KRAS as a co-mutation in KMT2A-r leukemia — the single most frequently co-mutated gene, and it still cannot start the disease

Discovery. In the 22-case whole-genome PCGP infant KMT2A-r ALL cohort, KRAS was the most frequently mutated gene after KMT2A itself, with n = 4 recurrent mutations — ahead of NRAS (n = 2), with FLT3, NF1, PTPN11 and PIK3R1 each appearing once 1.

Independent replication. In 124 infant B-ALL cases from Interfant-99/06, recurrent mutations were found exclusively in KRAS and NRAS: 38% of the 42-case discovery arm and 34% of the 82-case validation arm carried an activating mutation in one of the two 2. Unlike NRAS, KRAS was not reported as enriched in t(4;11) specifically; the t(4;11) enrichment in that cohort was an NRAS effect (32% vs 6%, P = 0.01) 2. See comut-nras-kmt2a-r.

Variant spectrum. Querying the public curation of the PCGP study (cBioPortal all_stjude_2015, 93 sequenced samples, retrieved 2026-09-18) returns 14 KRAS mutations across 12 samples, with the following distinct protein changes 3:

The presence of A146T and K117N matters practically: these are weaker, non-canonical activating alleles that hotspot-only panels covering codons 12/13/61 will miss, and they are also the alleles least well covered by the emerging KRAS-selective inhibitors. Note that 14 mutations across 12 samples means at least two samples carried more than one KRAS allele — consistent with the parallel subclonal architecture described in comut-ras-pathway-kmt2a-r.

The decisive experiment transduced human cord-blood CD34+ HSPCs with KMT2A::AFF1, KRAS-G12V, or both, and transplanted them intra-bone-marrow into immunodeficient mice 4.

• KRAS activation did not cooperate with KMT2A::AFF1 to immortalise CD34+ HSPCs in clonogenic or differentiation/proliferation assays.
• KMT2A::AFF1 alone, KRAS-G12V alone, and the combination all enhanced haematopoietic repopulation without impairing myeloid-lymphoid differentiation and without initiating leukemia.
• KRAS activation did enhance extramedullary haematopoiesis of KMT2A::AFF1-expressing lines and HSPCs, with leukocytosis and CNS infiltration — both clinical hallmarks of infant t(4;11) B-ALL — and transcriptional profiling of KMT2A::AFF1 patient samples supported a cell-migration gene signature underlying the mutant-KRAS phenotype 4.

The clinical genetics agrees on every point where it can be checked: RAS-mutant infants present with higher white-cell counts (P = 0.013) 5, the mutations are largely subclonal and arrive after the fusion 1, and NG2-high KMT2A-r infant blasts — a separate migration-associated marker — likewise associate with more circulating blasts and more frequent CNS disease and relapse 6. The consistent picture is that KRAS in this disease buys spread, not transformation.

In the PCGP cohort, of the 22 whole-genome cases only 10 were predicted to encode an in-frame reciprocal partner-gene fusion protein, and two of those involved genes with a known role in cancer. One of the two was KRAS-MLL 1. The public structural-variant curation of the same study independently contains a KRAS::KMT2A record 3.

So in at least one infant, the KRAS locus was not point-mutated but physically incorporated into the KMT2A rearrangement as the 5' partner of a reciprocal derivative product. Whether that product is expressed, stable, or functional is unknown and untested. It is included here because it is a reminder that co-lesion analysis restricted to point mutations will systematically miss structural involvement of the same genes. See cnv-reciprocal-derivative-chromosome-retention.

KRAS is the co-mutation most people reach for first because it is the most frequent and, since 2021, the most druggable RAS isoform. Three things argue against optimism in this specific disease.

1. The alleles are wrong for the available drugs. The approved and advanced KRAS-selective inhibitors target G12C (sotorasib, adagrasib) and G12D. In the observed spectrum, G12C does not appear at all in the KRAS variant list for this cohort (it appears in NRAS), and the spectrum is dominated by G12D/G12R/G12S/G12V, G13D, Q61H plus the atypical A146T and K117N 3. Pan-KRAS and G12D-selective agents would be required, none of which has infant pharmacology.
2. The clone is usually a minority and is often lost at relapse — two of five informative PCGP diagnosis-relapse pairs lost the pathway mutation, and two thirds of relapsing Interfant patients lost the diagnostic RAS mutation 1, 2.
3. The functional data say it is not load-bearing. A gene that cannot initiate and is not required for maintenance is a poor drug target even when frequent 4.

The more defensible route through KRAS in this disease is downstream and genotype-agnostic: MEK inhibition, which sensitises KMT2A-r blasts to prednisolone in both RAS-mutant and RAS-wild-type cells. That argument is developed in comut-ras-pathway-kmt2a-r.