Oncofusion

start / graph / gene-idh1

Proteinstrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.Q81-methylome-axis

IDH1 - the same oncometabolite, the same assay, and NO enrichment in KMT2A-rearranged AML (6.7% vs 7.6%, p = 0.53): the control that makes the IDH2 result interpretable

also called IDH1 · isocitrate dehydrogenase (NADP(+)) 1 · IDHC_HUMAN · Isocitrate dehydrogenase [NADP] cytoplasmic · IDP · IDPC · PICD · Oxalosuccinate decarboxylase

IDH1 is the cytoplasmic twin of IDH2, and when mutated it makes exactly the same DNA-methylation-raising molecule. In the biggest series that can test it, IDH1 mutations are no more common in KMT2A-rearranged leukaemia than in any other kind - which is the reason to be careful about the claim that IDH2 is.

The same thing, for a scientist

**IDH1 (O75874, 414 aa) exists in this graph as a negative control, and that is its value.** In the 3,863-case FoundationOne Heme AML series that finds IDH2 enriched in KMT2A-altered cases, **IDH1 is 6.7% of 521 KMT2A-altered versus 7.6% of 3,342 KMT2A-wild-type - not significant, and an independent Fisher exact recomputation gives odds ratio 0.876, p = 0.53** [E1][E2]. Rule 87 says the informative proportion is often not the mid-range one; here the informative row is a **null measured at high precision in the same assay, same cohort, same comparison** as the positive. Mutant IDH1 produces the identical oncometabolite D-2-hydroxyglutarate, competitively inhibits the same alpha-KG-dependent dioxygenases including the TET 5mC hydroxylases, and is grouped with IDH2 throughout the founding hypermethylation study [E3][E4]. **So a mechanism that runs purely through 2-HG and DNA hypermethylation predicts IDH1 to track IDH2, and it does not.** The candidate explanations are lesion type (43.1% of the KMT2A-altered arm is KMT2A-PTD and the table is not stratified), enzyme compartment (IDH1 is cytosolic and peroxisomal with experimental support; IDH2's mitochondrial assignment in UniProt is only ECO:0000250), statistical power at a lower base rate, and chance [E1][E5]. **None is excluded and the beat does not choose between them.** DepMap: IDH1 is not a dependency and not selective - mean gene effect -0.089 in KMT2A-rearranged lines, -0.076 in other leukaemia lines, -0.095 across 1,208 lines, rank 7,311 of 17,728 [E6]. Clinically IDH1 matters because **ivosidenib is an approved mutant-IDH1 inhibitor** with 30.4% CR/CRh in relapsed/refractory IDH1-mutant AML [E7] - and the flat co-mutation row is the reason there is no rationale to reach for it in KMT2A-rearranged disease.

What it looks like
loading AF-O75874-F1…
AlphaFold AF-O75874-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
Why a null belongs in this graph as its own entry

The granularity rule says one entry per thing. The thing here is not IDH1 the enzyme - it is the measured absence of IDH1 enrichment in KMT2A-rearranged AML, in the same instrument that measured the IDH2 enrichment. Written as a separate entry, it does three jobs that a sentence inside gene-idh2 could not:

  1. It is the internal control for the IDH2 claim. An assay that reports a difference for one gene and no difference for its near-identical paralogue is behaving like an assay rather than like a batch effect.
  2. It is a bounded negative with a real interval. 6.7% (Wilson 4.9-9.2) against 7.6% (6.7-8.5). This is not "nobody looked" - it is looked at 521 cases and 3,342 controls and found nothing, and the upper bound of the odds ratio excludes the size of effect seen for IDH2.
  3. It stops a future beat reaching for ivosidenib. An agent reading only gene-idh2 might reason "IDH inhibitors are approved, IDH is enriched, try them". The correct statement is IDH2, not IDH, and the difference is a whole drug.
What the paralogue difference could mean, and why the beat will not pick one
explanationwhat would settle it
Lesion type. KMT2A-PTD arises in older, MDS-related AML; if IDH2 co-occurs with PTD and IDH1 does not, the whole signal is about duplications. 43.1% of the arm is PTD.Re-run the comparison in a fusion-only KMT2A cohort. Cheapest and most decisive.
Compartment. IDH1 is cytosolic and peroxisomal (UniProt ECO:0000269, PubMed 10521434); IDH2 is mitochondrial (ECO:0000250 only). Different compartments, different 2-HG flux to the nucleus.Measure intracellular and nuclear 2-HG in IDH1- versus IDH2-mutant cells of matched genotype.
Power. IDH1 is less frequent, and 6.7% vs 7.6% at these n excludes a large effect but not a small one.Larger series, or meta-analysis across KMT2A cohorts.
Chance. One significant row among ~17 genes with no multiplicity adjustment reported.Replication. IDH2 clears Bonferroni (0.05/17 = 0.0029) at p = 4.1e-4, so chance is the weakest of the four, but it is not zero.

Recording the four explicitly is the point. A single entry asserting IDH2 is enriched because 2-HG raises methylation would be wrong on its own terms, because the same mechanism predicts IDH1 and IDH1 is flat.

Facts and figures
uniprotO75874
gene symbolIDH1
protein nameIsocitrate dehydrogenase [NADP] cytoplasmic
length AA414
alphafold IDAF-O75874-F1
accession resolutionUniProt REST gene_exact:IDH1 AND organism_id:9606 AND reviewed:true, run 2026-09-21; one row; primary gene name read back as IDH1 and length 414 from the same fetched record.
carries cxxc domainfalse
classificationNeomorphic metabolic enzyme, cytosolic/peroxisomal paralogue of IDH2; mutant IDH1 produces the same oncometabolite D-2-hydroxyglutarate. Not a CXXC reader.
shown for fusionNO. Measured and flat: 6.7% vs 7.6% in adult KMT2A-altered versus wild-type AML, recomputed Fisher p = 0.53. This is a measured null in the index comparison, not an unexamined question - but no infant KMT2A::AFF1 cohort has ever reported IDH1 status.
cell typeAdult AML patients (FoundationOne Heme, n = 3,863; median age 62) and the ivosidenib phase 1 population (n = 258 treated). DepMap leukaemia lines. No paediatric or infant material.
cell type evidenceCohort genomics and one single-arm trial. No functional experiment in any KMT2A-rearranged cell.
oncometaboliteD-2-hydroxyglutarate
comutation
{
  "adult_kmt2a_altered_aml_f1h": {
    "k": 35,
    "n": 521,
    "pct": 6.7,
    "wilson_95": [
      4.9,
      9.2
    ],
    "k_note": "back-calculated from a reported 6.7%",
    "source_pmid": "41514668"
  },
  "adult_kmt2a_wt_aml_f1h_NULL": {
    "k": 254,
    "n": 3342,
    "pct": 7.6,
    "wilson_95": [
      6.7,
      8.5
    ],
    "source_pmid": "41514668"
  },
  "test": "Fisher exact, same assay: odds ratio 0.876, p = 0.53, NO DIFFERENCE",
  "role": "Internal negative control for the IDH2 enrichment. Same oncometabolite, same mechanism class, same instrument, no effect."
}
localization
{
  "subcellular": "Cytoplasm/cytosol and peroxisome, both ECO:0000269 from PubMed 10521434 (pointer only - that paper was not opened in this beat)",
  "subnuclear": "not applicable",
  "localization_shown_for_fusion": "NO - nobody has looked in any KMT2A-fusion system"
}
essentiality
{
  "depmap_scan": "analyses/depmap-kmt2a-selective-scan-full",
  "difference_kmt2a_r_minus_other": -0.0139,
  "mean_kmt2a_rearranged": -0.0894,
  "mean_other_leukemia": -0.0756,
  "pan_essentiality_all_lines": -0.0954,
  "n_all_lines": 1208,
  "rank_of_17728": 7311,
  "percentile": 41.24,
  "release_note": "Open Targets dataVersion 26.06, scan generated 2026-09-19"
}
druggable by
[
  "drug-ivosidenib"
]
How this could bite us back
  • Ivosidenib is approved and available, and an agent or a clinician reading 'IDH is enriched in KMT2A-rearranged AML' could reach for it. The measured statement is that IDH2 is and IDH1 is not, and ivosidenib does not inhibit mutant IDH2 - reaching for the wrong paralogue's drug is a real and cheap error to make.
  • Ivosidenib carries IDH differentiation syndrome (3.9% grade >=3) and QT prolongation (7.8%), and revumenib carries differentiation syndrome and QTc as well. Any hypothetical combination stacks both toxicities in a population that has neither a trial nor a rationale.
Still open
  1. Would an IDH1 enrichment appear in a fusion-only KMT2A cohort, where the KMT2A-PTD dilution is removed? The null measured here is in a 43%-duplication arm.
  2. Is the IDH1/IDH2 asymmetry reproduced in any other KMT2A series? One paper is one paper, and this row is the internal control for the beat's only positive finding.
  3. Do IDH1-mutant and IDH2-mutant AMLs differ in nuclear 2-HG concentration, as the compartment explanation requires? Not measured anywhere found in this beat.
  4. Has IDH1 status ever been reported in an infant or paediatric KMT2A-rearranged cohort? This beat found none, so the index disease has no estimate for either paralogue.
Every source behind this entry (7)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In 3,863 adult AML cases on one hybrid-capture assay, IDH1 alteration frequency did not differ between the 521 KMT2A-altered cases and the 3,342 KMT2A-wild-type cases - 6.7% versus 7.6%, reported as not significant - in the same table in which IDH2 is significantly enriched and TET2 significantly depleted.

    Characterizing KMT2A Rearrangement in Acute Myeloid Leukemia: A Comprehensive Genomic Study. · Batayneh O, Moein M, Naji NS, et al. · Cancers (Basel) 2026
    PMID 41514668DOI
    What this evidence cannot show

    Read in full text from PMC12784994 on 2026-09-21 including Table 2. Adult AML, median age 62; nothing about infant KMT2A::AFF1 ALL. Targeted commercial panel, not whole genome. 43.1% of the KMT2A-altered arm are duplications rather than fusions and the table is not stratified by lesion type, so this null is also a null about a mixed arm. A non-significant result at this n excludes a large effect, not a small one.

    Evidence that argues against this (1)
    • A null in a mixed KMT2A-altered arm could conceal opposite effects in the fusion and duplication subsets that cancel; the paper publishes no subset analysis that would expose that. Batayneh O et al. 2026
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.dataset

    Recomputed from counts back-calculated from the published percentages, IDH1 in KMT2A-altered versus KMT2A-wild-type AML gives 35/521 (6.7%, Wilson 4.9-9.2) versus 254/3,342 (7.6%, Wilson 6.7-8.5), odds ratio 0.876, two-sided Fisher p = 0.53 - against IDH2's odds ratio 1.630, p = 4.1e-4 in the same table.

    Independent Fisher exact and Wilson-interval recomputation of the KMT2A-altered versus KMT2A-wild-type co-mutation table of PMID 41514668 · Oncofusion knowledge-graph project, Q81-methylome-axis · Recomputed 2026-09-21 from percentages read out of PMC12784994 Table 2; counts back-calculated as round(pct x n), accurate to about one case 2026
    What this evidence cannot show

    A recomputation from marginals inherits every property of the original assay and adds about one case of rounding error per cell. It validates internal consistency (it reproduces Table 2 for IDH2, TET2 and DNMT3A) but cannot detect an error in the authors' own counting.

    Evidence that argues against this (1)
    • A p of 0.53 is consistent with a true odds ratio anywhere from roughly 0.6 to 1.25, so this null does not exclude a modest IDH1 enrichment; it excludes one the size of IDH2's. Batayneh O et al. 2026
  3. E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    IDH1 and IDH2 mutations are treated as one mechanistic class in the founding study: both produce 2-hydroxyglutarate, both give global DNA hypermethylation and a specific hypermethylation signature in AML, both are mutually exclusive with TET2 mutation, and both impair TET2 catalytic function in cells.

    PMID 21130701DOI
    What this evidence cannot show

    Adult AML, 2010; methylation by HELP rather than bisulfite; no KMT2A-rearranged subgroup analysed. Grouping IDH1 with IDH2 is the paper's own analytical choice and is exactly the assumption that the flat IDH1 row in the 2026 KMT2A comparison puts under strain.

    Evidence that argues against this (1)
    • The 2026 KMT2A comparison shows the two paralogues behaving differently with respect to KMT2A rearrangement, which is evidence that the one-class treatment is an approximation - at least for co-occurrence, whatever is true of the methylome phenotype. Batayneh O et al. 2026
  4. E4established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    Tumour-derived IDH1 mutants accumulate 2-hydroxyglutarate, which competitively inhibits alpha-ketoglutarate-dependent dioxygenases including histone demethylases and the TET 5mC hydroxylases; in glioma, IDH1 mutation is associated with increased histone methylation and decreased 5-hydroxymethylcytosine.

    PMID 21251613DOI
    What this evidence cannot show

    Biochemistry and ectopic expression; the cellular correlation is in glioma. Nothing leukaemic and nothing in a KMT2A-rearranged system.

    Evidence that argues against this (1)
    • If IDH1 and IDH2 mutants are biochemically equivalent 2-HG producers, then the differing co-occurrence with KMT2A rearrangement cannot be a 2-HG effect and must come from lesion type, cell of origin or chance - which means the mechanism most often invoked for the IDH2 finding is the one this paper rules out as an explanation for the difference. Batayneh O et al. 2026
  5. E5established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    Human IDH1 resolves to a single reviewed UniProt entry O75874 (IDHC_HUMAN), primary gene name IDH1, 414 aa, with experimentally supported subcellular localisation to cytoplasm/cytosol and peroxisome (ECO:0000269, PubMed 10521434) - an experimental code its paralogue IDH2 does not carry for any location.

    What this evidence cannot show

    Fetched 2026-09-21 via gene_exact:IDH1 AND organism_id:9606 AND reviewed:true; one row; primary gene name and length read back from the same fetched record. Rule 91: PubMed 10521434 was NOT opened in this beat, so the evidence code is recorded as a pointer and not as a paper this beat read. The ft_domain and ft_zn_fing fields returned no features, so no residue coordinates are asserted and no viewer lane is created.

    Evidence that argues against this (1)
    • Rule 91 states directly that an ECO:0000269 means a curator read an experiment somewhere in that paper, not that the paper is about the fact; PubMed 10521434 was not opened here, so the localisation should be treated as a database pointer rather than a read observation. UniProt Consortium 2026
  6. E6moderate confidence 3 of 5: One strong study, or several smaller consistent ones.dataset

    IDH1 is neither essential nor KMT2A-selective in DepMap: mean Chronos gene effect -0.0894 in the 11 KMT2A-rearranged lines, -0.0756 in the 43 other leukaemia lines, -0.0954 across all 1,208 lines, rank 7,311 of 17,728.

    Genome-wide KMT2A-rearranged selectivity scan, 17,728 genes (DepMap CRISPR Chronos via the Open Targets Platform GraphQL API) · Oncofusion knowledge-graph project · analyses/depmap-kmt2a-selective-scan-full/results.json, generated 2026-09-19T10:12:06Z; Open Targets dataVersion year 26 month 06 2026
    What this evidence cannot show

    A knockout scan of a wild-type gene says nothing about a neomorphic mutant allele, which is what matters clinically for IDH1. None of the 11 KMT2A-rearranged lines is infant-derived; the arm is myeloid-enriched. Open Targets dataVersion 26.06; DepMap's own release identifier is not returned by that endpoint. Rule 84: rank 7,311 of 17,728 is in the selective half by construction for roughly half of all genes and carries no information on its own.

    Evidence that argues against this (1)
    • Ivosidenib produces durable remissions in IDH1-mutant AML, so mutant IDH1 is unambiguously a therapeutic dependency in the right genotype even though wild-type IDH1 knockout has no effect - the scan and the clinic are measuring different things. DiNardo CD et al. 2018
  7. E7established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    Ivosidenib, an oral mutant-IDH1 inhibitor, produced a 30.4% CR/CRh rate (95% CI 22.5-39.3) in 125 patients with relapsed or refractory IDH1-mutated AML, with IDH differentiation syndrome in 3.9% and QT prolongation in 7.8% of 179 treated relapsed/refractory patients.

    Durable Remissions with Ivosidenib in IDH1-Mutated Relapsed or Refractory AML. · DiNardo CD, Stein EM, de Botton S, et al. · N Engl J Med 2018
    PMID 29860938DOI
    What this evidence cannot show

    Adults with IDH1-mutant AML; no KMT2A-rearranged subgroup is reported and none would be expected given the flat co-mutation row. Single-arm phase 1 dose-escalation and expansion. NCT02074839.

    Evidence that argues against this (1)
    • No preexisting co-occurring single gene mutation predicted response or resistance in that study, so nothing in it supports selecting or excluding patients by co-mutation - including by KMT2A status. DiNardo CD et al. 2018
Connected to