Oncofusion

start / graph / gene-tp53

Genestrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.Q46-cooperating-lesions

TP53 - named in 84 entries of this graph, absent from the index genome study of the disease, and four different things depending on when you look

also called TP53 · P53 · p53 · cellular tumor antigen p53 · tumour suppressor p53 · P04637

TP53 is the most famous tumour-suppressor gene in cancer, and in newly diagnosed infant leukaemia driven by a KMT2A fusion it is essentially not involved: the large genome study of this disease does not report a single case. It appears later - when the disease comes back after chemotherapy, when leukaemia is caused by earlier cancer treatment, and possibly when a menin inhibitor stops working.

The same thing, for a scientist

TP53 (P04637, 393 aa) is named across this graph more often than any other co-mutation candidate and had no entry. The reason it needs one is that **the same gene means four different things in this disease depending on the clinical moment**, and pooling them - which is what a bare 'TP53 is mutated in KMT2A-rearranged leukaemia' does - produces a claim that is simultaneously true and useless. **1. At diagnosis in infants: effectively absent.** The index whole-genome study of infant MLL-rearranged ALL - 22 WGS cases, 47 infant MLL-R cases in total, plus a 232-gene panel - **does not mention TP53 anywhere in its text** [E1]. Independently, across nearly 1,000 COG paediatric AML participants, TP53 and DNMT3A, both common in adults, were 'conspicuously absent from virtually all pediatric cases' [E2]. **2. At relapse: enriched, and treatment-shaped.** Alterations in drug-response genes, most commonly TP53 and IKZF1, were present in 64% of relapsed KMT2A-r cases and in **79% of early relapse ALL (9-36 months)** against **9% of very early relapse (< 9 months)**, with a chemotherapy-exposure mutational signature confined to the early-relapse group [E3]. These are consequences of therapy, not cooperating lesions of the fusion. **3. In therapy-related disease: about 18%.** In 56 paediatric therapy-related haematologic neoplasms, roughly half carrying KMT2A rearrangement, about 18% harboured 17p deletion and/or TP53 mutation [E4]. **4. In menin-inhibitor resistance: proposed, contested, and one of the two sources is a preprint.** TP53 mutation was associated with de novo revumenib resistance in an isogenic AML model, with impaired TP53 target induction and deregulated BCL-2/MCL-1; the MCL-1 inhibitor MIK665, but **not** venetoclax, preferentially sensitised TP53-mutant cells [E5]. Against that, in patient-derived xenografts from multiply pretreated KMT2A::AFF1 ALL, TP53 and RAS-pathway comutations emerged with acquired resistance but **neither was sufficient to induce menin-inhibitor resistance in vitro** [E6]. And there is a **published contradiction that must not be quietly dropped**: a 1996 series reported TP53 mutations in 9 of 29 MLL-rearranged acute leukaemias, including 3 of 4 infants with myeloid/monoblastic disease [E7]. That is irreconcilable with the modern sequencing picture and this entry records it as a dispute with reasons, not as a refutation of either side.

What it looks like
TP53393 aa
DNA-binding domain (core)DNA-binding domain (core) · residues 102–2920100200300
loading AF-P04637-F1…
AlphaFold AF-P04637-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
The bounded absence, stated properly

Claim. The strings TP53 and p53 occur zero times in the full text of the index whole-genome study of infant MLL-rearranged ALL, read as 71,491 characters of tag-stripped text from efetch db=pmc on PMC4553269 on 2026-09-20 (Europe PMC fullTextXML returned HTTP 500 for this article) 1. That text includes the main body, the methods, the figure legends and the reference list.

What it bounds. The study's discovery cohort was 22 infants by WGS, its validation cohort 43 further infant ALL cases by a 232-gene panel plus SNP array, and 47 infant MLL-R cases in total. A zero across a WGS cohort of that size means TP53 mutation is not a recurrent feature of de novo infant KMT2A-rearranged ALL.

What it does not bound. It is one study. TP53 could have been on the 232-gene panel and found wild-type in every case without being named, so the absence is an absence of reporting in a study whose panel content is in the supplementary notes and was not read for this entry. And zero in 47 cases is compatible with a true frequency up to about 6% at the upper 95% bound.

Subtracted against this graph's own bibliography before being written, as the project rule requires: the disproving-shaped paper exists and is cited here as 7. Grep of the repository found no existing entry citing a TP53 mutation frequency in de novo infant KMT2A-rearranged ALL.

The 1996 contradiction, and why it is recorded rather than dismissed

Lanza and colleagues tested 29 MLL-rearranged acute leukaemias - 13 lymphoid, 16 myeloid/monoblastic - and reported TP53 mutations in 6/16 myeloid/monoblastic and 3/13 lymphoid; within the myeloid/monoblastic group, 3 of 4 infants were mutated. Their explicit conclusion was that at least two genetic lesions accumulate in the few weeks between conception or birth and infant leukaemia 7.

That is a 31% overall rate against a modern figure indistinguishable from zero. Four reasons to treat the old number cautiously, none of which is 'it is old':

  1. n = 4 infants in the subgroup carrying the striking claim. Three of four is one patient away from two of four.
  2. Mutation calling in 1996 was typically SSCP screening of exons 5-8 followed by Sanger sequencing, without a matched germline sample in most series. Without germline comparison, rare polymorphisms and PCR artefacts are called as mutations.
  3. MLL rearrangement in 1996 was defined by Southern blot, which detects breakage in the breakpoint cluster region without identifying a partner. Some cases so classified would not be called KMT2A-rearranged today.
  4. Referral bias. A 1996 series assembled to ask about TP53 in an unusual leukaemia is not a consecutive cohort.

But the modern side has a weakness too, and it belongs in the same paragraph: Lanza's TP53-mutant infants were myeloid/monoblastic, and the modern genome study is of infant ALL. The two are not measuring the same population, and there is no equivalent whole-genome study of infant KMT2A-rearranged AML. The honest status is unresolved for infant KMT2A-rearranged AML and effectively settled for infant KMT2A-rearranged ALL.

Where TP53 does carry weight in this disease
settingpopulationfinding
adult KMT2A-r BCP-ALL 8141 of 1,091 GRAALL patientsoverall mutational burden low, 'reminiscent of infant BCP-ALL'; TP53 and/or IKZF1 alteration defines a subset with 5-year CIR 69.3% vs 36.2% (P = .001) and OS 28.1% vs 60.7% (P = .006)
adult KMT2A-r AML 9185 HARMONY patients with NGSin de novo AML under 60, KRAS or TP53 mutation predicts lower CR (50% vs 86%) and shorter OS (median 7 vs 30 months)
relapsed paediatric KMT2A-r 336 relapsed ALL/AMLTP53/IKZF1 drug-response alterations in 79% of early relapse ALL, 9% of very early relapse
paediatric therapy-related neoplasms 456 t-HN, about half KMT2A-rabout 18% with 17p deletion and/or TP53 mutation
de novo infant KMT2A-r ALL 147 casesnot reported

The pattern across the table is consistent: TP53 in KMT2A-rearranged leukaemia is a marker of accumulated genomic history - age, prior therapy, prior relapse - rather than a partner of the fusion. The graph's 84 mentions of TP53 should be read against that.

The menin-inhibitor resistance question, with its sourcing stated

For. An isogenic AML model plus a cell-line panel found TP53 mutations associated with de novo revumenib resistance, impaired induction of TP53 transcriptional targets, and deregulated BCL-2 and MCL-1; MIK665 (MCL-1 inhibitor) but not venetoclax preferentially sensitised TP53-mutant cells to revumenib 5. This source is a bioRxiv preprint and is graded accordingly. If it holds, it is directly actionable: it says the obvious combination partner for a TP53-mutant KMT2A-rearranged leukaemia is an MCL-1 inhibitor and not the BCL-2 inhibitor the field reaches for.

Against. In PDXs from two multiply relapsed KMT2A::AFF1 patients who had never received a menin inhibitor, resistance was present upfront; targeted sequencing found emergent RAS-pathway and TP53 comutations, and neither was sufficient to induce menin-inhibitor resistance in vitro. The authors instead implicate KMT2C/D downregulation in one patient 6.

The two are compatible if TP53 loss is a permissive background rather than a resistance mechanism. They are not compatible with treating TP53 mutation as a menin-inhibitor resistance biomarker today, and this entry does not.

Facts and figures
uniprotP04637
gene symbolTP53
length AA393
alphafold IDAF-P04637-F1
cytoband17p13.1
resolution notegene_exact:TP53 AND organism_id:9606 AND reviewed:true with fields=accession,gene_primary,gene_synonym,protein_name,length returned one row: P04637, primary gene name TP53, synonym P53, 393 aa. Fetched from the UniProt REST API on 2026-09-20. DNA-binding region coordinates from the same record's feature table the same day.
domains noteUniProt P04637 annotates the DNA-binding region as a DNA binding feature rather than a Domain feature; the transactivation and tetramerisation regions are Region features and are not recorded here rather than guessed.
druggabilityNot directly druggable as a loss-of-function lesion. The actionable proposal in this disease is indirect and preprint-stage: TP53-mutant KMT2A-rearranged cells were reported MCL-1 dependent and sensitised to revumenib by MIK665 rather than by venetoclax.
subnuclear localizationPML NUCLEAR BODIES - the best-evidenced subnuclear address of any protein in this tranche, and one with a direct fusion-oncoprotein precedent already in this graph. UniProt P04637 annotates 'Nucleus, PML body' at ECO:0000269 on three references, and the founding observation is Guo et al. 2000: PML physically interacts with p53 in vitro and in vivo, COLOCALISES with p53 in the PML nuclear body, and PML's co-activation of p53 DEPENDS ON ITS ABILITY TO LOCALISE IN THE PML-NB - so the body is load-bearing for the function, not decorative [E_q77_loc2]. A second study adds that PML recruits p53 into PML bodies together with CHEK2 and protects it from MDM2 [E_q77_loc3]. TWO CAUTIONS TRAVEL WITH THIS. First, the third citation UniProt gives for the PML-body location is an IN VITRO LIQUID-LIQUID PHASE SEPARATION study [E_q77_loc4], which by this graph's own condensate standards cannot support a claim about a nuclear body in a cell. Second, an UNCODED isoform-scoped comment says isoform 8 'forms foci in the nucleus that are different from nucleoli' - a third, unnamed subnuclear compartment that every ECO-filtered read of this record discards [E_q77_loc1].
shown for fusionPARTIAL, and the fusion is PML::RARA rather than any KMT2A fusion. p53's PML-body residence and its functional dependence on that body are established in the acute-promyelocytic-leukaemia literature, which is the one fusion oncoprotein disease where a subnuclear rearrangement is the accepted mechanism. NOTHING has been measured in KMT2A::AFF1, KMT2A::MLLT3, MLL-AF9 mouse or any other KMT2A system - and that gap is conspicuous because this graph names TP53 in 84 entries.
domains
[
  {
    "name": "DNA-binding domain (core)",
    "start": 102,
    "end": 292,
    "pfam": null
  }
]
four settings
{
  "de_novo_infant_kmt2a_r_all": "not reported in 47 cases in the index whole-genome study; zero occurrences of TP53 or p53 in its full text",
  "relapse": "TP53 with IKZF1 among drug-response gene alterations in 64% of relapsed cases; 79% of early relapse ALL vs 9% of very early relapse; chemotherapy-exposure signature confined to early relapse",
  "therapy_related": "about 18% of paediatric therapy-related haematologic neoplasms carry 17p deletion and/or TP53 mutation, in a cohort about half of which is KMT2A-rearranged",
  "menin_inhibitor_resistance": "associated with de novo revumenib resistance in a preprint isogenic AML model with MCL-1 dependence; NOT sufficient to induce resistance in vitro in pretreated KMT2A::AFF1 PDXs"
}
open dispute
{
  "claim": "TP53 mutation in 9 of 29 MLL-rearranged acute leukaemias including 3 of 4 infants with myeloid/monoblastic disease",
  "source_pmid": "8824725",
  "status": "unresolved for infant KMT2A-rearranged AML; effectively settled against for infant KMT2A-rearranged ALL",
  "why_unresolved": "the 1996 cases were myeloid/monoblastic and there is no modern whole-genome study of infant KMT2A-rearranged AML to compare against"
}
depmap kmt2a selectivity
{
  "chronos_mean_kmt2a_rearranged": 0.4984,
  "chronos_mean_other_leukaemia": 0.2573,
  "difference": 0.2412,
  "pan_essentiality_all_lines": 0.4242,
  "selectivity_rank": 17652,
  "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."
}
localization
{
  "subcellular": "SIX COMPARTMENTS, ALL EXPERIMENTALLY CODED, AND A SEVENTH LAYER THE CODES HIDE. UniProt P04637 fetched 2026-09-21 annotates, all at ECO:0000269: Cytoplasm (nine references), Nucleus (nine), 'Nucleus, PML body' (three), Endoplasmic reticulum (one), Mitochondrion matrix (three) and 'Cytoplasm, cytoskeleton, microtubule organizing center, centrosome' (one). This is the most heavily evidenced localisation record in this tranche by an order of magnitude. UNDERNEATH IT sit SEVEN ISOFORM-SCOPED SUBCELLULAR COMMENTS - isoforms 1, 2, 3, 4, 7, 8 and 9 - and EVERY ONE OF THEM IS UNCODED [E_q77_loc1]. Isoform 9 is annotated CYTOPLASM ONLY. Isoform 1 is 'predominantly nuclear but localizes to the cytoplasm when expressed with isoform 4'. Isoform 4 'translocates to the cytoplasm following cell stress'. This is brief rule 91's KMT2A fragment case repeating on the most-studied protein in cancer biology: filtering P04637 on ECO:0000269 keeps the six generic compartments and DISCARDS every isoform-resolved statement, which is where the interesting content is.",
  "subnuclear": "PML NUCLEAR BODIES - the best-evidenced subnuclear address of any protein in this tranche, and one with a direct fusion-oncoprotein precedent already in this graph. UniProt P04637 annotates 'Nucleus, PML body' at ECO:0000269 on three references, and the founding observation is Guo et al. 2000: PML physically interacts with p53 in vitro and in vivo, COLOCALISES with p53 in the PML nuclear body, and PML's co-activation of p53 DEPENDS ON ITS ABILITY TO LOCALISE IN THE PML-NB - so the body is load-bearing for the function, not decorative [E_q77_loc2]. A second study adds that PML recruits p53 into PML bodies together with CHEK2 and protects it from MDM2 [E_q77_loc3]. TWO CAUTIONS TRAVEL WITH THIS. First, the third citation UniProt gives for the PML-body location is an IN VITRO LIQUID-LIQUID PHASE SEPARATION study [E_q77_loc4], which by this graph's own condensate standards cannot support a claim about a nuclear body in a cell. Second, an UNCODED isoform-scoped comment says isoform 8 'forms foci in the nucleus that are different from nucleoli' - a third, unnamed subnuclear compartment that every ECO-filtered read of this record discards [E_q77_loc1].",
  "how_shown": "co-immunoprecipitation and immunofluorescence colocalisation with PML in PML nuclear bodies, with PML-NB-localisation-defective mutants as the control [E_q77_loc2]; recruitment into PML bodies with CHEK2 and MDM2 protection assays [E_q77_loc3]; subcellular fractionation and imaging of mitochondrial translocation under oxidative stress and mitomycin C [E_q77_loc5]; in vitro droplet assays for the phase-separation citation [E_q77_loc4]",
  "systems": "human and murine cultured cells across two decades of p53 cell biology; the mitochondrial work includes mouse stroke and ischaemia models. NO KMT2A-rearranged system, no infant leukaemia material, and no measurement of p53 localisation in a blast of any genotype.",
  "shown_for_fusion_localisation": "NO for KMT2A - nobody has measured p53 localisation in a KMT2A-rearranged cell. But NOT a bare NO overall, and the exception is the point: the PML-body address of p53 is the subnuclear compartment that PML::RARA destroys, and this graph already carries localization-pml-nuclear-bodies-pmlrara for that reason. So p53's subnuclear biology has been read in ONE fusion oncoprotein disease - acute promyelocytic leukaemia - and in none of the KMT2A ones.",
  "node": "localization-pml-nuclear-bodies-pmlrara",
  "why_the_pml_body_matters_to_this_project": "PML::RARA disperses PML nuclear bodies into a microspeckled pattern, and this graph holds that as its cleanest worked example of a fusion oncoprotein whose mechanism IS a subnuclear reorganisation. p53's apoptotic co-activation by PML requires the body. That gives the project a ready-made positive control for the claim it is trying to make about KMT2A: a fusion changes a compartment, the compartment's resident tumour suppressor loses a function, and the phenotype follows. The uncomfortable comparison is that after decades nobody has done the equivalent measurement for a KMT2A fusion - see localization-menin-has-no-subnuclear-compartment-and-nobody-has-looked."
}
cell type
{
  "evidence_class_key": "concept-expressed-required-maintains-three-evidence-classes",
  "identifier_note": "TP53 is P04637, 393 aa, primary gene name TP53, synonym P53 (UniProt, 2026-09-21) - a single clean reviewed row. data/symbol-collisions.json was grepped for TP53 and P53 before the first query and has NO row for either; none is proposed, because gene_exact:TP53 resolves cleanly. The hazard for this gene is not the symbol, it is the ISOFORM: P04637 carries at least nine annotated isoforms with DIFFERENT compartments, and 'p53' in a paper almost never says which one.",
  "expressed_in": {
    "class": 1,
    "summary": "ubiquitous. The interesting cell-type statement about TP53 in this disease is not where it is expressed but where it is MUTATED, and this entry's four-settings block already answers that: effectively absent at diagnosis in infants, enriched at relapse and in therapy-related disease."
  },
  "required_in": {
    "class": 2,
    "claim": "IN MICE, p53 IS REQUIRED TO HOLD HAEMATOPOIETIC STEM CELLS QUIESCENT - a steady-state role distinct from the DNA-damage role it is famous for. Transcriptional profiling of HSCs from wild-type and p53-null mice identifies Gfi-1 and Necdin as p53 target genes in this context, and lentiviral over-expression or knockdown of either changes HSC quiescence [E_q77_ct1]. The authors' own framing is that p53's function during steady-state haematopoiesis had not previously been established.",
    "pmids": [
      "19128791"
    ],
    "system_named_in_the_claim": "murine steady-state haematopoiesis in wild-type and Trp53-null animals - no leukaemia, no fusion",
    "nodes": [
      "cell-type-hsc-bone-marrow"
    ],
    "the_direction_is_the_useful_part": "p53 PROMOTES quiescence. So losing TP53 - the event this entry tracks at relapse and in therapy-related disease - is predicted to push stem and progenitor cells OUT of quiescence and into cycle. That is the opposite direction from the quiescent persister state this graph holds as the relapse reservoir (cell-state-diagnosis-versus-relapse-quiescent-subclone), and nobody has reconciled the two in KMT2A-rearranged disease."
  },
  "context": "murine normal haematopoiesis; no human primary material and no fusion"
}
cell state
{
  "evidence_class_key": "concept-expressed-required-maintains-three-evidence-classes",
  "maintains_state": {
    "class": 3,
    "states": [
      "cell-state-quiescence",
      "cell-state-apoptotic-priming-bcl2"
    ],
    "claim": "Two states, in opposite registers, and the second is where this entry's therapeutic content already sits. (1) QUIESCENCE: in mice, p53 maintains HSC quiescence through Gfi-1 and Necdin, and this is a steady-state function rather than a damage response [E_q77_ct1]. (2) APOPTOTIC PRIMING: this entry's E5 already records that TP53-mutant AML cells show deregulated BCL-2 and MCL-1 and are preferentially sensitised to revumenib by the MCL-1 inhibitor MIK665 but NOT by venetoclax - which is a statement that TP53 status sets which anti-apoptotic protein the cell leans on. The subcellular half of that story is independently documented: p53 translocates to the MITOCHONDRIAL MATRIX under oxidative stress and opens the permeability transition pore [E_q77_loc5].",
    "pmids": [
      "19128791",
      "22726440"
    ],
    "system_named_in_the_claim": "murine steady-state haematopoiesis for quiescence; murine and human cells plus mouse stroke models for the mitochondrial arm - neither involves a KMT2A fusion"
  },
  "why_the_mitochondrial_pool_is_worth_recording_here": "This entry's only actionable proposal is an MCL-1 inhibitor combination, and MCL-1 is a mitochondrial outer-membrane protein. p53 has a documented, experimentally coded MITOCHONDRIAL MATRIX pool that acts non-transcriptionally. A TP53-mutant cell therefore loses two separable things - the transcriptional apoptotic programme and a direct mitochondrial activity - and the preprint that proposes the MIK665 combination reads out only the first. Which of the two the MCL-1 dependence follows from is untested and is a cheap experiment in an isogenic pair.",
  "confidence": "moderate",
  "caveats": "The quiescence result is murine, steady-state and germline-null, and its strongest arm is in HSCs lacking MEF/ELF4 rather than in wild-type animals. The mitochondrial result is largely from stress and ischaemia models, not leukaemia. Neither involves a KMT2A fusion or any human leukaemic material. And the apoptotic-priming arm of the claim rests on this entry's E5, which is a BIORXIV PREPRINT and is graded as such - it is included here to connect the subcellular record to the therapeutic proposal, not to raise the preprint's confidence.",
  "counter_evidence": [
    {
      "note": "The quiescence direction argues against the intuitive reading of TP53 loss in this disease. If p53 MAINTAINS HSC quiescence, TP53 loss should drive cells into cycle and make them MORE chemosensitive, not less - yet this entry's clinical table shows TP53 alteration predicting worse outcome in every setting measured (adult KMT2A-r BCP-ALL 5-year CIR 69.3% vs 36.2%; adult KMT2A-r AML CR 50% vs 86%). The two are reconcilable only if the clinical effect runs through apoptotic resistance rather than through cell cycle, which is exactly what the MCL-1 result would predict and exactly what has not been shown in this disease. This graph should not use the murine quiescence result to reason about TP53-mutant relapse without saying so.",
      "citation": {
        "title": "p53 regulates hematopoietic stem cell quiescence.",
        "authors": "Liu Y, Elf SE, Miyata Y, Sashida G, Liu Y, Huang G, Di Giandomenico S, Lee JM, Deblasio A, Menendez S, Antipin J, Reva B, Koff A, Nimer SD",
        "year": 2009,
        "venue": "Cell Stem Cell",
        "pmid": "19128791",
        "doi": "10.1016/j.stem.2008.11.006",
        "type": "paper"
      }
    }
  ]
}
How this could bite us back
  • TP53 mutation is rare at diagnosis and common after treatment in this disease, so a therapy that improves early control while adding genotoxic exposure could increase the proportion of relapses that are TP53-mutant - trading an early failure for a harder late one.
  • If MCL-1 inhibition is what sensitises TP53-mutant cells to a menin inhibitor, that is a combination of two agents with overlapping haematological toxicity in infants, and MCL-1 inhibitors carry a cardiac-toxicity history that has stopped programmes in adults.
  • Venetoclax is the BH3 agent this field actually reaches for, and the preprint reports it as the one that does NOT sensitise TP53-mutant cells. A combination chosen by availability rather than by mechanism could be the wrong one.
  • A TP53-mutant clone at relapse is a clone selected by prior therapy. Adding a targeted agent to chemotherapy in this disease should be accompanied by a plan to measure whether the surviving clone is more or less TP53-intact, and no trial in this disease reports that.
  • This entry's central negative - TP53 is not part of de novo infant KMT2A-rearranged ALL - rests on one study's full text. If TP53 turns out to have been on the validation panel and simply unreported, the negative weakens from 'not present' to 'not reported', and several sentences in this graph would need softening.
Still open
  1. Is TP53 mutated in infant KMT2A-rearranged AML? The only cohort that says yes is from 1996 and used Southern blot and SSCP; no whole-genome study of that population exists, and this is the single clearest unresolved question this entry found.
  2. Was TP53 on the 232-gene validation panel of the index study? If it was, the absence becomes a measured zero in 43 further cases rather than an absence of reporting - one supplementary table settles it.
  3. Does the preprint's MCL-1 dependence hold in KMT2A-rearranged ALL, and in primary infant material? The claim, if true, redirects combination strategy from venetoclax to an MCL-1 inhibitor in a subset defined by a gene that is rare at diagnosis and common at relapse.
  4. Do the TP53 alterations found at early relapse pre-exist at diagnosis below the detection floor, as the RAS lesions do? Duplex sequencing has been applied to the RAS question in this disease and not to the TP53 one.
Every source behind this entry (15)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    The strings TP53 and p53 do not occur anywhere in the full text of the index whole-genome sequencing study of infant MLL-rearranged ALL, which covered 22 infants by WGS and 47 infant MLL-rearranged cases in total with a 232-gene validation panel and SNP arrays.

    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

    A bounded absence from one study. The full text was retrieved via efetch db=pmc for PMC4553269 on 2026-09-20 after Europe PMC fullTextXML returned HTTP 500, tags stripped before entity unescaping, giving 71,491 characters; case-insensitive counts for 'TP53' and 'p53' were both zero. The supplementary notes listing the 232 panel genes were NOT retrieved, so this entry cannot say whether TP53 was on the panel and found wild-type or was simply not assayed. Zero events in 47 cases is compatible with a true frequency up to roughly 6% at the upper 95% confidence bound.

    Evidence that argues against this (1)
    • A 1996 series reported TP53 mutations in 9 of 29 MLL-rearranged acute leukaemias including 3 of 4 infants with myeloid/monoblastic disease, which is irreconcilable with a zero and is treated in this entry as an unresolved dispute rather than a superseded result. Lanza C et al. 1996
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Across nearly 1,000 Children's Oncology Group paediatric AML participants profiled by whole-genome, targeted DNA, mRNA and microRNA sequencing and CpG methylation, mutations in DNMT3A and TP53, which are common in adults, were conspicuously absent from virtually all paediatric cases.

    PMID 29227476DOI
    What this evidence cannot show

    Paediatric AML as a whole, not the KMT2A-rearranged subset specifically, and 'virtually all' is the authors' own qualifier rather than a stated count. The cohort is trial-enrolled and therefore excludes children not entered on COG protocols. Abstract read; full text not retrieved for this entry.

    Evidence that argues against this (1)
    • Paediatric therapy-related haematologic neoplasms, which are enriched for KMT2A rearrangement, show about 18% with 17p deletion and/or TP53 mutation - so 'absent in paediatric disease' holds for de novo and not for treatment-related disease in the same age group. Kovach AE et al. 2024
  3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In 36 infants and children with relapsed KMT2A-rearranged ALL or AML, somatic alterations in drug-response genes, most commonly TP53 and IKZF1, were present in 64% and were highly enriched in early relapse ALL (79%, 9-36 months after diagnosis) but rare in very early relapse ALL (9%, under 9 months), with a chemotherapy-exposure mutational signature detected in early relapse ALL and not in very early ALL or AML relapse.

    The genomic landscape of relapsed infant and childhood KMT2A-rearranged acute leukemia. · Ahlgren L, Pilheden M, Sturesson H, et al. · Nature Communications 2025
    PMID 41062506DOI
    What this evidence cannot show

    A relapse-only cohort of 36 patients split into timing subgroups, so each subgroup is small. The 64% figure covers a curated 'drug-response genes' category, not TP53 alone; this entry therefore does not quote a TP53-specific relapse frequency. Mutational-signature attribution is unstable on small mutation counts. Abstract read; full text not retrieved for this entry.

    Evidence that argues against this (1)
    • The very-early relapses, which lack both the chemotherapy signature and the drug-response gene alterations, are the relapses with the worst prognosis - so the subgroup with the clearest TP53 involvement is not the subgroup with the worst outcome, and TP53 cannot be the general explanation of relapse in this disease. Ahlgren L et al. 2025
  4. E4moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    In 56 retrospectively identified paediatric therapy-related haematologic neoplasms, approximately half harboured a KMT2A rearrangement, and approximately 18% harboured 17p deletions and/or TP53 mutations.

    PMID 39116419DOI
    What this evidence cannot show

    The abstract reports the 18% figure for the cohort as a whole, not for the KMT2A-rearranged half specifically, and the paper explicitly states that among therapy-related neoplasms WITHOUT KMT2A rearrangement the genetic drivers were heterogeneous - which is where the 17p/TP53 events may be concentrated. Retrospective, single-cohort, and the assays varied across cases (microarray, NGS and/or RNA sequencing). Abstract read.

    Evidence that argues against this (1)
    • The same abstract states that many paediatric therapy-related neoplasms are driven by genetic events independent of clonal haematopoiesis, which cuts against reading TP53 in this setting as the adult clonal-haematopoiesis story transplanted into children. Kovach AE et al. 2024
  5. E5weak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.preprint

    In an AML cell-line panel and an isogenic model, TP53 mutations were associated with de novo resistance to the menin inhibitor revumenib, with impaired induction of TP53 transcriptional targets and deregulated BCL-2 and MCL-1 expression, and the MCL-1 inhibitor MIK665 - but not venetoclax - preferentially sensitised TP53-mutant AML cells to revumenib.

    PMID 40196700DOI
    What this evidence cannot show

    A PREPRINT, not peer reviewed, and graded weak on that basis alone. AML cell lines and an isogenic model, not KMT2A-rearranged infant ALL and not primary patient material; the NPM1-mutant and KMT2A-rearranged contributions to the panel are not separable from the abstract. A cell-line association between TP53 status and drug response is confounded by everything else that differs between lines, which is why the isogenic arm matters and why its size and design should be read before acting. Abstract read.

    Evidence that argues against this (1)
    • In patient-derived xenografts from multiply pretreated KMT2A::AFF1 ALL, TP53 comutations emerged alongside acquired menin-inhibitor resistance but were explicitly NOT sufficient to induce menin-inhibitor resistance in vitro. Mahdavi L et al. 2026
  6. E6moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    In patient-derived xenografts of KMT2A-rearranged leukaemia, highly pretreated KMT2A::AFF1 ALL samples were much less sensitive to menin inhibition than earlier samples from the same patients despite no prior menin-inhibitor exposure, transcriptomic analysis showed sustained on-target activity, and emergent RAS-pathway and TP53 comutations were individually insufficient to induce menin-inhibitor resistance in vitro.

    Upfront menin-inhibitor resistance in multiply pretreated leukemias. · Mahdavi L, Alikarami F, Goodrow H, et al. · Experimental Hematology 2026
    PMID 41027490DOI
    What this evidence cannot show

    Serial samples from two patients for the resistance arm, five PDX models for the sensitivity arm - a small study whose value is the design (paired earlier and later samples from the same patient) rather than the n. 'Not sufficient in vitro' is a negative result in an engineered system and does not exclude a contribution in vivo or in combination. Abstract read; full text not retrieved for this entry.

    Evidence that argues against this (1)
    • The authors' own alternative explanation - KMT2C/D downregulation, with KMT2C-edited cells from the same patient selected under menin inhibition - is offered for one patient only, so the negative finding about TP53 is not accompanied by an established positive mechanism. Mahdavi L et al. 2026
  7. E7weak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.paper

    A 1996 series of 29 MLL-rearranged acute leukaemias reported TP53 mutations in 6 of 16 myeloid/monoblastic cases and 3 of 13 lymphoid cases, with 3 of the 4 infants in the myeloid/monoblastic group mutated, and concluded that MLL rearrangement plus TP53 mutation accumulate within the few weeks corresponding to the development of infant leukaemia.

    Distribution of TP53 mutations among acute leukemias with MLL rearrangements. · Lanza C, Gaidano G, Cimino G, et al. · Genes, Chromosomes and Cancer 1996
    PMID 8824725DOI
    What this evidence cannot show

    Graded weak on design, not on age. The striking infant claim rests on 4 patients. Mutation detection in 1996 was typically SSCP screening of a subset of exons with Sanger confirmation and frequently without a matched germline sample, so polymorphisms and artefacts can be scored as mutations. MLL rearrangement was defined by Southern blot, which identifies breakpoint-cluster-region breakage without a partner gene, so some cases would not meet a modern KMT2A-rearranged definition. Cohort assembly is not stated to be consecutive. The abstract was read; the full text was not retrieved for this entry.

    Evidence that argues against this (1)
    • Modern whole-genome sequencing of 47 infant MLL-rearranged ALL cases does not report TP53 at all, and nearly 1,000 paediatric AML participants showed TP53 conspicuously absent - so the 1996 rate is not reproduced by any subsequent cohort. Andersson AK et al. 2015
  8. E8strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Among 141 adults with KMT2A-rearranged B-cell precursor ALL from 1,091 Philadelphia-negative BCP-ALL patients in three consecutive GRAALL trials, molecular profiling showed a low mutational burden reminiscent of infant BCP-ALL, and the presence of TP53 and/or IKZF1 alterations defined a subset with significantly poorer 5-year cumulative incidence of relapse (69.3% vs 36.2%, P = .001) and overall survival (28.1% vs 60.7%, P = .006).

    PMID 37595275DOI
    What this evidence cannot show

    Adults, not infants, and TP53 is combined with IKZF1 into a single composite exposure so no TP53-specific hazard can be extracted. Trial-enrolled patients, so not population-representative. The finding that adult KMT2A-r BCP-ALL also has a low mutational burden is the load-bearing observation for this graph and it is stated without a number in the abstract. Abstract read; full text not retrieved.

    Evidence that argues against this (1)
    • That the same low-burden phenotype appears in adults weakens the standing inference that the quiet genome is a consequence of the prenatal, infant developmental window; the quietness may be a property of the fusion rather than of infancy. Kim R et al. 2023
  9. E9moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    In 185 adult KMT2A-rearranged AML patients with panel NGS from the HARMONY database, KRAS and TP53 were the prognostically most relevant mutated genes in de novo AML under 60, with mutation of either predicting a lower complete remission rate (50% vs 86%, P < 0.001) and inferior overall survival (median 7 vs 30 months, P < 0.001).

    PMID 38965370DOI
    What this evidence cannot show

    Adults, AML, registry cohort assembled across trials, and KRAS and TP53 are pooled into one exposure for the complete-remission and survival comparison, so neither gene's individual effect can be read off it. Abstract read; full text not retrieved.

    Evidence that argues against this (1)
    • Paediatric AML shows TP53 conspicuously absent across nearly 1,000 cases, so this adult result cannot be carried into the paediatric or infant disease that this graph is about. Bolouri H et al. 2018
  10. E_q77_loc1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    UniProt P04637 (fetched 2026-09-21) carries SEVEN ISOFORM-SCOPED subcellular comments - for isoforms 1, 2, 3, 4, 7, 8 and 9 - and NOT ONE of them carries an evidence code, while all six of the unscoped compartment annotations carry ECO:0000269. Isoform 9 is annotated CYTOPLASM only; isoform 1 is 'predominantly nuclear but localizes to the cytoplasm when expressed with isoform 4'; isoform 4 'translocates to the cytoplasm following cell stress'; and isoform 8 'forms foci in the nucleus that are different from nucleoli'.

    What this evidence cannot show

    A measurement of the database, reported as such, read from the comment array with molecule-scoping and evidence codes attached on 2026-09-21 rather than from the flattened field, which loses both. This is brief rule 91's KMT2A N320/C180 case on a different protein and it is worse here, because there are seven scoped comments rather than two and one of them names a subnuclear structure - nuclear foci distinct from nucleoli - that appears nowhere else in the record. Uncoded does not mean wrong; it means a curator wrote it without attaching an experiment, so it must be recorded as uncoded and not discarded.

    Evidence that argues against this (1)
    • The generous reading - that the isoform comments are curator summaries of the same experiments behind the coded lines - cannot be excluded from the record alone, and this item does not claim otherwise. What it claims is narrower and checkable: an automated pipeline filtering P04637 on ECO:0000269 returns six generic compartments and zero isoform-resolved statements, so any downstream analysis of p53 localisation built that way is blind to the isoform layer by construction. That is a property of the filter, and the same filter is what this project's own localisation coverage work has used. Guo A et al. 2000
  11. E_q77_loc2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    PML acts as a transcriptional co-activator of p53: PML physically interacts with p53 in vitro and in vivo and COLOCALISES with p53 in the PML nuclear body, and the co-activatory role DEPENDS ON PML's ABILITY TO LOCALISE IN THE PML-NB. p53-dependent DNA-damage-induced apoptosis, transcriptional activation by p53, p53 DNA binding and induction of p53 targets including Bax and p21 after gamma-irradiation all require PML.

    The function of PML in p53-dependent apoptosis. · Guo A, Salomoni P, Luo J, Shih A, Zhong S, Gu W, Pandolfi PP · Nat Cell Biol 2000
    PMID 11025664DOI
    What this evidence cannot show

    Rule 91 check performed and PASSED for this reference: the paper UniProt cites for 'Nucleus, PML body' does contain the colocalisation and, better, contains the localisation-dependence control that turns a colocalisation into a functional claim. Limits: cultured cells, gamma-irradiation as the stimulus, and no haematopoietic or leukaemic material. The PML-body dependence is demonstrated for PML's function, so the strict reading is that p53 co-activation requires PML to be in the body, which is one inferential step from 'p53's own pool in the body is the active one'.

    Evidence that argues against this (1)
    • The same UniProt line cites an IN VITRO PHASE-SEPARATION paper as a third source for the PML-body location, and that is a category error the record does not flag. A droplet assay in a tube cannot establish residence in a nuclear body, and by this graph's own condensate standards it cannot even distinguish phase separation from binding to clustered sites. Anyone taking the three ECO:0000269 references for this compartment as three independent cellular observations would be counting one of them wrongly. Chen C et al. 2022
  12. E_q77_loc3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    PML recruits p53 into PML nuclear bodies together with CHEK2 and protects p53 from MDM2-mediated inhibition and degradation - so the body is a site of p53 stabilisation as well as of its co-activation.

    The promyelocytic leukemia protein protects p53 from Mdm2-mediated inhibition and degradation. · Louria-Hayon I, Grossman T, Sionov RV, Alsheich O, Pandolfi PP, Haupt Y · J Biol Chem 2003
    PMID 12810724DOI
    What this evidence cannot show

    Cultured cells; the CHEK2 recruitment element is the part UniProt's note cites and it is one component of a paper about MDM2 antagonism. No haematopoietic material. Recorded here because it makes the PML body a functional compartment for p53 in two independent registers - transcription and stability - rather than a place p53 is merely seen.

    Evidence that argues against this (1)
    • Both PML-body results are from the Pandolfi laboratory line of work (Pandolfi is an author on this paper and on the founding one), so the graph should not count them as two independent groups. The independent corroboration that would matter - a PML-body measurement of p53 in a leukaemic blast, in either APL or a KMT2A-rearranged cell - does not exist. Guo A et al. 2000
  13. E_q77_loc4weak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.paper

    One of the three ECO:0000269 references UniProt gives for p53's 'Nucleus, PML body' location is an IN VITRO liquid-liquid phase separation study: Chen et al. 2022 report that p53 undergoes phase separation driven by its unstructured basic region and that oncogenic mutations in the tetramerisation domain prevent it.

    PMID 36108750DOI
    What this evidence cannot show

    Rule 91 FIRES, and it fires on the compartment that matters most for this entry. The paper is reconstituted biophysics; it is not a study of PML nuclear bodies and it does not measure where p53 is in a cell. Confidence is WEAK deliberately. This graph's standing condensate position applies in full: 1,6-hexanediol sensitivity and classical FRAP cannot distinguish phase separation from binding to spatially clustered sites, and only model-free calibrated half-bleaching can, so an in vitro droplet result licenses nothing about a nuclear body. Recorded so that the PML-body annotation is not read as resting on three cellular observations when it rests on two.

    Evidence that argues against this (1)
    • The cellular result the annotation should rest on is Guo et al. 2000, which has the colocalisation AND the localisation-dependence control. Weighting that one and demoting the droplet paper leaves the PML-body call intact - this item is a correction to the provenance of a claim, not to the claim. Guo A et al. 2000
  14. E_q77_loc5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    p53 has an experimentally coded MITOCHONDRIAL MATRIX pool that acts non-transcriptionally: it translocates to mitochondria under oxidative stress and opens the mitochondrial permeability transition pore to trigger necrosis, and it separately translocates to mitochondria in response to mitomycin C.

    p53 opens the mitochondrial permeability transition pore to trigger necrosis. · Vaseva AV, Marchenko ND, Ji K, Tsirka SE, Holzmann S, Moll UM · Cell 2012
    PMID 22726440DOI
    What this evidence cannot show

    Largely oxidative-stress, ischaemia and stroke models in mouse plus human cells; not a leukaemia system and not a KMT2A-rearranged one. Included in a localisation block because this entry's only actionable therapeutic proposal is an MCL-1 inhibitor combination and MCL-1 acts at the mitochondrial outer membrane - so the compartment is mechanistically adjacent to the proposal rather than incidental.

    Evidence that argues against this (1)
    • Reading the mitochondrial pool into the menin-inhibitor-resistance proposal would be an unlicensed join. The preprint behind that proposal (this entry's E5) reads out TRANSCRIPTIONAL target induction and BCL-2/MCL-1 expression; it does not measure p53 localisation at all, and nothing shows that the MCL-1 dependence of TP53-mutant KMT2A-rearranged cells runs through the loss of a mitochondrial p53 pool rather than through the loss of transcriptional output. The two hypotheses are separable in an isogenic pair with a transactivation-dead versus mitochondrially-excluded p53, and that experiment has not been done in this disease. UniProt Consortium 2026
  15. E_q77_ct1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In MICE, p53 regulates HAEMATOPOIETIC STEM CELL QUIESCENCE during steady-state haematopoiesis - a role the authors state had not previously been established, distinct from p53's DNA-damage function. Transcriptional profiling of HSCs from wild-type and p53-null mice identifies Gfi-1 and Necdin as p53 target genes in this setting, and lentiviral over-expression or knockdown of either alters HSC quiescence; the effect is most pronounced in the enhanced quiescence of HSCs lacking MEF/ELF4.

    p53 regulates hematopoietic stem cell quiescence. · Liu Y, Elf SE, Miyata Y, Sashida G, Liu Y, Huang G, Di Giandomenico S, Lee JM, Deblasio A, Menendez S, Antipin J, Reva B, Koff A, Nimer SD · Cell Stem Cell 2009
    PMID 19128791DOI
    What this evidence cannot show

    Murine, steady-state, germline Trp53-null, and the headline quiescence phenotype is sharpest in a compound genetic background (Mef/Elf4-null) rather than in wild-type animals. No leukaemia, no fusion, no human material. Gfi-1 already has an entry in this graph, which makes the target link checkable rather than decorative.

    Evidence that argues against this (1)
    • The direction does not fit this entry's clinical table, and the mismatch should be stated rather than smoothed. p53 promoting quiescence predicts that TP53 loss pushes cells into cycle and therefore into chemosensitivity, while every clinical row in this entry shows TP53 alteration predicting worse outcome. The reconciliation would have to run through apoptotic resistance rather than cell cycle - which is what the MCL-1 preprint would predict and what nobody has demonstrated in KMT2A-rearranged disease. Until that is done, the murine quiescence result should not be used to reason about TP53-mutant relapse here. Vaseva AV et al. 2012
Connected to