BRAF-Mutated Acute Myeloid Leukemia (AML) Represents a Prognostically Poor Subgroup Enriched for Myelodysplasia-Related Subtypes and Distinct from Other RAS Pathway Mutant AML

Background: MAPK/RAS signaling pathway mutations are common somatic events in de novo and secondary acute myeloid leukemia (AML), occurring mainly in NRAS, KRAS, and/or PTPN11. BRAF mutations occur with lower frequency in AML and are reported to be universally associated with poor outcome. To date, no comprehensive analysis exists that decisively assesses clinical and molecular characteristics of BRAF-mutated (BRAFm) AML, including insights into clonal architecture, co-occurring mutations, and disease immunophenotype.

Methods: We identified 42 AML pts harboring BRAF mutations through targeted molecular profiling of patients (pts) from our cooperative group frontline trials (Cancer and Leukemia Group B/Alliance for Clinical Trials in Oncology, enriched for de novo AML) and large comprehensive cancer centers (enriched for relapsed/refractory (R/R) and secondary AML). All pts had detailed clinical annotations, cytogenetic information, and NGS-based targeted sequencing data available. Six BRAFm pts underwent additional genomic profiling including paired whole exome sequencing and total transcriptome sequencing. To delineate clonal composition and cell states in BRAFm AML, 9 pt samples were characterized using single-cell DNA+Protein multi-omic profiling.

Results: Our study cohort included 42 BRAFm pts with newly diagnosed (n=19, 45%), R/R (n=8, 19%) and secondary AML (n=15, 36%), with a median age of 67 years (range 19-84). BRAF mutations were most common in cases of myelodysplasia-related (MR)-AML (WHO 5th edition), accounting for 32 (76%) pts. Bulk NGS suggested that BRAF mutations were either dominant clonal or subclonal events, with variant allele frequency (VAF) of 1-83%. Interestingly, the majority (n=30, 71%) of pts had non-V600 BRAF mutations, instead harboring non-canonical mutations including G469 (n=12), D594 (n=7) and others (n=12). The most frequent co-occurring mutations were in TET2 (36%), ASXL1 (33%), NRAS (29%) and KRAS (26%). Whole exome sequencing revealed that BRAFm AML samples harbored a median mutational burden of 15 non-synonymous coding variants (range, 9-30), including 3 MR mutations and 6 RAS-pathway mutations. Single-cell multi-omic analysis revealed that BRAF mutations can co-occur with other RAS pathway mutations, including NRAS and KRAS, on a single cell level. However, PTPN11 mutations were found in mutually exclusive clones to BRAF mutations. Moreover, we found that BRAFm clones can exist in multiple differentiation states, including CD34⁺ and CD11b⁺ compartments unlike FLT3- or other RAS- mutant clones which show strong genotype-immunophenotype correlations. Functional studies have also evaluated the impact of the non-canonical BRAF mutations on clonal fitness and disease development.

Clinically, BRAFm pts were treated with high-intensity (HI)+/-venetoclax (VEN), [n=15, 42%], low intensity [LI, n=10, 29%] or LI+VEN [n=10, 29%] regimens. BRAFm AML pts had extremely poor survival, regardless of therapy with an overall composite remission (CR+CRi) rate of 39%. The entire cohort had a median overall survival (OS) of 7 months with a majority (n=31/35, 89%) of pts deceased at last follow-up. The median survival for the newly diagnosed cohort was 14 months. No significant differences in OS were observed based on treatment regimen. Importantly, we found no correlation between BRAF variant allele burden and OS. Given the enrichment of BRAF mutations in MR-AML, we assessed if RAS pathway mutations generally represented poor outcome prognosticators in this subgroup. Analysis of a control cohort of MR-AML pts with (n=129) and without (n=403) RAS pathway mutations treated on cytarabine/daunorubicin-based frontline protocols revealed no survival difference between mutated and wild-type pts. This finding suggests the negative survival impact observed in BRAFm AML pts is specifically due to the presence of the BRAF mutation.

Conclusions: BRAF mutations are associated with poor prognosis regardless of clonal burden, without significant therapeutic advantage of currently available treatments. Our profiling results suggest that while BRAFm AML may harbor a RAS pathway addiction, BRAF mutations have distinct impacts compared to other RAS pathway mutations. Our findings highlight the need to further understand the role of BRAF mutations in AML pathogenesis and assess the utility of novel therapeutic strategies in BRAFm AML.