Oncofusion

start / graph / ana-q199-no-gwas-in-the-catalog-studies-kmt2a-rearranged-or-infant-leukaemia-while-fusion-defined-subtypes-have-their-own-scans

Analysisstrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.

None of the 166 leukaemia-family GWAS in the Catalog names KMT2A, MLL, 11q23 or infants, although ETV6::RUNX1, TCF3::PBX1 and hyperdiploid ALL each have their own scans - and a 987-case scan could only detect odds ratios of about 1.4-1.6 or more

also called no KMT2A GWAS · germline risk of infant leukaemia is unmapped

Researchers have mapped inherited risk for several childhood leukaemia subtypes, but not for the KMT2A-rearranged leukaemia of babies. A study that size would only pick up fairly strong inherited effects.

The same thing, for a scientist

Whole-word search of study title, trait, mapped trait and sample-size text across 166 leukaemia-family studies in the GWAS Catalog release of 2026-10-05: KMT2A/MLL/11q23 = 0 studies, 'infant' = 0. Positive control: the same search returns 6 studies for ETV6::RUNX1, TCF3::PBX1 and hyperdiploid. Sensitivity: with 5 controls per case, 80% power and alpha 5e-8, a 987-case scan detects OR >= 1.39 (MAF 0.30) or 1.57 (MAF 0.10).

Go deeper
What the instrument can and cannot see

The search covers the Catalog's curated text fields only. The positive control matters: it proves the search can find a fusion-defined subtype when one exists, so a zero for KMT2A is a zero and not a blind spot. It does not show that KMT2A-rearranged cases are absent from the large unsplit scans - they are very likely inside them, unlabelled.

Facts and figures
scriptanalyses/q199-germline-genetics-of-the-fusion-pathway/run.py
absence classliterature-absence-bounded
catalog rows1198394
leukaemia studies166
positive control studies6
power
5,321 cases (the largest …1.2405,321 cases (the largest B-ALL scan in the Catalog): 1.24987 cases (the size of th…1.570987 cases (the size of the 2023 recombinome's infant-ALL stratum, as a thought experiment): 1.57a 300-case scan2.110a 300-case scan: 2.11
power — the numbers behind the chart
{
  "a 300-case scan": {
    "maf_0.10": 2.11,
    "maf_0.30": 1.78
  },
  "987 cases (the size of the 2023 recombinome's infant-ALL stratum, as a thought experiment)": {
    "maf_0.10": 1.57,
    "maf_0.30": 1.39
  },
  "5,321 cases (the largest B-ALL scan in the Catalog)": {
    "maf_0.10": 1.24,
    "maf_0.30": 1.16
  }
}
How this could bite us back
  • No actionable inherited-risk or pharmacogenomic marker is available specifically for KMT2A-rearranged infant leukaemia: any such claim today would rest on subtype-agnostic scans.
Still open
  1. Do any of the large unsplit ALL scans report KMT2A-rearranged cases as a subgroup in their supplements, and could a meta-analysis across them reach useful power?
  2. Is there germline DNA (dried blood spots, remission samples) with matched fusion status in a consortium such as Interfant that could support even a 300-case scan for large effects?
Every source behind this entry (2)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

    Across 166 leukaemia-family studies in the GWAS Catalog (release of 2026-10-05), none names KMT2A, MLL, 11q23, mixed-lineage or infants in its title, trait, mapped trait or sample-size text, while the same search finds 6 studies for the fusion-defined subtypes ETV6::RUNX1, TCF3::PBX1 and hyperdiploid.

    What this evidence cannot show

    Whole-word patterns; acronyms (ALL, AML, CLL, MDS) matched case-sensitively. Two earlier selections were wrong and are recorded so they are not repeated: 'leuk' matched 'interLEUKin' and a case-insensitive ALL matched the word 'all', each pulling ~1,400 irrelevant studies into the leukaemia set. The largest B-ALL scan (5,321 cases) subdivides only by ETV6::RUNX1 and hyperdiploidy.

    Evidence that argues against this (1)
    • Absence from the Catalog is not absence from the literature. KMT2A-rearranged cases are almost certainly present, unlabelled, inside the unsplit childhood-ALL scans; a candidate-gene study, an unpublished dbGaP cohort, or a scan that reports subtypes only in a supplement would not appear. GWAS Catalog 2026
  2. E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.internal

    With 5 controls per case, a scan of 987 cases (the size of the 2023 recombinome's infant-ALL stratum, taken as a thought experiment and not as an existing cohort) has 80% power at alpha 5e-8 only for odds ratios of 1.39 or more at allele frequency 0.30, or 1.57 at 0.10; the catalogued median odds ratios are 1.76 for ARID5B and 1.67 for IKZF1 (detectable) but 1.34 for CEBPE and 1.34 for GATA3 (not).

    What this evidence cannot show

    Allelic-test normal approximation, equal allele-frequency assumption in controls, no covariates or imputation loss. Published odds ratios are the median over the Catalog's genome-wide-significant rows and are subject to winner's curse (inflated for the discovery study).

    Evidence that argues against this (1)
    • The 987-case figure is not an available dataset: the recombinome is a referral cohort assembled for diagnostics, with no genotype data. It is used only to size the question. The conclusion that a KMT2A-rearranged infant GWAS is feasible only for large effects holds for any cohort of that size. Meyer C et al. 2023
Connected to