarcoma/breast-carcinoma fusion is transferred by homology of the identical retained domain. Crystallography used a mutant that blocks runaway polymerisation to permit crystallisation.

    Evidence that argues against this (1)
    • The one structure-function experiment done on this fusion measures DIMERS, not polymers, and its functional readout is kinase activation rather than transcriptional repression. Wai et al. showed that ETV6-NTRK3 homodimerises AND heterodimerises with wild-type ETV6, is autophosphorylated on tyrosine, and transforms NIH3T3 cells (soft-agar growth, tumours in SCID mice); deleting the ETV6 HLH/SAM domain abolished dimer formation with either partner and abolished transformation, while an ATP-binding mutant abolished transformation with the dimerisation module intact. Heterodimerisation with endogenous full-length ETV6 is not what an open-ended head-to-tail homopolymer predicts as the operative species, and no experiment in the ETV6::NTRK3 literature measures polymer length or tests the mid-loop/end-helix surfaces that define the polymer model. Transferring a requirement established for ETV6-mediated repression to a receptor tyrosine kinase fusion is a cross-family generalisation, not a result. Wai DH et al. 2000
  3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    ETV6-NTRK3 is expressed in 12 of 13 (92%) human secretory breast carcinomas and in no other ductal carcinomas, and retroviral transfer of the fusion into murine mammary epithelial cells produced tumour-forming cells that still made glands and expressed epithelial antigens.

    Expression of the ETV6-NTRK3 gene fusion as a primary event in human secretory breast carcinoma · Tognon C, Knezevich SR, Huntsman D, et al. · Cancer Cell 2002
    PMID 12450792DOI
    What this evidence cannot show

    n=13, a rare tumour assembled from multiple centres; 'primary event' rests on the fusion's presence in essentially all cases plus the transformation experiment, not on lineage tracing in a human breast. The murine model used retroviral overexpression, which exceeds physiological fusion levels.

    Evidence that argues against this (1)
    • The fusion marks secretory morphology, not breast tissue. Skalova et al. described 16 salivary gland tumours (13 parotid) with the histology and immunophenotype of secretory carcinoma of the breast and found the t(12;15)(p13;q25) ETV6-NTRK3 translocation in all but one analysable case, while 12 conventional salivary acinic cell carcinomas, a pleomorphic adenoma and a low-grade cribriform cystadenocarcinoma were negative and all 3 tested mammary secretory carcinomas were positive; 4 of 13 followed patients had local recurrence and 2 died. 'Expressed in 12 of 13 secretory breast carcinomas and in no other ductal carcinomas' is a within-breast specificity statement and says nothing about organ specificity - the same fusion defines a distinct entity in the parotid gland, which also weakens the inference that the murine mammary-epithelium transduction experiment identifies the natural cell of origin. Skalova A et al. 2010
  4. E4established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    Larotrectinib produced an overall response rate of 75% by independent review and 80% by investigator assessment in 55 patients aged 4 months to 76 years with 17 unique TRK fusion-positive tumour types, with 71% of responses ongoing and 55% of patients progression-free at one year, and no discontinuations for drug-related adverse events.

    Efficacy of Larotrectinib in TRK Fusion-Positive Cancers in Adults and Children · Drilon A, Laetsch TW, Kummar S, et al. · New England Journal of Medicine 2018
    PMID 29466156DOINCT02122913
    What this evidence cannot show

    A pooled analysis of three single-arm protocols with no control group; 55 patients spread over 17 tumour types means most histologies contributed one or two patients, so the ETV6::NTRK3-specific response rate is not separately estimable from this report. Median duration of response and PFS had not been reached, so durability was immature at publication.

    Evidence that argues against this (1)
    • The second TRK inhibitor to report in the same setting achieved a lower objective response rate (57%), so 75-80% should be read as the upper end of the range rather than a class figure. Doebele RC et al. 2020
  5. E5established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    Entrectinib produced objective responses in 31 of 54 patients (57%, 95% CI 43.2-70.8) with advanced or metastatic NTRK fusion-positive solid tumours across three pooled phase 1-2 trials.

    PMID 31838007DOI
    What this evidence cannot show

    Pooled single-arm basket analysis across heterogeneous histologies and NTRK1/2/3 fusions with different partners, so the figure is not specific to ETV6::NTRK3. Entrectinib is a multi-kinase agent (TRK, ROS1, ALK), which complicates attribution of response to TRK inhibition alone.

    Evidence that argues against this (1)
    • A pooled single-arm response rate cannot show the failure mode, which is on-target and already documented in an ETV6-NTRK3 tumour. Drilon et al. treated a patient whose salivary acinic cell carcinoma was reclassified as mammary analogue secretory carcinoma after next-generation sequencing showed an ETV6-NTRK3 fusion; entrectinib produced a dramatic and durable response, followed by acquired resistance that coincided with emergence of a novel NTRK3 G623R mutation, with structural modelling predicting steric interference with drug binding and cell-based assays confirming reduced sensitivity. Objective response in 31 of 54 patients therefore measures initial tumour shrinkage across 10 histologies and three NTRK genes, not durable control of ETV6::NTRK3 disease. Drilon A et al. 2016
  6. E6strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In the paediatric phase 1 study of larotrectinib, the protocol was amended to enrol children with locally advanced infantile fibrosarcoma who would otherwise have required disfiguring surgery to achieve complete resection, establishing a surgery-sparing use of targeted therapy in this fusion.

    PMID 29606586DOINCT02637687
    What this evidence cannot show

    A phase 1 dose-escalation report with 24 patients (17 TRK fusion-positive) whose primary endpoint was safety, not efficacy or surgical outcome; the surgery-sparing claim reflects the eligibility amendment and early experience rather than a prospective surgical endpoint. Long-term consequences of TRK inhibition in infants (neurodevelopment, growth) were not addressable at this follow-up.

    Evidence that argues against this (1)
    • Conventional chemotherapy had already achieved the surgery-sparing endpoint before any TRK inhibitor existed, so the comparator for larotrectinib is not mutilating surgery. In the European paediatric Soft tissue sarcoma Study Group's prospective series of 50 infants (median age 1.4 months) with localised infantile fibrosarcoma, ETV6-NTRK3 present in 87.2% of those tested, first-line vincristine-actinomycin D in inoperable (IRS-III/R2) disease gave a 68.0% response rate, mutilating surgery was performed in only 3 of 50 children, alkylating- and anthracycline-based chemotherapy was avoided in 71.0% of those needing chemotherapy, and 3-year event-free and overall survival were 84.0% and 94.0%. An eligibility amendment admitting children 'who would otherwise have required disfiguring surgery' therefore describes a recruitment category rather than an unmet need that only targeted therapy can meet. Orbach D et al. 2016
Connected to