In pancreatic neuroendocrine tumours MEN1 and DAXX/ATRX are the two commonest mutations of the same tumour - and both are LOST there while leukaemia needs both intact
also called the panNET mirror · MEN1 44% DAXX/ATRX 43%
There is a rare pancreatic cancer in which the two most frequently broken genes are menin - the protein this whole project is about - and the pair of chaperones this beat is about. In that cancer both are destroyed; in this leukaemia the cells need both of them working. The same two things, wired in opposite directions.
The same thing, for a scientist
Exome sequencing of 68 pancreatic neuroendocrine tumours found that the two most frequently mutated genes are **MEN1 (44%, inactivating)** and **DAXX or ATRX (43%)** - and both are associated with *better* prognosis [E1]. This graph's central axis is the KMT2A-menin interface, and this beat's subject is DAXX/ATRX. **The two are the top two lesions of one tumour type and nothing in this corpus had noticed.** `endocrine-men1-pancreatic-neuroendocrine-tumour` exists and does not contain the string DAXX or ATRX. **The analogy is a sign inversion, which is what makes it useful.** In panNET both are tumour suppressors whose loss drives the tumour. In KMT2A-rearranged leukaemia menin is not lost - it is *required*, because the fusion needs it to reach chromatin, which is why a menin inhibitor works at all - and DepMap says the same of DAXX in the B-lineage arm [see `ana-daxx-lymphoid-dependency-under-three-nulls`]. Two genes, one tumour where losing them is the lesion and another where keeping them is the dependency. **The third way of breaking the axis is in a third disease.** In paediatric glioblastoma the H3.3-ATRX-DAXX pathway is hit in 44% of tumours, with H3F3A itself mutated at K27M or G34R/V in 31%, and **ATRX/DAXX mutation in 100% of G34 tumours** [E2]. So the same chromatin axis is broken three ways in three cancers of young people: **mutate the histone** (glioma), **delete the chaperone** (glioma, panNET), or - in KMT2A-rearranged leukaemia - **fuse the writer** and leave the histone and chaperone intact. **And the chaperone-loss arm has a consequence the leukaemia arm has never been tested for.** 61% of panNETs show alternative lengthening of telomeres, and every ALT-positive tumour carried an ATRX or DAXX mutation or loss of the nuclear protein [E3]. `"alternative lengthening of telomeres"[tiab] AND KMT2A[tiab]` returns 0 and the same against MLL returns 0, against 921 records for the bare term [E4].
The mapping, stated as a table so the disanalogies are visible
| panNET | paediatric GBM | KMT2A-rearranged leukaemia | |
|---|---|---|---|
| MEN1 | inactivated in 44% | not a driver | required - the fusion's chromatin anchor, and an approved drug target |
| DAXX/ATRX | mutated in 43% | mutated in 31%, 100% of G34 | wild type; DepMap says required in the B-lineage arm (3 donors) |
| H3.3 itself | wild type | K27M or G34R/V in 31% | wild type |
| direction of the lesion | loss of function | loss of function | gain of function in a fourth gene |
| telomere phenotype | ALT in 61% | ALT in ATRX-mutant tumours | never measured |
| prognostic sign | MEN1 and DAXX/ATRX mutation = better survival | H3.3/ATRX mutation = worse | n/a |
The principal disanalogy, and it is large. panNET and G34 glioma are diseases of losing the chaperone; this beat's leukaemia observation is a requirement for it. Those are not the same biology and a reader who collapses them will predict the opposite of the right thing. What transfers is narrower: the H3.3 deposition axis is a recurrent oncogenic target in young-onset tumours of quiet genomes, and its two failure modes have opposite consequences for telomeres.
The hypothesis this generates, and why it is cheap
If DAXX is genuinely required in KMT2A-rearranged B-lineage blasts, the panNET literature says what to look for when it is removed: telomere dysfunction and a drift toward ALT. That is a prediction with a $200 assay attached - a C-circle assay, or telomere FISH for ultrabright foci - and it has never been run on a KMT2A-rearranged line.
It also supplies the missing bite-back for any DAXX-directed intervention in this disease. ATRX/DAXX loss is the canonical route to ALT in human tumours. A drug that removes DAXX function in a leukaemic blast is administering the exact lesion that supplies telomerase-independent telomere maintenance in two other cancers. That is not a hypothetical toxicity; it is a documented consequence of the same perturbation in humans.
And it gives the deep DAXX group in the DepMap panel a testable identity: if RCH-ACV and YCUB-2 are in that group because they are ALT-positive, one assay on thirteen cell lines explains a cluster that the dependency analysis could not.
What this corrects in an existing entry
endocrine-men1-pancreatic-neuroendocrine-tumour is a substantial entry about MEN1 in panNET, written for the menin-inhibitor safety argument. It cites eight PMIDs and none is Jiao 2011; it does not contain the strings DAXX, ATRX or alternative lengthening of telomeres. That is not an error in the entry - its question was drug exposure and tumour latency, not the somatic genetics - but it means this graph held a panNET entry and a DAXX entry for several cycles with the strongest link between them unwritten.
This entry does not restate that one. It supplies the co-mutation fact, the telomere consequence and the sign inversion, and links to it.
| domain | oncology |
|---|---|
| source concept | The H3.3-ATRX-DAXX deposition axis as a recurrent oncogenic target in young-onset, quiet-genome tumours |
| principal disanalogy | The sign is inverted. panNET and G34 glioma LOSE the chaperone; the leukaemia observation is a REQUIREMENT for it. Collapsing the two predicts the opposite of the right thing. What transfers is the axis and the telomere consequence, not the direction. |
| absence class | never-measured |
| relationship to existing entries | `endocrine-men1-pancreatic-neuroendocrine-tumour` covers MEN1 in panNET for the drug-exposure argument and contains no mention of DAXX, ATRX or ALT. `gene-atrx` already lists ALT in its bite_back as a general risk of ATRX loss; this entry supplies the human tumour data behind that and extends it to a testable prediction in this disease. Neither Jiao 2011, Schwartzentruber 2012 nor Heaphy 2011 was cited anywhere in this graph before 2026-09-24, verified with scripts/check-citation-reuse.py. |
mapping
[
{
"bio": "KMT2A-rearranged leukaemia requires menin (gain-of-function dependency; drug target)",
"analog": "panNET inactivates MEN1 in 44% (loss-of-function tumour suppressor)"
},
{
"bio": "KMT2A-rearranged B-lineage lines are the most DAXX-dependent in the DepMap B-ALL panel",
"analog": "panNET inactivates DAXX or ATRX in 43%"
},
{
"bio": "the chromatin lesion in KMT2A-r leukaemia is a FUSION of a histone writer, with the histone and chaperone intact",
"analog": "the chromatin lesion in paediatric GBM is a point mutation in the histone (H3F3A K27M/G34R) or deletion of its chaperone"
},
{
"bio": "telomere maintenance mechanism of KMT2A-r blasts: never measured",
"analog": "61% of panNETs are ALT-positive and all of those carry an ATRX or DAXX lesion"
},
{
"bio": "MEN1 and DAXX/ATRX have never been reported in the same tumour genome in leukaemia",
"analog": "MEN1 and DAXX/ATRX are the top two mutated genes of one tumour type"
}
]generated hypotheses
[ "hyp-daxx-dependence-in-kmt2a-r-b-all-is-a-telomere-phenotype" ]
measurements — the numbers behind the chart
[
{
"quantity": "mutation_frequency",
"subject": "MEN1 | pancreatic neuroendocrine tumours",
"value": 44,
"units": "percent of tumours",
"n": 68,
"universe": "10 exome-sequenced plus 58 screened panNETs, Jiao 2011",
"null": null,
"note": "somatic inactivating; the most frequently mutated gene in the tumour type"
},
{
"quantity": "mutation_frequency",
"subject": "DAXX or ATRX | pancreatic neuroendocrine tumours",
"value": 43,
"units": "percent of tumours",
"n": 68,
"universe": "as above",
"null": null,
"note": "the second most frequently mutated; combined across the two subunits"
},
{
"quantity": "mutation_frequency",
"subject": "ATRX or DAXX | paediatric glioblastoma",
"value": 31,
"units": "percent of tumours",
"n": 48,
"universe": "paediatric GBM exomes, Schwartzentruber 2012",
"null": null,
"note": "100% among the G34R/G34V H3.3-mutant subgroup"
},
{
"quantity": "alt_positive_fraction",
"subject": "pancreatic neuroendocrine tumours",
"value": 61,
"units": "percent of tumours",
"n": null,
"universe": "panNET cohort, Heaphy 2011, telomere FISH morphology",
"null": null,
"note": "every ALT-positive tumour carried an ATRX or DAXX lesion"
},
{
"quantity": "pubmed_record_count",
"subject": "\"alternative lengthening of telomeres\"[tiab] AND KMT2A[tiab]",
"value": 0,
"units": "records",
"n": null,
"universe": "PubMed titles and abstracts, 2026-09-24",
"null": 921,
"note": "null column is the bare-phrase operand control; MLL gives 0 and leukemia gives 105"
}
]- This is the strongest argument in the beat against removing DAXX or ATRX therapeutically: loss of exactly these proteins is the canonical human route to alternative lengthening of telomeres, documented in 61% of one tumour type. An intervention against DAXX risks handing the blast a telomerase-independent immortality mechanism, which is the opposite of the intended effect and would not show up in a two-week viability assay.
- Both lesions are associated with BETTER prognosis in panNET, which means the human data on losing these proteins comes from tumours that behave comparatively well. That is reassuring about toxicity and says nothing useful about efficacy.
- Germline ATRX loss causes an intellectual disability and alpha-thalassaemia syndrome, so partial loss of this arm has a known human developmental phenotype - a serious consideration for a drug given to infants.
- Do KMT2A-rearranged leukaemia cells use ALT? Nobody has measured it in any KMT2A-rearranged line or patient sample. A C-circle assay across the thirteen B-ALL DepMap lines would simultaneously test the analogy and explain, or eliminate, the deep DAXX group in this beat's dependency analysis.
- Has any panNET or glioma cohort been examined for a KMT2A rearrangement, or any KMT2A-rearranged leukaemia cohort for a somatic DAXX or ATRX mutation? The co-occurrence has never been looked for in either direction, and paediatric ALL exomes exist in dbGaP.
- MEN1 loss and DAXX/ATRX loss are both associated with BETTER survival in panNET. Is there any setting in which losing a chromatin dependency improves outcome in leukaemia, and does that say anything about what a complete rather than partial menin blockade would do?
- Why is the H3.3 lesion in glioma in the histone and in panNET in the chaperone? If the two are interchangeable routes to the same state, an H3F3A mutation should be findable in some panNET and an ATRX-wild-type route should exist in glioma - and the answer bears on whether a leukaemic cell could reach the same state through its fusion alone.
- E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
In exomic sequencing of 10 non-familial pancreatic neuroendocrine tumours followed by screening of the most commonly mutated genes in 58 more, 44% carried somatic inactivating mutations in MEN1 and 43% carried mutations in DAXX or ATRX, making these the two most frequently mutated genes in the tumour type; mutations in both were associated with better prognosis.
DAXX/ATRX, MEN1, and mTOR pathway genes are frequently altered in pancreatic neuroendocrine tumors. · Jiao Y, Shi C, Edil BH, et al. · Science 2011What this evidence cannot show
Abstract read from efetch 2026-09-24; full text not read. n = 68 tumours total, of which only 10 were exome-sequenced and 58 were screened on the genes the first 10 nominated, so the 43-44% figures are from a targeted screen following a small discovery set and the discovery step could have missed a gene commoner than either. 2011 sequencing depth and calling. PanNET is an adult tumour and the survival association is from this cohort alone. The paper says nothing about leukaemia and the comparison to KMT2A-rearranged disease is entirely this entry's.
Evidence that argues against this (1)
- The prognostic direction has not held up uniformly in later cohorts, and the biologically deflating reading of the co-mutation is that MEN1 and DAXX/ATRX are simply the two chromatin genes a quiet-genome neuroendocrine tumour has room to lose, rather than partners in one mechanism. The abstract itself frames them as separate findings - 'proteins implicated in chromatin remodeling' - and asserts no interaction between them. Heaphy CM et al. 2011
- E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
Exome sequencing of 48 paediatric glioblastomas found somatic mutations in the H3.3-ATRX-DAXX chromatin remodelling pathway in 44% of tumours, with recurrent H3F3A mutations producing K27M or G34R/G34V substitutions in 31%, ATRX or DAXX mutations in 31% overall and in 100% of tumours carrying a G34R or G34V H3.3 mutation, and TP53 mutations in 86% of samples with H3F3A and/or ATRX mutations.
Driver mutations in histone H3.3 and chromatin remodelling genes in paediatric glioblastoma. · Schwartzentruber J, Korshunov A, Liu XY, et al. · Nature 2012What this evidence cannot show
Abstract read from efetch 2026-09-24; full text not read. n = 48 exomes with a 784-tumour validation screen. Glioblastoma is not a haematopoietic disease and no part of this comparison is an experiment. The '100% of G34 tumours' figure is over a small subgroup of 48. This entry uses the paper for the fact that the histone and its chaperone are alternative lesions in the same pathway, which the abstract states explicitly.
Evidence that argues against this (1)
- The glioma analogy breaks at the most load-bearing point. There, ATRX/DAXX mutation co-occurs with a mutation in the HISTONE ITSELF and with TP53 loss in 86% of cases, so the chaperone lesion arrives inside a specific two- or three-hit context. KMT2A-rearranged infant leukaemia has neither an H3F3A mutation nor, characteristically, TP53 loss at diagnosis, so the context that makes chaperone loss oncogenic in glioma is absent here and the pathway being 'the same' is a statement about gene names. Jiao Y et al. 2011
- E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
61% of pancreatic neuroendocrine tumours display the telomere abnormalities characteristic of alternative lengthening of telomeres, and every tumour exhibiting them carried an ATRX or DAXX mutation or had lost nuclear ATRX or DAXX protein; ATRX mutations also correlate with abnormal telomeres in central nervous system tumours.
Altered telomeres in tumors with ATRX and DAXX mutations. · Heaphy CM, de Wilde RF, Jiao Y, et al. · Science 2011What this evidence cannot show
Abstract read from efetch 2026-09-24; full text not read. The telomere readout is a FISH-based morphological criterion, not a C-circle assay, and 'characteristic of ALT' is an inference from appearance. The direction established is chaperone-loss to ALT; nothing here shows the converse, and nothing shows what happens when a chaperone is acutely depleted in a cell that already uses telomerase, which is the situation in a leukaemic blast.
Evidence that argues against this (1)
- ALT is rare in haematological malignancy and essentially all acute leukaemias are telomerase-positive, so the telomere prediction may simply not apply: a blast with active telomerase has no selective pressure to engage ALT when its chaperone is removed, and the phenotype of DAXX loss there could be nothing to do with telomeres. This is why the proposed C-circle assay is framed as a test that can eliminate the explanation rather than as one expected to confirm it. PubMed (NCBI E-utilities) 2026
- E4established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database
As of 2026-09-24, PubMed returns 0 records for '"alternative lengthening of telomeres"[tiab] AND KMT2A[tiab]', 0 for the same against MLL[tiab] and 0 against '"acute lymphoblastic leukemia"[tiab]', against 921 records for the bare term and 105 for the term against leukemia[tiab]; every query returned a non-empty querytranslation, so these are genuine zeros and not instrument failures.
Bounded PubMed absence for alternative lengthening of telomeres in KMT2A-rearranged and lymphoblastic leukaemia, run through scripts/pubmed_safe.py · PubMed (NCBI E-utilities), via this project's pubmed_safe wrapper 2026What this evidence cannot show
These are counts of TITLES AND ABSTRACTS. A telomere-length measurement reported only in a supplementary figure of a leukaemia paper is invisible to this query, which is the exact failure rule 80 in this project's brief documents. The operand controls are clean: 921 for the bare phrase, 105 against leukemia, 2103 for bare ATRX, so the instrument can return a one. Querytranslations were read back for every query. Aliases not searched: 'ALT' alone, which is unusable as a token, and 'telomerase-independent'. The bare term against leukemia returns 105 records which were NOT title-checked; the zero claimed here is only for the KMT2A, MLL and ALL conjunctions.
Evidence that argues against this (1)
- This zero is bounded by a two-term conjunction and by abstracts. The much larger `"alternative lengthening of telomeres"[tiab] AND leukemia[tiab]` set of 105 records was not read, and a paediatric ALL cohort with telomere measurements sitting inside those 105 would refute the absence without ever naming KMT2A in its abstract. This is the same shape as the SET::NUP214 H3K79 case in this project's brief, where a clean four-spelling zero was false because the measurement was in Figure 6A. PubMed (NCBI E-utilities) 2026
supplies the DAXX/ATRX co-mutation and the ALT consequence, neither of which that entry contains
MEN1 is the top mutated tumour suppressor in panNET and the required chromatin anchor in KMT2A-r leukaemia - the same gene with an inverted sign
DAXX/ATRX mutated in 43% of panNET and 31% of paediatric GBM, and wild type but apparently required here
- analogous to →ATRX - DAXX's partner and the other arm of H3.3 deposition; in this disease its only signal is in the four lymphoid cell lines
ATRX loss is the canonical human route to ALT, documented in 61% of panNET; never measured in this disease
- analogous to →Histone H3.3 - one protein, two genes, and the reason this graph should not file it as gene-h3f3a
three routes to one axis: mutate the histone (glioma), delete the chaperone (glioma, panNET), fuse the writer (leukaemia)
generated by the panNET/glioma analogy: ATRX/DAXX loss is the canonical human route to ALT
the ALT bite-back that blocks this target class comes from the panNET and glioma data in that entry