Oncofusion

start / graph / gene-hmga2

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

HMGA2 (P52926) - a direct KMT2A::AFF1 target through the let-7b axis, and a recurrent fusion partner whose own fusions work by escaping a microRNA rather than by making a chimeric protein

also called HMGA2 · HMGIC · HMGI-C · P52926 · HMGA2_HUMAN · High mobility group protein HMGI-C · High mobility group AT-hook protein 2

HMGA2 is normally held down by a small regulatory RNA. In infant leukaemia the fusion protein switches that brake off in two separate ways, and HMGA2 then releases a cell-cycle brake of its own. Separately, HMGA2 is itself the front half of fusion genes in several benign tumours - but those fusions work by cutting off the piece of message the brake grabs onto, which is a different kind of mechanism from ours.

The same thing, for a scientist

Flagged twice in previous cycles as the family member serving the PI's analogous-fusion-families dimension, and written on that basis. **The fetch inverted the reason for the entry: HMGA2 is a direct target of the index fusion, in the index population, in a paper this graph had never cited.** **Identity.** P52926, `HMGA2_HUMAN`, **109 aa**, *High mobility group protein HMGI-C*; alternative name *High mobility group AT-hook protein 2*; sole gene synonym **HMGIC**. UniProt annotates **no DOMAIN features** and a single **uncoded** location, Nucleus. FUNCTION: transcriptional regulator; cell-cycle regulation through **CCNA2**; chromosome condensation at the meiotic G2/M transition of spermatocytes; postnatal myogenesis and satellite-cell activation (by similarity); and positive regulation of **IGF2** through PLAG1 and PLAG1-independently. **THE INDEX-DISEASE FINDING, uncited here until now.** In infant ALL with KMT2A fusion, the microRNA **let-7b (MIRLET7B) is significantly downregulated by DNA hypermethylation of its promoter**, and HMGA2 — one of the oncogenes MIRLET7B represses — is **reciprocally upregulated, particularly in KMT2A-AFF1 (MLL-AF4)-positive ALL**. And the fusion drives it by **two routes**: *"In addition to the suppression of MIRLET7B, KMT2A fusion proteins positively regulate the expression of HMGA2."* HMGA2 in turn is a **negative regulator of CDKN2A** (p16^INK4A^). The HMGA2 inhibitor **netropsin**, combined with **5-azacytidine**, upregulated and sustained CDKN2A and suppressed growth of KMT2A-AFF1-expressing lines more than 5-azacytidine alone. The authors propose the **MIRLET7B–HMGA2–CDKN2A axis** as a molecular target for infant ALL with KMT2A-AFF1. **AND ITS OWN FUSIONS WORK BY A DIFFERENT MECHANISM ENTIRELY.** UniProt records four separate chromosomal aberrations: **lipomas**, t(3;12)(q27-q28;q13-q15) with **LPP**, and *"HMGA2 is also fused with a number of other genes in lipomas"*; **pulmonary chondroid hamartomas**, t(3;12) with LPP; **parosteal lipomas**, t(3;12)(q28;q14) with LPP; and **uterine leiomyoma**, t(12;14)(q15;q23-24) with **RAD51B**. But the operative event is not a chimeric protein with new activity — it is **loss of the 3′ UTR**. Intragenic breakpoints remove the **let-7 complementary sites**, stabilising the transcript and overexpressing the protein. Quantified: in uterine leiomyomas with 12q15 rearrangements, truncated transcripts predominate in **roughly one third** — 5 of 13 with highly reduced 3′UTR levels against **8 of 13 retaining the complete 3′ UTR with all let-7 sites**. **That makes HMGA2 the cleanest mechanistic CONTRAST in this graph's analogous-families set.** KMT2A fusions matter because the chimera does something new; HMGA2 rearrangements matter because the transcript escapes a microRNA. Same cytogenetic vocabulary, opposite kind of mechanism — and the two meet at the let-7 axis, which is how HMGA2 ends up downstream of KMT2A::AFF1 without being a KMT2A fusion partner. **A shared partner links the two families.** **LPP**, HMGA2's commonest partner in lipomas, is itself an MLL fusion partner: t(3;11)(q28;q23) in AML-M5 arising three years after treatment for follicular lymphoma, with the breakpoint in **intron 8 of both MLL and LPP** — and the paper makes the connection itself, noting LPP *"is frequently fused to HMGIC in lipomas"*. **In blood, HMGA2 overexpression alone is sufficient for a clonal phenotype.** Transgenic mice carrying a 3′UTR-truncated *Hmga2* cDNA showed proliferative haematopoiesis across all lineages, hypercellular marrow, splenomegaly with extramedullary erythropoiesis, erythropoietin-independent erythroid colony formation, and a **growth advantage in competitive repopulation and serial transplantation** — i.e. clonal expansion at the stem-cell level. The same paper notes HMGA2 overexpression in PNH clones, some myeloproliferative neoplasms, and **haematopoietic clones arising from gene-therapy procedures**. **Not an interactome member.** HMGA2 is absent from the 277-partner IntAct REST union, as is LPP. This entry is about a transcriptional target and an analogous fusion family, not about a binding partner.

Facts and figures
uniprotP52926
uniprot entry nameHMGA2_HUMAN
gene symbolHMGA2
protein nameHigh mobility group protein HMGI-C
length AA109
domains noteUniProt P52926 annotates no DOMAIN features. The AT-hook motifs that define the protein and give it its alternative name are not DOMAIN features on this record, so no coordinates are asserted. Empty array rather than an omitted field, because the viewer requires these to be arrays.
localizationUniProt P52926: Nucleus, and the assignment is UNCODED - no experimental evidence code. Same weakness as gene-hmga1, and a real one for a protein whose relevance is chromatin-architectural.
subnuclear localizationNONE. No subnuclear compartment is assigned at all, coded or uncoded. NOT measured in any KMT2A-rearranged cell.
localization shown for fusionfalse
cell typeFor the index-disease finding: infant ALL leukaemic cells and KMT2A-AFF1-expressing cell lines. For the fusion biology: lipomas, pulmonary chondroid hamartomas, parosteal lipomas and uterine leiomyomas. For the haematopoietic phenotype: transgenic mouse marrow, plus reported overexpression in PNH clones, MPNs and gene-therapy-derived clones.
cell stateProliferation is the state with a measured phenotype: HMGA2 regulates the cell cycle through CCNA2, represses CDKN2A, and its overexpression confers a growth advantage in competitive repopulation and serial transplantation. Also meiotic G2/M in spermatocytes. Nothing measured under a menin or DOT1L inhibitor.
epigenetic mark namedNone written or read. Its route into this graph's epigenetic layer is indirect and specific: KMT2A fusions suppress MIRLET7B by DNA hypermethylation of its promoter, which de-represses HMGA2.
epigenetic readoutNot a writer, reader or eraser. It is an AT-hook architectural factor and, in this disease, a transcriptional OUTPUT of the fusion via a methylation-silenced microRNA.
not an interactome memberHMGA2 is absent from the 277-partner IntAct REST union, as is LPP. This entry concerns a transcriptional target and an analogous fusion family, not a binding partner.
shown for fusionfalse
absence classmeasured-not-analysed
uniprot synonyms
[
  "HMGIC"
]
alternative names
[
  "High mobility group AT-hook protein 2"
]
domains
[]
the index disease axis
{
  "citation": "PMID 26403224, Wu, Eguchi-Ishimae, Yagi et al., British Journal of Haematology 2015, doi 10.1111/bjh.13763",
  "step_1": "in infant ALL with KMT2A fusion, MIRLET7B (let-7b) is significantly downregulated by DNA hypermethylation of its promoter region",
  "step_2": "HMGA2, one of the oncogenes repressed by MIRLET7B, is reciprocally upregulated in infant ALL leukaemic cells, particularly in KMT2A-AFF1 (MLL-AF4)-positive ALL",
  "step_3": "and separately - in addition to suppressing MIRLET7B - KMT2A fusion proteins POSITIVELY REGULATE HMGA2 expression, so the fusion drives it by two routes",
  "step_4": "HMGA2 is a negative regulator of CDKN2A, encoding p16INK4A",
  "step_5": "the HMGA2 inhibitor netropsin combined with 5-azacytidine upregulated and sustained CDKN2A and suppressed growth of KMT2A-AFF1-expressing cell lines, more than 5-azacytidine alone",
  "the_authors_proposal": "the MIRLET7B-HMGA2-CDKN2A axis as a possible molecular target for the therapy of infant ALL with KMT2A-AFF1",
  "and_it_was_uncited_in_this_graph": "0 entries before this one"
}
its own fusions work by a different mechanism
{
  "the_aberrations_uniprot_records": {
    "lipomas": "t(3;12)(q27-q28;q13-q15) with LPP; 'HMGA2 is also fused with a number of other genes in lipomas'",
    "pulmonary chondroid hamartomas": "t(3;12)(q27-q28;q14-q15) with LPP",
    "parosteal lipomas": "t(3;12)(q28;q14) with LPP, one case",
    "uterine leiomyoma": "t(12;14)(q15;q23-24) with RAD51B"
  },
  "the_operative_event_is_3prime_UTR_LOSS_not_a_chimeric_protein": "intragenic breakpoints remove the let-7 complementary sites in the 3' UTR, stabilising the transcript and overexpressing the protein",
  "quantified": "in uterine leiomyomas with 12q15 rearrangements, truncated transcripts predominate in roughly one third - 5 of 13 with highly reduced 3'UTR levels, against 8 of 13 retaining the complete 3' UTR with all let-7 complementary sites (PMID 20082846)",
  "why_this_is_the_useful_contrast": "KMT2A fusions matter because the chimera acquires a new activity; HMGA2 rearrangements matter because the transcript escapes a microRNA. Same cytogenetic vocabulary, opposite kind of mechanism - and the two families meet at the let-7 axis, which is how HMGA2 is downstream of KMT2A::AFF1 without being a KMT2A fusion partner."
}
a shared partner links the two families
{
  "LPP": "lipoma preferred partner, 3q28, HMGA2's commonest fusion partner in lipomas",
  "and_it_is_an_MLL_partner_too": "t(3;11)(q28;q23) in AML-M5 arising three years after treatment for follicular lymphoma, i.e. therapy-related; breakpoint in intron 8 of both MLL and LPP; two in-frame MLL-LPP transcripts plus a reciprocal LPP-MLL (PMID 11433529)",
  "the_paper_makes_the_link_itself": "it notes that LPP 'contains substantial identity to the focal adhesion protein, zyxin, and is frequently fused to HMGIC in lipomas'",
  "gene-lpp": "does not exist in this graph",
  "A_DISCREPANCY_TO_CHECK_NOT_TO_REPEAT": "that 2001 paper states the predicted MLL-LPP fusion protein 'includes the A/T hook motifs and methyltransferase domain of MLL'. A retained methyltransferase domain contradicts this graph's model, in which every KMT2A fusion loses the SET domain, and an intron-8 breakpoint sits inside the BCR well N-terminal of it. This is quoted rather than repeated as fact - it is most likely era-specific domain nomenclature, and it needs the full text to settle."
}
hmga2 overexpression alone is sufficient in blood
{
  "citation": "PMID 21460244, Ikeda, Mason et al., Blood 2011, doi 10.1182/blood-2011-02-334425",
  "model": "transgenic mice carrying a 3'UTR-truncated Hmga2 cDNA",
  "phenotype": "proliferative haematopoiesis with increased numbers in all peripheral blood lineages, hypercellular bone marrow, splenomegaly with extramedullary erythropoiesis, erythropoietin-independent erythroid colony formation, and a growth advantage over wild-type cells in competitive repopulation and serial transplantation - clonal expansion at the stem-cell level",
  "and_the_human_settings_it_names": "PNH PIGA-negative clones in two cases, some myeloproliferative neoplasms, and haematopoietic cell clones arising from gene-therapy procedures",
  "note": "this is a MOUSE transgenic overexpression model and the phenotype is MPN-like, not lymphoblastic"
}
germline
{
  "Silver-Russell syndrome 5": "UniProt records HMGA2 variants as causing a form of Silver-Russell syndrome - severe intrauterine growth retardation, poor postnatal growth, triangular face, broad forehead, body asymmetry and minor malformations. Relevant to bite_back: this is a gene whose loss impairs human growth."
}
alias split measured
{
  "HMGA2[tiab]": 1995,
  "HMGIC[tiab]": 209,
  "\"HMGI-C\"[tiab]": 100,
  "intersection HMGA2 and HMGIC": "0 in the fetched sets",
  "intersection HMGA2 and \"HMGI-C\"": "1 in the fetched sets",
  "A_TRUNCATION_CAVEAT_THAT_MATTERS": "the HMGA2 id set was fetched at retmax=800 against a count of 1,995, so it is TRUNCATED and the intersections above are lower bounds, not measurements. No union size is quoted for that reason. This is the same retmax trap that produced a meaningless union figure for B2M in an earlier cycle, and it is recorded rather than glossed.",
  "what_is_safe_to_say": "the historic tokens are small (209 and 100) and the approved symbol dominates by count; whether they overlap materially was not established"
}
context counts
{
  "(HMGA2 OR HMGIC OR \"HMGI-C\") AND (MLL OR KMT2A)": 7,
  "AND leukemia": 44,
  "AND lipoma": 96,
  "AND leiomyoma": 103,
  "AND \"AT-hook\"": 404,
  "AND LPP": 31,
  "AND truncat*": 58,
  "all 7 MLL/KMT2A records read by title": true
}
How this could bite us back
  • This entry touches an intervention and the intervention is the weakest thing in it. Netropsin is described in the source as 'the HMGA2 inhibitor', but it is a DNA minor-groove binder with no selectivity for HMGA2 binding sites over any other AT-rich sequence - so growth suppression of a KMT2A-AFF1 line by netropsin plus 5-azacytidine is fully consistent with non-specific minor-groove occupancy plus global demethylation. It must not be cited as evidence that inhibiting HMGA2 treats infant ALL, and this project has a registry row for exactly this error shape.
  • 5-azacytidine is a real drug with real toxicity in infants; the combination result is in cell lines only, with no primary blast and no animal data in the source abstract.
  • HMGA2 is a growth gene whose germline disruption causes Silver-Russell syndrome 5, with severe intrauterine growth retardation and poor postnatal growth. Any HMGA2-directed strategy in a child would be acting on a gene required for normal growth, and nothing here bounds that cost.
  • The haematopoietic evidence points away from this disease: deregulated HMGA2 in blood produces MPN-like multi-lineage myeloproliferation, not lymphoblastic leukaemia, and in gene-therapy clones it confers clonal fitness without transformation in most reported cases.
Still open
  1. Is HMGA2 upregulation in KMT2A-AFF1 infant ALL a driver or a passenger? The axis paper shows the fusion drives HMGA2 and that HMGA2 represses CDKN2A, but the only functional test is netropsin plus 5-azacytidine in cell lines - and netropsin is a non-selective minor-groove binder.
  2. What does 'KMT2A fusion proteins positively regulate HMGA2' mean mechanistically? Direct occupancy at the HMGA2 locus has not been shown here, and if the effect is downstream of the same let-7 loss then the two-route model collapses to one.
  3. Why does `gene-lpp` not exist? LPP is HMGA2's commonest partner AND an MLL partner in therapy-related AML, which makes it the shared node between two fusion families and the natural next entry.
  4. Does the MLL-LPP fusion really retain a methyltransferase domain? The 2001 abstract says so and this graph's model says no fusion does. The full text would settle whether that is era-specific nomenclature or a genuine exception.
  5. Is let-7b hypermethylation present in KMT2A-AFF1 infant ALL independently of the 2015 report? A single-source methylation claim underpins the whole axis, and this graph has a standing rule about single-lab load-bearing evidence.
Every source behind this entry (4)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    HMGA2 is a direct transcriptional target of the index fusion through a microRNA axis. In infant ALL cells with KMT2A fusion, the microRNA let-7b (MIRLET7B) is significantly downregulated by DNA hypermethylation of its promoter region, and HMGA2 - one of the oncogenes repressed by MIRLET7B - is reciprocally upregulated in infant ALL leukaemic cells, particularly in KMT2A-AFF1 (MLL-AF4)-positive ALL. In addition to suppressing MIRLET7B, KMT2A fusion proteins positively regulate HMGA2 expression. HMGA2 is a negative regulator of CDKN2A, which encodes p16INK4A, and the HMGA2 inhibitor netropsin combined with the demethylating agent 5-azacytidine upregulated and sustained CDKN2A expression, resulting in growth suppression of KMT2A-AFF1-expressing cell lines - more apparent than 5-azacytidine alone. The authors propose the MIRLET7B-HMGA2-CDKN2A axis as a possible molecular target for infant ALL with KMT2A-AFF1.

    HMGA2 as a potential molecular target in KMT2A-AFF1-positive infant acute lymphoblastic leukaemia. · Wu H, Eguchi-Ishimae M, Yagi C, et al. · British Journal of Haematology 2015
    PMID 26403224DOI
    What this evidence cannot show

    Abstract-level; no figure or methods section was read, so the number of infant ALL samples, the magnitude of HMGA2 upregulation, the cell lines used and the netropsin concentrations are not established here. 'KMT2A fusion proteins positively regulate the expression of HMGA2' is asserted in the abstract without the mechanism being stated - whether that is direct promoter occupancy or an indirect consequence is not resolved, and this entry does not claim direct binding. The growth-suppression result is in cell lines, not in primary infant blasts or any animal. This paper was uncited in this graph before now, which is the gap it closes rather than a point in its favour.

    Evidence that argues against this (2)
    • A two-route model is weaker than it sounds, because the two routes are not independently evidenced to the same degree. Promoter hypermethylation of MIRLET7B is a measurement; 'KMT2A fusion proteins positively regulate HMGA2' is a summary claim with no mechanism in the abstract, and if it turned out to be downstream of the same let-7 loss then there is one route, not two. The entry's headline should not rest on the second route.
    • The therapeutic half is the weakest part and this graph has been burned by exactly this shape before. Netropsin is a DNA minor-groove binder with no selectivity for HMGA2 sites over any other AT-rich sequence, so growth suppression of a KMT2A-AFF1 line by netropsin plus 5-azacytidine is consistent with general AT-rich minor-groove occupancy and global demethylation rather than with an HMGA2-specific effect. Calling netropsin 'the HMGA2 inhibitor' imports a specificity the compound does not have - the same error pattern this project recorded for family-named compounds such as a 'KDM3 inhibitor' whose selectivity sat on a different paralogue.
  2. E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    HMGA2's own recurrent fusions work by removing the 3' UTR and escaping let-7 repression, not by creating a chimeric protein with new activity. UniProt records chromosomal aberrations involving HMGA2 in lipomas (t(3;12)(q27-q28;q13-q15) with LPP, and fusion with a number of other genes), pulmonary chondroid hamartomas (t(3;12) with LPP), parosteal lipomas (t(3;12)(q28;q14) with LPP) and uterine leiomyoma (t(12;14)(q15;q23-24) with RAD51B). The operative consequence is loss of the let-7 complementary sites in the 3' UTR: intragenic breakpoints produce truncated transcripts whose increased stability drives protein overexpression. Measured in uterine leiomyomas with rearrangements of 12q14-q15, truncated transcripts were apparently predominant in roughly one third - 5 of 13 showed highly reduced 3' UTR levels, with full-length transcripts almost undetectable in two, while 8 of 13 retained the complete 3' UTR with all let-7 complementary sites.

    PMID 20082846DOI
    What this evidence cannot show

    Abstract-level, one study, n=13 leiomyomas, by real-time RT-PCR rather than by sequencing the breakpoints. The result is explicitly partial: the MAJORITY of rearranged tumours in this series - 8 of 13 - retained the complete 3' UTR with all let-7 sites, so 3'UTR truncation cannot be the mechanism in most of them and something else drives overexpression there. Uterine leiomyoma is a benign mesenchymal tumour and nothing in this evidence item is in a haematopoietic cell. The characterisation of the whole family's mechanism as '3'UTR escape' generalises from this one tumour type plus the UniProt aberration notes.

    Evidence that argues against this (2)
    • UniProt itself supplies the counter-argument, in the same record: 'Chromosomal rearrangements involving HMGA2 do not seem to be the principle pathobiological mechanism in uterine leiomyoma.' So in the best-studied of the four tumour types, the rearrangement may not be the driver at all - which weakens using this family as a clean mechanistic contrast to KMT2A, where the fusion unambiguously is.
    • The contrast this entry draws may be a false dichotomy. Truncating the 3' UTR of a gene that still makes a full-length protein and creating a chimeric protein are not mutually exclusive descriptions of a translocation, and some HMGA2 rearrangements do produce in-frame fusion transcripts with LPP. A cleaner statement is that the DOSAGE consequence is established and the chimeric-protein consequence is not, rather than that the mechanism is categorically different.
  3. E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    LPP is a fusion partner of both HMGA2 and MLL, linking the two families through a shared partner. A t(3;11)(q28;q23) was identified in a patient who developed AML-M5 three years after treatment for follicular lymphoma, and the LPP (lipoma preferred partner) gene on 3q28 was identified as the MLL fusion partner by inverse PCR. The breakpoint occurred in intron 8 of MLL and of LPP; two in-frame MLL-LPP transcripts fusing MLL exon 8 to LPP exon 9 were detected, the smaller carrying a 120 bp deletion of MLL sequence, and a reciprocal LPP-MLL transcript was also detected. The paper notes that LPP contains substantial identity to the focal adhesion protein zyxin and is frequently fused to HMGIC in lipomas.

    Human LPP gene is fused to MLL in a secondary acute leukemia with a t(3;11)(q28;q23). · Dahéron L, Veinstein A, Brizard F, et al. · Genes, Chromosomes & Cancer 2001
    PMID 11433529DOI
    What this evidence cannot show

    A SINGLE CASE, abstract-level, from 2001, in therapy-related AML-M5 following topoisomerase-II-inhibitor treatment - not infant ALL and not the index fusion. The transcripts are predicted fusion proteins inferred from RT-PCR, with no functional characterisation. NOTE A DISCREPANCY THIS ENTRY DELIBERATELY DOES NOT REPEAT AS FACT: the abstract states the predicted MLL-LPP protein 'includes the A/T hook motifs and methyltransferase domain of MLL', whereas this graph's model holds that every KMT2A fusion loses the SET methyltransferase domain, and an intron-8 breakpoint sits inside the breakpoint cluster region well N-terminal of it. That is most likely era-specific domain nomenclature and it needs the full text to settle; it is quoted rather than asserted.

    Evidence that argues against this (1)
    • A shared partner is a weak form of analogy and may be coincidence of gene architecture rather than biology. LPP is a large gene at 3q28 in a region prone to rearrangement, and the two fusions use it for different things - the last two LIM domains in the MLL case, and in lipomas a role in which truncated LPP protein is reported as crucial. Two diseases using the same locus does not mean the locus contributes the same function to both.
  4. E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    HMGA2 overexpression alone is sufficient to produce a clonal haematopoietic phenotype in mice. Transgenic mice carrying a 3'UTR-truncated Hmga2 cDNA expressed increased HMGA2 protein in several tissues including haematopoietic cells and showed proliferative haematopoiesis with increased numbers in all peripheral blood lineages, hypercellular bone marrow, splenomegaly with extramedullary erythropoiesis, and erythropoietin-independent erythroid colony formation; marrow cells from these mice had a growth advantage over wild-type cells in competitive repopulation and serial transplantation, indicating clonal expansion at the stem-cell level. The same report notes HMGA2 overexpression in the clonal PIGA-negative cells of two cases of paroxysmal nocturnal haemoglobinuria, in some myeloproliferative neoplasms, and in haematopoietic cell clones arising from gene-therapy procedures - in nearly all of which the overexpression is due to deletions or translocations removing the 3' UTR.

    PMID 21460244DOI
    What this evidence cannot show

    Abstract-level. This is a MOUSE TRANSGENIC OVEREXPRESSION model, and the phenotype is MPN-like myeloproliferation, NOT lymphoblastic leukaemia - so it does not model infant B-ALL and does not show that HMGA2 contributes to the index disease. Transgene expression was increased 'in various tissues', so the phenotype is not cell-autonomous to the haematopoietic compartment on this evidence. The human settings named - PNH, MPN, gene-therapy clones - are cited by the authors and were not independently verified here.

    Evidence that argues against this (2)
    • The phenotype argues against HMGA2 mattering in THIS disease rather than for it. Deregulated HMGA2 in blood produces a myeloproliferative, multi-lineage, stem-cell-level expansion - which is close to the opposite of KMT2A-AFF1 infant ALL, a lineage-restricted lymphoblastic disease with a short latency and a fetal origin. If HMGA2 overexpression were doing substantial work downstream of the fusion, this is not the phenotype one would predict.
    • The gene-therapy observation cuts against the therapeutic reading in a specific way: HMGA2 overexpression there arises from insertional 3'UTR disruption and confers clonal advantage WITHOUT causing leukaemia in most reported cases. That makes HMGA2 look like a clonal-fitness gene rather than a transforming one, and a fitness gene is a poor target because its inhibition would be expected to have a wide normal-tissue cost - consistent with germline HMGA2 variants causing Silver-Russell syndrome with severe intrauterine growth retardation.
Connected to