Oncofusion

start / graph / gene-cdkn2a

Genemoderate confidence 3 of 5: One strong study, or several smaller consistent ones.Q46-cooperating-lesions

CDKN2A - one gene, two reviewed UniProt records with the same primary symbol, and a deletion that is three times rarer here than in the infant leukaemias next door

also called CDKN2A · CDKN2 · MTS1 · MLM · p16INK4a · p16-INK4 · CDK4I · p14ARF · ARF · P42771 · Q8N726

CDKN2A is the brake on cell division that is deleted in a large share of childhood leukaemias. In infant KMT2A leukaemia it is deleted about three times less often than in infant leukaemias without the fusion - one more sign that this disease does not need the usual extra hits.

The same thing, for a scientist

CDKN2A is deleted in about **8.5% (4/47)** of infant KMT2A-rearranged ALL, against **28% (5/18)** of infant ALL *without* a KMT2A rearrangement (P = 0.058), and the paper notes the non-KMT2A infant figure approaches what is seen in standard non-Philadelphia childhood ALL [E1]. In the 22-case whole-genome discovery cohort CDKN2A/CDKN2B was the second most frequently affected gene among the structural events not caused by the rearrangement itself, in 3 cases, behind PAX5 in 5 [E1]. In **non-infant** KMT2A-rearranged disease the split is by lineage: CDKN2A/B deletions in **3 of 9 (33%)** ALL cases and **none** of the AML cases [E1]. So CDKN2A in this disease is a **comparative** finding rather than a frequency finding. It is present, it is not rare enough to call absent, and its interest is that it is **depleted relative to the leukaemias it sits beside** - which is the pattern a genome that does not need a second hit should produce. This entry also records an identifier trap that the project's rules predict but had not yet met in this form.

What it looks like
loading AF-P42771-F1…
AlphaFold AF-P42771-F1 ↗
colourshape

This is a prediction, not a measurement. No one has seen this protein in this shape. The colours say how much the software trusts itself, and that is worth reading before anything else — the orange stretches are places the model is effectively saying this part has no fixed shape.

very high (90+)confident (70–90)low (50–70)very low (<50)
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
Two reviewed records, both with primary gene name CDKN2A

gene_exact:CDKN2A AND organism_id:9606 AND reviewed:true, requesting gene_primary and length, returns two rows:

accessionprimarysynonymsprotein namelength
P42771CDKN2ACDKN2 MTS1Cyclin-dependent kinase inhibitor 2A (p16-INK4a) (CDK4I) (MTS-1)156
Q8N726CDKN2ACDKN2 MLMTumor suppressor ARF (Alternative reading frame) (p14ARF)132

The project's standing rule (rule 62) is: keep only the row whose primary name equals the symbol asked for; if more than one does, stop and look. Here both do, and the rule fires.

Looking is what resolves it, and the answer is biological rather than an error. The CDKN2A locus encodes two proteins from overlapping exons read in different frames: p16INK4a, which inhibits CDK4/6 and holds RB1 hypophosphorylated, and p14ARF, which stabilises p53 by sequestering MDM2. They share no amino-acid sequence. They are not isoforms in the ordinary sense and they are not paralogues; they are two proteins from one gene.

Consequences for anything written about this gene:

  1. A statement about 'CDKN2A' in a leukaemia paper is almost always about a genomic deletion, which removes both products, so the ambiguity usually does not matter clinically - and that is exactly why it is dangerous in a protein-level context, where it does.
  2. data.uniprot in this entry is P42771 and data.length_aa is 156, i.e. the p16INK4a product, because that is the protein the cell-cycle claim in this disease refers to. Q8N726 is recorded in data.second_reading_frame so that a reader who needs ARF is not silently handed p16.
  3. Any future entry making a p14ARF claim must assert Q8N726 / 132 aa and must not inherit this entry's accession. The lengths differ by 24 residues, which is enough for verify-accessions.py to catch a swap but small enough that nobody would notice it by eye.

This is a fourth shape alongside the three the brief already logs: not a synonym match (TCF3 to TCF7L1), not an alias pointing at a paralogue (CSNK2A1 in CSNK2A2), not swapped protein names (KMT2B/KMT2D). It is two distinct proteins legitimately sharing one primary gene symbol.

The comparison that carries the finding
populationCDKN2A/B deletionsource
infant KMT2A-rearranged ALL4/47 (8.5%)1
infant ALL without KMT2A rearrangement5/18 (28%)1
comparisonP = 0.0581
non-infant KMT2A-rearranged ALL3/9 (33%)1
non-infant KMT2A-rearranged AML01

Three readings, in decreasing confidence:

Solid. CDKN2A deletion is present in infant KMT2A-rearranged ALL at a low single-figure-to-8.5% rate. It is not absent, and entries elsewhere in this graph that list CDKN2A among the things this disease does not have should say 'not recurrent' rather than 'absent'.

Suggestive. It is depleted relative to infant ALL without the fusion. P = 0.058 on 4/47 versus 5/18 is a trend, not a result, and 18 comparator cases is a thin denominator.

Speculative and worth stating anyway. The age split within KMT2A-rearranged ALL - 8.5% in infants, 33% in older children - runs the same direction as everything else in this beat: older KMT2A-rearranged disease accumulates the ordinary co-lesions and infant disease does not.

Why a CDKN2A deletion would matter if it were common

CDKN2A loss removes the INK4a brake on CDK4/6 and, through the ARF reading frame, weakens p53 stabilisation - so a single deletion attacks both the RB and the p53 axes at once. In B-ALL generally it is associated with relapse risk, and CDK4/6 inhibitors are the obvious pharmacological response.

In this disease the argument does not get started, for a reason that is the beat's recurring theme: at 8.5%, a CDK4/6-inhibitor strategy selected on CDKN2A status would address fewer than one in ten patients, and no study has reported the clonality of these deletions - so it is not even known whether the 8.5% is 8.5% of patients with a clonal loss or a mixture including subclonal events.

This graph already holds the relevant negative from the other direction: TP53 is not part of de novo infant KMT2A-rearranged ALL (gene-tp53). If ARF loss were the route by which this disease disabled p53, CDKN2A deletion would be expected at a higher rate than 8.5%. It is not, and the simplest reading is that this leukaemia does not disable p53 at diagnosis by any route.

Facts and figures
uniprotP42771
gene symbolCDKN2A
length AA156
alphafold IDAF-P42771-F1
cytoband9p21.3
resolution notegene_exact:CDKN2A AND organism_id:9606 AND reviewed:true with fields=accession,gene_primary,gene_synonym,protein_name,length returned TWO rows, both with primary gene name CDKN2A: P42771 (156 aa, p16-INK4a) and Q8N726 (132 aa, p14ARF). Fetched from the UniProt REST API on 2026-09-20. This entry asserts P42771/156 because its cell-cycle claim concerns p16INK4a; a p14ARF claim must assert Q8N726/132 independently.
domains noteUniProt P42771 carries no Domain features; the four ankyrin repeats are annotated as Repeat features. No domain is recorded rather than a guessed boundary.
druggabilityCDKN2A loss is the canonical rationale for CDK4/6 inhibition, but at 8.5% of patients, with no clonality data and no published CDK4/6-inhibitor experiment in KMT2A-rearranged infant ALL, the rationale is not actionable in this disease today.
subnuclear localizationp14ARF is NUCLEOLAR, held there by phase separation with NPM1 and released by DNA damage or by nucleus-translocated cytochrome c. p16INK4a has no published subnuclear compartment. Writing 'CDKN2A is nucleolar' or 'CDKN2A is nucleoplasmic' without naming the frame is wrong in one direction or the other.
shown for fusionNO, for either frame. No localisation, fractionation or imaging of p16INK4a or p14ARF has been performed in a KMT2A-rearranged cell in any source read for this beat. The absence is consequential rather than clerical: CDKN2A deletion is one of the commonest cooperating lesions recorded in this graph, a genomic deletion at 9p21 usually removes BOTH frames, and the two products are lost from two different compartments with two different consequences - a CDK4/6-inhibitor-relevant nucleoplasmic brake and a nucleolar, NPM1-tethered p53 arm. An entry that records 'CDKN2A deleted' without the frame understates the lesion. Zero in titles and abstracts; supporting figures not searched.
identifier trapTWO TRAPS, one already registered and one new. (1) Registered: CDKN2A returns TWO reviewed human records - P42771 (156 aa, p16INK4a) and Q8N726 (132 aa, Tumor suppressor ARF/p14ARF) - both carrying CDKN2A as primary gene, sharing no sequence because they are two reading frames of one locus; rule 62's 'stop when more than one matches' is a false alarm here. Re-verified live against UniProt REST 2026-09-21. (2) NEW, and it is a rule-74 antibody case: the IVD-approved anti-p16INK4a clone E6H4 can produce a distinctive STRONG NUCLEAR WITH NUCLEOLAR pattern that is easily confused with the HPV-associated block-positivity pattern, and that pattern is caused by frameshift deletions or insertions in exon 2 of the INK4a/ARF locus which convert the reading frame of the 3-prime p14ARF sequence into p16INK4a, producing a p14ARF::p16INK4a CHIMERIC PROTEIN. So an anti-p16 antibody, on a p16-inactivating mutation, reports a nucleolar signal that is neither p16 nor ARF. The clone is the identifier, not the label.
epigenetic readoutTARGET OF A MARK, NOT A WRITER OF ONE - and the two halves of that mark story have OPPOSITE evidential standing, which is the finding. CDKN2A encodes p16INK4a (P42771, 156 aa) and, in a second reading frame, p14ARF (Q8N726, 132 aa); neither protein carries a SET, JmjC, HAT, HDAC, bromo, chromo, Tudor, PHD, CXXC or DNMT module, so nothing is written or read by this locus. What is written ON it: (1) H3K27me3, AND THE MEASUREMENT IS AT THE LOCUS IN A KMT2A-FUSION BACKGROUND. In MURINE MLL/ENL-transduced cells, ChIP places EZH2 highly enriched around the p16 transcription start site together with H3K27 methylation marks, and - the part that makes it a specificity claim rather than a housekeeping one - the same enrichment is ABSENT in E2A/HLF-transduced cells; EZH2 inhibition by DZNep or shRNA reduces leukaemia-initiating-cell frequency, p16 knockdown cancels the survival benefit, and Hoxa9 levels and p16 levels are NOT inversely related between MLL/ENL and Hoxa9/Meis1 cells, so a Hoxa9-independent route to p16 repression must also exist [E_q97_epi1]. Independently, in MURINE MLL-ENL AML, Hhex binds the Cdkn2a locus and directly interacts with PRC2 to enable H3K27me3-mediated repression of both frames [E_q97_epi2]. And pharmacologically, the dual EZH2/EZH1 inhibitor UNC1999 derepresses Cdkn2a in MLL-rearranged cells with a DECREASE in H3K27me3 and a CONCURRENT GAIN in H3K27ac, preferentially at distal enhancers, with the inactive analogue UNC2400 as the off-target control [E_q97_epi3]. (2) DNA METHYLATION, AND THE BRIEF THAT SENT ME HERE OVERSTATED IT. Promoter methylation of p16 is common in the right lineage - 34.4% of 61 children and 26.7% of 30 adults with B-lineage ALL, against p14 at 14.8%/10.0% and p15 at 34.4%/40.0% - but that study explicitly found NO significant association between methylation and any cytogenetic group, MLL rearrangement included, and no association between deletion and methylation [E_q97_epi4]. So DNA methylation of this locus is a feature of childhood B-ALL generally and is NOT, on published evidence, a KMT2A-fusion-linked event. The H3K27me3 arm is the fusion-linked one and it is murine and MYELOID (rule 73). WHY THIS MATTERS HERE: gene-bmi1 E4 in this graph argues that because CDKN2A is deleted in only ~8.5% of infant KMT2A-rearranged ALL, any repression of the p16/p14 brake in those patients must be epigenetic rather than genomic. That inference is supported for H3K27me3 only by murine myeloid MLL-fusion models, and the one human B-lineage methylation study available declines to associate methylation with MLL rearrangement. NOBODY HAS MEASURED H3K27me3 OR PROMOTER METHYLATION AT CDKN2A IN A HUMAN KMT2A::AFF1 INFANT BLAST.
second reading frame
{
  "accession": "Q8N726",
  "length_aa": 132,
  "protein_name": "Tumor suppressor ARF",
  "common_name": "p14ARF",
  "localization": "nucleolar, sequestered by NPM1 via phase separation",
  "note": "the entry asserts P42771/156 for p16INK4a per the registry rule; the ARF frame is recorded here rather than as a second accession, and the localisation difference between the two frames is the substance of this amendment"
}
domains
[]
cooperating lesion profile
{
  "disease": "infant KMT2A-rearranged B-ALL",
  "lesion_class": "genomic deletion of CDKN2A/CDKN2B, removing both p16INK4a and p14ARF",
  "frequency": "4/47 (8.5%); 3 of the 22 whole-genome cases",
  "comparator": "5/18 (28%) in infant ALL without KMT2A rearrangement, P = 0.058",
  "age_split_within_kmt2a_r": "3/9 (33%) in non-infant KMT2A-rearranged ALL; 0 in non-infant KMT2A-rearranged AML",
  "clonality": "not reported by any study read for this entry",
  "verdict": "present but not recurrent, and depleted relative to the neighbouring leukaemias - the comparison is the finding, not the rate"
}
localization
{
  "subcellular": "Nucleus - for BOTH products, but they are two different proteins from two reading frames of one locus and they do NOT share a compartment. p16INK4a (P42771, 156 aa) is nucleoplasmic and nuclear-cytoplasmic in tissue, and its canonical diagnostic readout is 'block positivity', strong nuclear or nuclear-cytoplasmic staining. p14ARF (Q8N726, 132 aa) is NUCLEOLAR: it is sequestered in nucleoli by nucleophosmin/NPM1, and that sequestration is what holds it inactive under non-stress conditions.",
  "subnuclear": "p14ARF: NUCLEOLUS, and the mechanism of residence is measured. p14ARF assembles into a gel-like meso-scale network upon phase separation with NPM1, mediated by hydrophobic residues in an alpha-helix and beta-strands in a partially folded N-terminal portion; these contacts enhance p14ARF NUCLEOLAR PARTITIONING, restrict NPM1 diffusion within condensates and nucleoli, and reduce cellular proliferation. Release from the nucleolus is a regulated event: DNA damage shifts ARF out of NPM, and nucleus-translocated mitochondrial cytochrome c binds nucleolar NPM, driving an extended-to-compact conformational change that liberates ARF by changing nucleolar liquid-liquid phase separation. p16INK4a: NO subnuclear compartment is assigned in any source read for this beat - and a nucleolar p16 IHC pattern means something else entirely (see identifier_trap).",
  "how_shown": "NMR and structural work plus in-cell partitioning and diffusion measurements for the ARF-NPM1 assembly (PMID 39528457); nucleolar LLPS and conformational analysis for cytochrome c-driven ARF release (PMID 36220893); diagnostic immunohistochemistry with clone E6H4 plus sequencing for the chimeric-protein pattern (PMID 41339993)",
  "systems": "human cell lines and in vitro reconstitution; human tumour tissue for the IHC - no KMT2A-rearranged system",
  "node": "localization-nucleolus-kmt2a-fusion"
}
localization counted absence
{
  "query": "(p14ARF OR \"p14(ARF)\" OR \"ARF tumor suppressor\")[tiab] AND (nucleolus OR nucleolar OR localiz* OR nucleoplasm*)[tiab]",
  "records": 174,
  "records_in_a_kmt2a_context": 0,
  "operand_controls": {
    "(p16INK4a OR \"p16(INK4a)\")[tiab] AND (localiz* OR nuclear OR cytoplasm* OR nucleolar)[tiab]": 701
  },
  "date": "2026-09-21",
  "tool": "PubMed E-utilities esearch",
  "bound": "zero in titles and abstracts; supporting figures not searched"
}
depmap kmt2a selectivity
{
  "chronos_mean_kmt2a_rearranged": 0.2737,
  "chronos_mean_other_leukaemia": 0.0968,
  "difference": 0.177,
  "pan_essentiality_all_lines": 0.1597,
  "selectivity_rank": 17552,
  "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."
}
cell type
{
  "evidence_class_key": "concept-expressed-required-maintains-three-evidence-classes",
  "two_frames_kept_distinct": "P42771 / 156 aa is p16INK4a; Q8N726 / 132 aa is p14ARF (murine p19Arf), read from the same locus in an alternative frame and sharing no sequence. Both frames appear in the cell-type result below and they behave DIFFERENTLY, which is the reason the distinction is kept here rather than collapsed.",
  "expressed_in": {
    "class": 1,
    "summary": "both frames are repressed in healthy young haematopoietic stem cells and are DE-REPRESSED when polycomb repression fails. They are not constitutively expressed genes whose level marks a cell type; they are silenced genes whose de-repression marks a cell STATE.",
    "pmids": [
      "16954369"
    ]
  },
  "required_in": {
    "class": 2,
    "claim": "NOT APPLICABLE as usually written - CDKN2A is a tumour suppressor and the relevant measurement is the effect of its LOSS, not of its requirement. In MICE, de-repressed p16Ink4a and p19Arf in Bmi1-deficient animals are tightly associated with LOSS OF SELF-RENEWING HAEMATOPOIETIC STEM CELLS, and deleting both Ink4a and Arf substantially restores the self-renewal capacity of Bmi1-/- HSCs [E_q69_ct1].",
    "pmids": [
      "16954369"
    ],
    "system_named_in_the_claim": "murine Bmi1-deficient haematopoietic stem cells - normal haematopoiesis under a polycomb lesion, not leukaemia and not human",
    "the_cell_type_answer": "the cell type in which the CDKN2A locus is decisive is the SELF-RENEWING HSC. That is the same compartment the Bmi1 thread in this graph is about, and the same compartment cell-state-stem-like-self-renewal names."
  },
  "not_required_in": {
    "claim": "The DepMap line for this gene points the WRONG WAY BY CONSTRUCTION and must not be read as a selectivity result. Chronos is POSITIVE in both arms - +0.2737 in KMT2A-rearranged lines and +0.0968 in other leukaemias - i.e. knocking CDKN2A out makes cells grow BETTER, which is what a tumour suppressor does. The nominal rank of 17,552 of 17,728 records that the growth advantage is LARGER in the KMT2A-rearranged arm, not that CDKN2A is 'anti-selective' in the sense the scan's ranking was built for.",
    "category_note": "counting this gene in an 'anti-selective half' tally is a category error: the scan ranks DEPENDENCIES and this is not one. The correct reading of a positive difference on a tumour suppressor is that the KMT2A-rearranged lines gain MORE from losing it - consistent with the well-documented enrichment of CDKN2A deletion at relapse in this disease, which the entry's cooperating_lesion_profile already records."
  },
  "nodes": [
    "cell-type-hsc-bone-marrow"
  ],
  "context": "murine Bmi1-deficient HSCs for the cell-type result; murine craniofacial development for the TASP1 link; DepMap cancer lines for the growth-advantage measurement"
}
cell state
{
  "evidence_class_key": "concept-expressed-required-maintains-three-evidence-classes",
  "maintains_state": {
    "class": 3,
    "states": [
      "cell-state-stem-like-self-renewal"
    ],
    "claim": "p16Ink4a and p19Arf ENFORCE EXIT from the self-renewing state, and polycomb exists in part to hold them off. Bmi1 regulates HSCs by acting as a critical failsafe against p16Ink4a- and p19Arf-dependent PREMATURE LOSS of HSCs; deleting both frames substantially restores Bmi1-/- HSC self-renewal [E_q69_ct1]. The state CDKN2A sets is therefore the NEGATIVE of self-renewal.",
    "pmids": [
      "16954369"
    ],
    "system_named_in_the_claim": "murine Bmi1-deficient haematopoietic stem cells"
  },
  "regulates_state": {
    "class": 3,
    "claim": "The Cdkn2a locus is also the read-out of a TASPASE1-dependent proliferative state. In MICE, taspase1-mediated cleavage of the general transcription factor TFIIA maintains LIMITED transcription of p16Ink4a and p19Arf, non-cleavable TFIIA accumulates at both promoters and drives their transcription, and compound Cdkn2a deficiency - especially p16Ink4a deficiency - markedly reduces the craniofacial anomalies of TASP1-deficient mice [E_q69_cs1].",
    "pmids": [
      "25664857"
    ],
    "why_it_is_recorded_here": "TASP1 is in this same tranche and is the protease that cleaves KMT2A. A regulatory circuit running TASP1 -> TFIIA -> Cdkn2a -> proliferation connects two of this beat's genes through a mechanism that has nothing to do with KMT2A cleavage, which is the interesting part: the cell-cycle arm of Taspase1 biology is at least partly MLL-independent."
  },
  "bounded_query": {
    "date": "2026-09-21",
    "tool": "NCBI esearch db=pubmed",
    "query": "Ink4a[ti] AND (\"stem cell\" OR haematopoietic OR hematopoietic)[ti] AND 2006[dp]",
    "count": 2,
    "on_topic_after_title_check": 1,
    "wider_shape": "p16INK4a[tiab] AND (\"stem cell ageing\" OR \"stem cell aging\" OR senescence)[tiab] AND (hematopoie* OR haematopoie*)[tiab] returns 72 records; the year-bounded title query above was used to reach the specific HSC self-renewal result rather than the large ageing literature.",
    "bound": "titles and abstracts only; supporting figures not searched (brief rule 80)"
  },
  "confidence": "moderate",
  "caveats": "Both results are murine and neither is a leukaemia. The Bmi1 study measures the locus's effect under a polycomb lesion - it does not establish what p16/p19 do in a WILD-TYPE HSC, only what happens when the repression that normally silences them is removed. The Taspase1 study is craniofacial development, not blood. No measurement of either frame's protein exists in a KMT2A-rearranged cell, which this entry's shown_for_fusion field already records as an explicit NO for both frames.",
  "counter_evidence": {
    "note": "Deleting Ink4a and Arf does NOT rescue the Bmi1-deficient phenotype fully. The bone marrow microenvironment of Bmi1-/- mice is severely defective at supporting haematopoiesis, and deleting both frames does not considerably restore it, leading to sustained postnatal HSC depletion in Bmi1-/-Ink4a-Arf-/- mice. So the CDKN2A locus accounts for the cell-autonomous half of the phenotype and not the niche half - and a therapeutic or mechanistic reading that treats p16/p19 as the whole of what polycomb restrains in this compartment is contradicted within the same paper.",
    "citation": {
      "title": "Differential impact of Ink4a and Arf on hematopoietic stem cells and their bone marrow microenvironment in Bmi1-deficient mice.",
      "authors": "Oguro H, Iwama A, Morita Y, Kamijo T, van Lohuizen M, Nakauchi H",
      "year": 2006,
      "venue": "J Exp Med",
      "pmid": "16954369",
      "doi": "10.1084/jem.20052477",
      "type": "paper"
    }
  }
}
epigenetic bounded absence
{
  "claim": "No study names CDKN2A and a chromatin mark or DNA methylation together under the CURRENT HGNC symbol for the fusion gene, and the H3K27me3-at-CDKN2A literature in leukaemia is two papers wide.",
  "date_measured": "2026-09-21",
  "tool": "NCBI esearch db=pubmed (eutils), querytranslation read back from every response",
  "queries": [
    {
      "query": "CDKN2A[tiab] AND methylation[tiab] AND KMT2A[tiab]",
      "querytranslation": "\"CDKN2A\"[Title/Abstract] AND \"methylation\"[Title/Abstract] AND \"KMT2A\"[Title/Abstract]",
      "count": 0,
      "errorlist_note": "warninglist outputmessages [No items found.]; quotedphrasesnotfound empty, so all three terms are indexed"
    },
    {
      "query": "p16[tiab] AND methylation[tiab] AND MLL[tiab]",
      "querytranslation": "\"p16\"[Title/Abstract] AND \"methylation\"[Title/Abstract] AND \"MLL\"[Title/Abstract]",
      "count": 7,
      "role": "THE CONTROL, AND IT IS THE FINDING. Same shape, retired symbol plus protein name, returns 7 where the approved-symbol form returns 0. Titles read: PMID 19837270 (the B-ALL deletion/methylation study cited here), 24612037 (the EZH2/p16 LSC study cited here), 20113419, 12970785, 11167824, 10706859 and 23669186 - the last a gastric-cancer methylation review, i.e. MLL bleeding in as a non-gene string."
    },
    {
      "query": "CDKN2A[tiab] AND H3K27me3[tiab] AND leukemia[tiab]",
      "querytranslation": "\"CDKN2A\"[Title/Abstract] AND \"H3K27me3\"[Title/Abstract] AND \"leukemia\"[Title/Abstract]",
      "count": 2,
      "title_checked": "PMID 26728554 (Hhex/PRC2/Cdkn2a in MLL-ENL AML) and 25395428 (UNC1999 in MLL-rearranged leukaemia). BOTH on topic, both cited here, both MURINE."
    },
    {
      "query": "CDKN2A[tiab] AND H3K27me3[tiab] AND MLL[tiab]",
      "querytranslation": "\"CDKN2A\"[Title/Abstract] AND \"H3K27me3\"[Title/Abstract] AND \"MLL\"[Title/Abstract]",
      "count": 2,
      "role": "identical pair - the leukaemia and MLL clauses select the same two records"
    },
    {
      "query": "p16[tiab] AND H3K27me3[tiab] AND lymphoblastic[tiab]",
      "querytranslation": "\"p16\"[Title/Abstract] AND \"H3K27me3\"[Title/Abstract] AND \"lymphoblastic\"[Title/Abstract]",
      "count": 0,
      "role": "THE B-LINEAGE ARM IS EMPTY. Positive control through the identical shape: p16[tiab] AND H3K27me3[tiab] returns 77, and CDKN2A[tiab] AND H3K27me3[tiab] returns 78, so the instrument reaches this mark at this locus in quantity - it is the lymphoblastic scoping that empties it."
    },
    {
      "query": "CDKN2A[tiab] AND \"bivalent\"[tiab]",
      "querytranslation": "\"CDKN2A\"[Title/Abstract] AND \"bivalent\"[Title/Abstract]",
      "count": 5,
      "role": "run because data/symbol-collisions.json records that bivalent means a two-motif protein interaction in the menin subfield; here none of the 5 is a menin-pocket paper, so the contamination did not fire on this pair. Not counted as support either - unread beyond titles."
    }
  ],
  "symbols_searched": [
    "CDKN2A",
    "p16",
    "KMT2A",
    "MLL",
    "lymphoblastic",
    "leukemia"
  ],
  "symbols_not_searched": [
    "p14ARF",
    "INK4A",
    "MTS1",
    "CDKN2B",
    "p15",
    "MLL1",
    "ALL1",
    "HRX"
  ],
  "absence_class": "never-measured",
  "absence_class_basis": "Bounded by titles and abstracts only; supporting figures not searched (rule 80), and an H3K27me3 ChIP at one locus is exactly the kind of supporting panel rule 80 says an abstract index cannot see. The zero that matters - no human B-lineage measurement - is therefore a floor, not a proof.",
  "reopening_condition": "Any H3K27me3 or EZH2 ChIP/CUT&RUN track at the human CDKN2A locus in a KMT2A-rearranged B-ALL sample or line, or any 450K/EPIC methylation call at CDKN2A stratified by KMT2A status, reopens this. GSE127508, already in this project's queue, is one place to look."
}
How this could bite us back
  • A CDK4/6 inhibitor selected on CDKN2A deletion would address under one in ten patients with this disease, and the deletion's clonality is unmeasured, so the addressable fraction could be smaller still.
  • CDKN2A deletion removes p14ARF as well as p16INK4a, so a CDK4/6 inhibitor given to a CDKN2A-deleted patient is acting in a cell whose p53 response is also weakened - the context in which cell-cycle arrest most readily becomes senescence escape rather than death.
  • Any protein-level claim about this gene inherits a two-record ambiguity in which both records carry the same primary symbol. Nothing in a returned record would warn a reader that they have the wrong one, and the length difference is 24 residues.
  • a deletion removes two proteins from two compartments - A 9p21 deletion usually removes both reading frames: a nucleoplasmic CDK4/6 brake (p16INK4a) and a nucleolar, NPM1-tethered p53 arm (p14ARF). A CDK4/6 inhibitor replaces only the first. Recording 'CDKN2A deleted' without naming the frame understates the lesion and overstates what palbociclib-class drugs restore.
  • the diagnostic readout can be a chimera - Anti-p16 clone E6H4 reports strong nuclear WITH NUCLEOLAR staining on exon-2 frameshifts, where the epitope sits on a p14ARF::p16INK4a chimeric protein. Any p16 IHC used to stratify or to confirm locus status must have the nucleolar pattern read out separately.
  • Any intervention that de-represses the CDKN2A locus in order to arrest leukaemic cells will hit self-renewing haematopoietic stem cells with the same lesion. In Bmi1-deficient mice de-repressed p16Ink4a and p19Arf cause premature LOSS of self-renewing HSCs, and deleting both frames restores self-renewal (PMID 16954369) - so the normal marrow's stem compartment is precisely the tissue this axis is built to eliminate. In infants that is an eighty-year liability.
  • The DepMap numbers on this entry cannot be quoted as a selectivity result. Chronos is POSITIVE in both arms (+0.2737 KMT2A-rearranged, +0.0968 other leukaemias): knocking CDKN2A out helps cells grow, as a tumour suppressor should. The nominal rank of 17,552 of 17,728 means the growth advantage is bigger in the KMT2A-rearranged arm, not that the gene is an anti-selective dependency, and counting it in an anti-selective tally is a category error.
  • The therapeutic logic here runs backwards from most of this graph: the intervention (an EZH1/2 inhibitor) works by RESTORING a tumour suppressor, so its efficacy depends on CDKN2A being intact - and CDKN2A is homozygously deleted in 11.5% of children and 30.0% of adults with B-ALL (PMID 19837270). Those patients cannot respond by this mechanism and the deletion is testable before dosing.
  • UNC1999 derepression comes with a gain of H3K27ac at distal enhancers (PMID 25395428), i.e. the drug opens enhancers genome-wide and not only at the intended brake. In a lineage where this graph already records a translocation hazard for EZH2 inhibition, a global enhancer-activating side effect is the wrong direction.
Still open
  1. Are the CDKN2A deletions in infant KMT2A-rearranged ALL clonal or subclonal? Nothing published says, and in this disease the answer has changed the interpretation of every other co-lesion.
  2. Is the depletion relative to non-KMT2A infant ALL real, or an age effect within the infant group? The index study reports that CNA burden tracked age independently of KMT2A status, which is a live confounder for the one comparison this entry rests on.
  3. Does ARF loss substitute for TP53 mutation in this disease? The rates argue against it - 8.5% CDKN2A deletion alongside no reported TP53 mutation - but no study has looked at p53 pathway status functionally in infant KMT2A-rearranged blasts.
  4. Has any CDK4/6 inhibitor been tested in a CDKN2A-deleted KMT2A-rearranged model? No such experiment was found in the searches run for this beat.
  5. In KMT2A-rearranged blasts with an intact 9p21 locus, is p14ARF nucleolar and NPM1-bound, and does the fusion or a menin inhibitor change that? The ARF arm is a p53 route that a CDK4/6 inhibitor does not touch and that a compartment assay could report on directly.
  6. Does the human CDKN2A locus carry H3K27me3 in a KMT2A::AFF1 infant blast? Every H3K27me3-at-Cdkn2a measurement in a KMT2A-fusion background is murine and myeloid, and this graph's argument that the p16/p14 brake is held down epigenetically in infants rests on that import.
  7. Is p16 promoter methylation enriched in KMT2A-rearranged childhood B-ALL specifically? The one study measuring it reports no association with any cytogenetic group, but on 61 children it could not have resolved a subgroup effect.
  8. What represses p16 in MLL/ENL cells that is NOT HOXA9? PMID 24612037 reports Hoxa9 and p16 levels failing to correlate inversely, which means the HOXA9-EZH2-p16 chain this graph implies is incomplete.
Every source behind this entry (12)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In the St. Jude infant ALL cohort, CDKN2A/B deletions were present in 4 of 47 (8.5%) infant MLL-rearranged cases against 5 of 18 (28%) infant non-MLL-rearranged cases (P = 0.058), the latter approaching the rate in standard non-Philadelphia childhood ALL; CDKN2A/CDKN2B was affected in 3 of the 22 whole-genome discovery cases, second only to PAX5 among structural events not caused by the MLL rearrangement; and in non-infant MLL-rearranged leukaemia CDKN2A/B deletions were found in 3 of 9 ALL cases (33%) and in none of the AML cases.

    The landscape of somatic mutations in infant MLL-rearranged acute lymphoblastic leukemias. · Andersson AK, Ma J, Wang J, et al. · Nature Genetics 2015
    PMID 25730765DOI
    What this evidence cannot show

    The comparator arm is 18 infant non-MLL-R cases and the non-infant ALL arm is 9 cases, so both contrasts rest on small denominators and P = 0.058 is explicitly not significant. The paper also reports that the higher CNA burden in infant non-MLL-R cases was associated with AGE and independent of MLL-R status (P = 0.005), which is a confounder for any comparison between the two infant arms. Copy number was assessed by SNP array plus WGS; no allele-specific or clonality analysis of the CDKN2A deletions is reported. Full text read via efetch db=pmc (PMC4553269).

    Evidence that argues against this (1)
    • The same study reports that the elevated CNA count in infant non-MLL-rearranged cases tracks age rather than MLL status, so the CDKN2A depletion in the MLL-rearranged arm may be partly an age effect within the infant group rather than a fusion effect. Andersson AK et al. 2015
  2. E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    A query for reviewed human UniProtKB records with gene_exact:CDKN2A returns two records, P42771 (156 aa, p16-INK4a) and Q8N726 (132 aa, tumour suppressor ARF / p14ARF), both carrying CDKN2A as their primary gene name.

    What this evidence cannot show

    Fetched 2026-09-20. This is a database fact about how UniProt models the CDKN2A locus and it is correct rather than a curation error: the two products are translated from overlapping exons in different reading frames and share no amino-acid sequence. It is recorded because the project's identifier rule tells a reader to stop when more than one row carries the primary symbol, and a reader who instead takes the first row gets p16 while believing they have 'CDKN2A'.

    Evidence that argues against this (1)
    • Almost all leukaemia literature reporting CDKN2A refers to a genomic deletion that removes both products, so in the clinical context the ambiguity is usually inconsequential - which is precisely why a protein-level claim built on the wrong record would attract no scrutiny. UniProt Consortium 2026
  3. E3weak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.paper

    In 185 adult KMT2A-rearranged AML patients with panel-based NGS, the independent prognostic factors for overall survival among intensively treated patients were age over 60, secondary AML, DNMT3A mutation and KRAS mutation - CDKN2A was not among them.

    PMID 38965370DOI
    What this evidence cannot show

    Cited as an absence from a named list, which is the weakest form of negative evidence and is graded accordingly. Panel-based NGS detects sequence variants; CDKN2A's lesion class is deletion, so this cohort may not have been able to detect CDKN2A loss at all - an absence of testing, not a negative result. Adults with AML, not infants with ALL. Abstract read.

    Evidence that argues against this (1)
    • A panel NGS study cannot exclude a copy-number lesion, so CDKN2A's absence from this prognostic model carries no information about whether CDKN2A deletion matters in adult KMT2A-rearranged AML. Andersson AK et al. 2015
  4. E_q65_loc1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    p14ARF - the second reading frame of CDKN2A - assembles into a gel-like meso-scale network upon phase separation with the nucleolar chaperone NPM1 through hydrophobic contacts in a partially folded N-terminal alpha-helix and beta-strands; these contacts enhance p14ARF's NUCLEOLAR PARTITIONING, restrict NPM1 diffusion within condensates and nucleoli, and reduce cellular proliferation. Under non-stress conditions NPM1 stabilises p14ARF in nucleoli, preventing its degradation and blocking p53 activation.

    p14ARF forms meso-scale assemblies upon phase separation with NPM1. · Gibbs E, et al. · Nat Commun 2024;15:9531 2024
    PMID 39528457DOI
    What this evidence cannot show

    Structural biology plus in-cell measurements in non-leukaemic systems; the phase-separation claim is supported by more than punctate appearance - restricted NPM1 diffusion inside condensates is measured - but this graph's condensate standard still holds that clustered binding is a live rival model. Nothing here is in a KMT2A-rearranged cell. Abstract only.

    Evidence that argues against this (1)
    • The nucleolar tether is not permanent and can be broken from an unexpected direction: nucleus-translocated mitochondrial cytochrome c binds nucleolar NPM, triggers an extended-to-compact conformational change and liberates NPM-sequestered ARF by altering nucleolar liquid-liquid phase separation, so ARF's nucleolar residence is a regulated state and not an address. González-Arzola K et al. 2022
  5. E_q65_loc2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    Nucleus-translocated mitochondrial cytochrome c binds nucleolar nucleophosmin and drives an extended-to-compact conformational change that releases NPM-sequestered ARF, changing nucleolar liquid-liquid phase separation - a mitochondria-to-nucleolus route for controlling where the ARF frame of CDKN2A is.

    PMID 36220893DOI
    What this evidence cannot show

    Non-leukaemic cells and a stress-induced, non-canonical cytochrome c localisation; the generality claimed by the authors goes beyond what one system shows. Cited for the mechanism of ARF release, not for the breadth. Abstract only.

    Evidence that argues against this (1)
    • The better-characterised ARF release trigger is DNA damage acting through NPM directly; adding a cytochrome c route does not displace it, and no experiment ranks the two in any one cell. Gibbs E et al. 2024
  6. E_q65_loc3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    The IVD-approved anti-p16INK4a immunohistochemistry clone E6H4 can produce a strong nuclear WITH NUCLEOLAR staining pattern, easily confused with the HPV-associated block-positivity pattern, and that pattern is caused by frameshift deletions or insertions within exon 2 of the INK4a/ARF locus which convert the reading frame of the 3-prime p14ARF sequence into p16INK4a, producing a p14ARF::p16INK4a chimeric protein; recognising the pattern is essential to avoid misdiagnosis of HPV association and inappropriate treatment.

    PMID 41339993DOI
    What this evidence cannot show

    Diagnostic pathology in HPV-associated cancer, not leukaemia, and CDKN2A exon-2 frameshifts are not the commonest lesion at this locus in KMT2A-rearranged disease, where biallelic DELETION predominates. Recorded here because it is a clean instance of rule 74 in this graph's own gene set: a nucleolar signal from an anti-p16 antibody reports neither p16 nor ARF but a chimera of the two frames. The PubMed record carries no volume or pages (Virchows Arch, 2025, online first). Abstract only.

    Evidence that argues against this (1)
    • A p16 IHC pattern cannot be read as a localisation of p16INK4a protein at all in this situation - the epitope is on a chimeric protein whose compartment is determined by the ARF-derived sequence, so this evidence item argues against, not for, using IHC as a p16 localisation assay. Gibbs E et al. 2024
  7. E_q69_ct1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In MICE, de-repressed p16(Ink4a) and p19(Arf) in Bmi1-deficient animals are tightly associated with a loss of self-renewing haematopoietic stem cells, and deletion of both Ink4a and Arf substantially restores the self-renewal capacity of Bmi1(-/-) HSCs, so Bmi1 regulates HSCs by acting as a critical failsafe against p16(Ink4a)- and p19(Arf)-dependent premature loss of HSCs; the bone marrow microenvironment of Bmi1(-/-) mice is separately and severely defective at supporting haematopoiesis and is NOT considerably restored by deleting both frames, leading to sustained postnatal HSC depletion in Bmi1(-/-)Ink4a-Arf(-/-) mice.

    Differential impact of Ink4a and Arf on hematopoietic stem cells and their bone marrow microenvironment in Bmi1-deficient mice. · Oguro H, Iwama A, Morita Y, Kamijo T, van Lohuizen M, Nakauchi H · J Exp Med 2006
    PMID 16954369DOI
    What this evidence cannot show

    Abstract read in full; full text not read. Murine, normal haematopoiesis under a polycomb lesion, not leukaemia and not human (brief rule 73). It measures the locus's effect when its repression is REMOVED, not its function in a wild-type HSC. Murine p19Arf corresponds to human p14ARF; the two frames are kept distinct in this entry and the mouse nomenclature is preserved in the claim.

    Evidence that argues against this (1)
    • The rescue is partial, and the authors say so: the Bmi1-deficient bone marrow microenvironment remains defective after Ink4a/Arf deletion and HSC depletion continues postnatally. The locus therefore explains the cell-autonomous arm of polycomb's HSC function and not the non-autonomous one, which bounds any claim that p16/p19 de-repression is what polycomb loss 'does' to this compartment. Oguro H et al. 2006
  8. E_q69_cs1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In MICE, taspase1-mediated cleavage of the general transcription factor TFIIA ensures proper coordination of rapid cell proliferation and morphogenesis during head formation by maintaining LIMITED transcription of the negative cell-cycle regulators p16Ink4a and p19Arf from the Cdkn2a locus; loss of TASP1 function causes catastrophic craniofacial malformations associated with inadequate cell proliferation, compound Cdkn2a deficiency - especially p16Ink4a deficiency - markedly reduces those anomalies, and ChIP shows non-cleaved TFIIA accumulating at the p16Ink4a and p19Arf promoters to drive their transcription.

    Taspase1-dependent TFIIA cleavage coordinates head morphogenesis by limiting Cdkn2a locus transcription. · Takeda S, Sasagawa S, Oyama T, Searleman AC, Westergard TD, Cheng EH, Hsieh JJ · J Clin Invest 2015
    PMID 25664857DOI
    What this evidence cannot show

    Abstract read in full; full text not read. Murine craniofacial development - not haematopoietic, not leukaemia, not human (brief rule 73). Its place in this entry is as a regulatory circuit that reaches the Cdkn2a locus from TASP1, a gene in the same tranche, through TFIIA rather than through MLL. It licenses no claim about p16 or p14ARF in a leukaemic cell.

    Evidence that argues against this (1)
    • The Taspase1 literature's other principal claim is that its cell-cycle role runs through MLL and MLL2 non-cleavage, with MLL(nc/nc;2nc/nc) MEFs also impaired in proliferation and p16(Ink4a) up-regulated. The TFIIA paper by contrast names TFIIA as the PRINCIPAL substrate for the head phenotype. Which substrate carries the Cdkn2a effect in which tissue is not settled, so this circuit should not be transferred to blood. Takeda S et al. 2006
  9. E_q97_epi1moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    In MURINE MLL/ENL-transduced cells, chromatin immunoprecipitation places EZH2 highly enriched around the transcription start site of p16 together with H3K27 methylation marks, and the same enrichment is absent in E2A/HLF-transduced cells; EZH2 inhibition by DZNep or shRNA reduces leukaemia-initiating-cell frequency and p16 knockdown cancels the survival advantage.

    PMID 24612037
    What this evidence cannot show

    Abstract fetched and read in full 2026-09-21; full text not read. MURINE, retroviral MLL/ENL - a myeloid model and not the index fusion or lineage (rule 73). DZNep is a promiscuous SAH-hydrolase inhibitor, not a selective EZH2 inhibitor, and the shRNA arm is what carries the specificity. The ChIP is reported for H3K27 methylation marks without a stated methylation state in the abstract, so H3K27me3 specifically is inferred rather than quoted.

    Evidence that argues against this (1)
    • The paper's own result undercuts the graph's tidiest version of this axis. Hoxa9 expression and p16 expression were NOT inversely related between MLL/ENL and Hoxa9/Meis1 cells, so recruitment of EZH2 to p16 cannot be explained by high HOXA9 alone and some other, unidentified mechanism must also repress p16. This graph's gene-bmi1 entry argues HOXA9 covers for BMI1 at the CDKN2A locus; the source of that idea reports the correlation failing. Cancer Sci 2014
  10. E_q97_epi2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In MURINE MLL-ENL-induced AML, Hhex binds the Cdkn2a locus and directly interacts with PRC2 to enable H3K27me3-mediated epigenetic repression; loss of Hhex de-represses p16INK4a and p19ARF, which are then required for the resulting growth arrest and myeloid differentiation.

    PMID 26728554
    What this evidence cannot show

    Abstract fetched and read in full 2026-09-21; full text not read. MURINE MLL-ENL MYELOID leukaemia (rule 73) - not KMT2A::AFF1 and not B-lineage. Hhex, not the fusion, is the DNA-binding factor placed at the locus, so this establishes that Cdkn2a carries PRC2-dependent H3K27me3 in a KMT2A-fusion AML and NOT that the fusion itself puts it there.

    Evidence that argues against this (1)
    • Hhex is dispensable for normal myelopoiesis in the same paper, which is what makes it a candidate target - but it also means the H3K27me3 at Cdkn2a in this model is downstream of a leukaemia-specific factor, so the mark's presence cannot be generalised to wild-type haematopoiesis or to a different fusion without being re-measured. No human sample is reported. Genes Dev 2016
  11. E_q97_epi3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    The dual EZH2/EZH1 inhibitor UNC1999 derepresses polycomb targets including Cdkn2a in MLL-rearranged leukaemia cells, preferentially at distal regulatory elements, with derepression correlating with a decrease in H3K27me3 and a concurrent gain in H3K27 acetylation; the inactive analogue UNC2400 does not, and oral UNC1999 prolongs survival in a murine MLL-AF9 model.

    PMID 25395428
    What this evidence cannot show

    Abstract fetched and read in full 2026-09-21; full text not read. Already cited three times in this graph (gene-ezh2 E4, mark-h3k27me3 E0, risk-ezh2-inhibition-b-lineage-translocations E2), so the paper is known here - what is new is the Cdkn2a-specific reading. MURINE MLL-AF9 in vivo and MLL-rearranged lines; this is pharmacological derepression, which shows the mark is load-bearing for Cdkn2a expression but does not show the fusion deposits it.

    Evidence that argues against this (1)
    • A mark that can be removed with a drug in a myeloid model is not a mark measured in the index disease, and this graph already holds an entry - risk-ezh2-inhibition-b-lineage-translocations - arguing that EZH2 inhibition in B-lineage cells carries its own hazard. Derepressing CDKN2A pharmacologically in an infant would be a cell-cycle brake applied through a globally acting chromatin drug. Blood 2015
  12. E_q97_epi4moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    In 91 newly diagnosed B-lineage ALL patients (61 children, 30 adults), promoter methylation of p16 was present in 34.4% of children and 26.7% of adults, p14 in 14.8% and 10.0%, p15 in 34.4% and 40.0%; NO significant association was observed between methylation and deletion, or between either and any known cytogenetic prognostic factor, MLL rearrangement included.

    PMID 19837270
    What this evidence cannot show

    Abstract fetched and read in full 2026-09-21; full text not read. HUMAN and B-lineage, which is the right setting - but 61 children is small for a four-way cytogenetic split, MLL-rearranged cases are a minority of any unselected childhood B-ALL cohort, and no infant subgroup is reported. A null association at this n is weak evidence of no association, NOT evidence of none; the honest reading is that the study was not powered to see a KMT2A-specific effect. Methylation was assayed at a promoter, which is a different measurement from a genome-wide CpG-island call.

    Evidence that argues against this (1)
    • This is the entry own counter-evidence to the brief that commissioned it. The dispatch premise was that CDKN2A is silenced by DNA methylation and H3K27me3 in leukaemia; the only human B-lineage study that measures both deletion and methylation at this locus finds methylation unassociated with MLL rearrangement. The DNA-methylation half of the premise is true of childhood B-ALL and unproven for the fusion. Cancer Genet Cytogenet 2009
Connected to