Haplo-Insufficiency of RAS Genes Is the Determinant of Outcome in AML

Acute Myeloid Leukemia (AML) is a heterogeneous and highly refractory disease. Mutations of NRAS and KRAS genes are the most frequently mutated genes in AML but have failed to provide utility as prognostic biomarkers or therapeutic targets. Colgan et al. (ASH 2024) have shown that NRAS mutations lack prognostic significance, while in contrast, the presence of KRAS mutations is associated with adverse outcomes. We expand on this observation and, using a genomic data base of over 2000 children and adults with AML, provide a comprehensive evaluation of sequence and structural alterations associated with RAS mutations, demonstrating that haplo-insufficiency of RAS, either because of bi-allelic RAS mutations or associated loss of heterozygosity (LOH), is the determinant of disease outcome.

Genome and transcriptome data obtained through a combination of whole genome, transcriptome, targeted capture, and Archer sequencing was interrogated for RAS mutations. Survival analysis was performed in R (v4.3.2).

In a cohort of 2041 patients, 722 (35.3%) patients had NRAS or KRAS mutations: 548 (26.8%) NRAS and 240 (11.8%) KRAS. 66 (9.1%) had mutations in both genes. 54 (7.5%) of those with RAS mutations had mutations in 2 different codons: 10 G12+G13, 16 G12+Q61, 15 G13+Q61, and 14 non-canonical pairings. Within NRAS mutations, 39 (7.1%) patients had mutations in 2 different codons: 6 G12+G13, 12 G12+Q61, 14 G13+Q61, and 7 non-canonical pairings. 16 (6.7%) patients with KRAS mutations had mutations in 2 different codons: 4 G12+G13, 4 G12+Q61, 1 G13+Q61, and 7 non-canonical pairings. Interrogation of whole genome and transcriptome sequencing data demonstrated that the two mutations were always in different sequencing reads, suggesting bi-allelic vs. multi-clonality as the underlying mechanism.

Outcome evaluation for patients with RAS mutations demonstrated 5-year event-free survival (EFS) of 26% for those with single-gene double mutations vs. that of 49% for those with single mutations, which were identical to wild-type (WT) (p<0.03). Of the 548 NRAS mutants, double mutants had an EFS of 35% vs. that of 47% for the single mutants and 48% for NRAS WT (p=0.22). Of the 240 KRAS mutants, EFS for double mutants was 7% vs. that of 40% for single and 49% for KRAS WT (p=0.001). For either RAS gene, codon pairing did not impact survival (p=0.51 NRAS, p=0.26 KRAS).

The significance of single-gene double mutations was interrogated in KMT2A-rearranged AML (KMT2A-r) cases. Of the 371 KMT2A-r cases, 106 (28.6%) had NRAS mutations and 93 (25.1%) had KRAS mutations. All KMT2A-r patients with single-gene double RAS mutations (n=16) relapsed by 2.1 years, compared to an EFS of 30% for single mutants and 35% in those with no RAS mutations (p=0.003). Among NRAS mutants, all double mutants (n=10, 9.4%) relapsed by 1.3 years, compared to an EFS of 44% for single mutants and 46% for NRAS WT. Among KRAS mutants, all double mutants (n=7, 7.5%) relapsed by 2.1 years, compared to an EFS of 25% for single mutants and 40% for WT (p=0.0025).

Bi-allelic mutations would lead to lack of normal RAS function and resultant haplo-insufficiency. We reasoned a RAS mutation in the setting of a LOH may result in similar phenotype. As LOH would lead to allelic imbalance and high variant allele frequency (VAF), we identified 16 cases with a single RAS mutation with VAF >60%. Whole genome sequencing of DNA from a case with high VAF NRAS mutation demonstrated copy-neutral LOH of 1P as the underlying mechanism of allelic imbalance. Patients with RAS mutations with high VAF had an EFS of 25%, similar to those with single-gene double mutations. In combination, patients with single-gene double RAS mutations or with high VAF constitute 10% of RAS mutants and provide significant clinical information. We validated the association of haplo-insufficiency with outcome in the adult BeatAML data (n=942), where patients with single-gene double RAS mutations or with high VAF had an overall survival of 10%, similar to our observation in younger patients.

We demonstrate that single-gene, double, likely bi-allelic RAS mutations or associated LOH have adverse outcomes in AML. Presence of such haplo-insufficient RAS variants are particularly important in KMT2A-r, where they all relapse. Despite history of therapeutic failures of RAS inhibitors in RAS mutant AML, this observation brings into question whether only those with haplo-insufficient RAS are uniquely susceptible to RAS inhibitors.