Tumor-agnostic genomic and clinical analysis of BRAF fusions identify actionable targets

Authors

Chen MF

Yang SR

Tao JJ

Desilets A

Diamond EL

Wilhelm C

Rosen E

Gong Y

Mullaney K

Torrisi J

Young RJ

Somwar R

Yu HA

Kris MG

Riely GJ

Arcila ME

Ladanyi M

Donoghue MTA

Rosen N

Yaeger R

Drilon A

Murciano-Goroff YR

Offin M

Doi

10.1158/1078-0432.CCR-23-3981

PMID: 38922339 · DOI: 10.1158/1078-0432.CCR-23-3981 · Journal: Clinical Cancer Research (2024)

TL;DR

This study performed a tumor-agnostic genomic and clinical analysis of all BRAF fusions detected across 97,024 samples sequenced at Memorial Sloan Kettering Cancer Center. The authors identified 241 BRAF fusion-positive tumors from 212 patients spanning 52 histologies with 82 unique 5’ fusion partners, of which 39 were novel. BRAF fusions were enriched in pilocytic astrocytomas (56% prevalence), and durable responses to MAPK-pathway directed therapies (MEK inhibitors in particular) were observed in pilocytic astrocytoma patients. Fifteen patients acquired BRAF fusions as a resistance mechanism to prior targeted therapy, predominantly EGFR TKIs in lung adenocarcinoma.

Cohort & data

  • 97,024 samples from 69,337 patients sequenced between January 2014 and November 2022 at MSK via MSK-IMPACT, MSK-ACCESS, and MSK-Fusion panels.
  • 212 patients with oncogenic BRAF fusions (195 de novo, 17 detected post-targeted therapy; 15 confirmed acquired).
  • 241 BRAF fusion-positive tumors across 52 histologies.
  • Data deposited in cBioPortal as braf_msk_archer_2024 and braf_msk_impact_2024.
  • 24 patients with de novo BRAF fusions treated with MAPK-pathway directed therapies were evaluable for clinical outcomes (RECIST v1.1).

Key findings

  • De novo BRAF fusion prevalence was <1% (195/69,337) across all cancers; highest in pilocytic astrocytomas (56%, N=29), gangliogliomas, low-grade neuroepithelial tumors, and acinar cell carcinoma of the pancreas (each >=5%).
  • 82 unique 5’ fusion partners identified; 39 (48%) were novel. 29% of partners were non-recurrent (identified only once).
  • Most common 5’ partners: KIAA1549 (25%, N=48), SND1 (10%, N=20), AGK (6%, N=12), MKRN1 (5%, N=9), TRIM24 (5%, N=9).
  • 90% of pilocytic astrocytoma BRAF fusions involved KIAA1549-BRAF.
  • Fusion partner distribution was histology-specific: SND1 in prostate adenocarcinoma (21%), TRIM24 in colorectal cancer (43%), SND1 in pancreatic adenocarcinoma (56%).
  • Co-alterations: TP53 mutations (22%), TERT mutations (18%), CDKN2A deletions (14%), CDKN2B deletions (12%). BRAF fusions were mutually exclusive with other MAPK pathway alterations.
  • In colorectal cancers, frequent co-mutations included RNF43 (64%), TP53 (57%), KMT2D (43%), MSH3 (42%), ARID1A (36%).
  • Of 20 evaluable patients treated with MAPK therapies: PR (N=2), SD (N=11), PD (N=7).
  • Median time on MEK inhibitors alone (N=9): 8 months (range 1-26). Median time on combination MEK+BRAF inhibitors (N=11): 1 month (range 0-18).
  • All 6 pilocytic astrocytoma patients remained on MEK inhibitor therapy >6 months (median 11 months, range 6-26).
  • 15 patients acquired BRAF fusions after targeted therapy: 10 with EGFR-mutant LUAD, and 5 across other histologies (COAD, THPA, BLCA, LUAD with NTRK1 fusion, LUAD with BRAF V600E).
  • Median time from EGFR TKI start to acquired BRAF fusion detection: 23 months (range 9-37).

Genes & alterations

  • BRAF: Oncogenic fusions (class II alterations) across 52 histologies; 82 unique 5’ partners. Fusions activate MAPK pathway via loss of auto-inhibitory N-terminal region leading to constitutive dimerization.
  • EGFR: Exon 19 deletions (N=8), L858R (N=1), G719C/L861Q (N=1) in patients who subsequently acquired BRAF fusions as resistance mechanism to EGFR TKIs.
  • TP53: Most common co-alteration (22% across all histologies; 57% in colorectal cancers).
  • CDKN2A / CDKN2B: Deletions co-occurring with BRAF fusions (14% and 12%, respectively).
  • TERT: Mutations co-occurring in melanomas (64%), thyroid cancers (73%), and gliomas (11%).
  • KIAA1549: Most common 5’ fusion partner (25%), dominant in pilocytic astrocytomas (90%).
  • RNF43, KMT2D, MSH3, ARID1A: Frequent co-mutations in BRAF fusion-positive colorectal cancers.
  • NTRK1, FGFR3: Driver fusions in patients who subsequently acquired BRAF fusions at resistance.

Clinical implications

  • BRAF fusions represent an emerging actionable target across histologies, though response rates to currently available MAPK therapies are lower than for other fusion-driven cancers (ALK, ROS1, RET: 60-80% ORR).
  • MEK inhibitors (trametinib, selumetinib) provided durable benefit in pilocytic astrocytomas (all 6 patients >6 months on therapy); combination BRAF+MEK inhibition was less durable (median 1 month).
  • BRAF fusions are a recurrent (2-4%) mechanism of acquired resistance to EGFR TKIs (osimertinib, erlotinib, afatinib), underscoring the importance of post-progression genotyping.
  • Combination strategies (EGFR TKI + MEK inhibitor) at resistance showed preliminary signals: one patient remained on erlotinib + trametinib for 12 months.
  • New therapeutic approaches targeting BRAF fusions are in development, including BRAF dimer blockers, ERK inhibitors, and PROTACs.
  • Partner-agnostic diagnostic strategies (RNA-based NGS) are critical given the extraordinary diversity of BRAF fusion partners.

Limitations & open questions

  • Retrospective, single-institution design with heterogeneous patient populations analyzed at various timepoints during care.
  • Small numbers of patients per treatment type limit statistical power for response comparisons.
  • No head-to-head comparison of MEK inhibitor monotherapy versus combination BRAF+MEK inhibition.
  • The study did not systematically compare diagnostic assay performance (DNA vs. RNA vs. ctDNA NGS) for BRAF fusion detection sensitivity.
  • Optimal combination strategy for acquired BRAF fusions at resistance (e.g., EGFR TKI + MEK inhibitor vs. other approaches) remains undetermined.
  • The influence of specific fusion partners on treatment response and biology requires further investigation with functional studies and larger clinical datasets.
  • Response appeared histology-dependent, paralleling BRAF V600E experience, but the mechanistic basis for this differential sensitivity is unclear.

Citations from this paper used in the wiki

  • “We found 241 BRAF fusion-positive tumors from 212 patients with 82 unique 5’ fusion partners spanning 52 histologies. 39 fusion partners were not previously reported, and 61 were identified once.” (Abstract)
  • “The prevalence of de novo BRAF fusions was <1% (195/69,337). […] The tumor type most enriched for BRAF fusions was pilocytic astrocytoma (N=29, prevalence 56%).” (Results)
  • “BRAF fusions were most commonly co-altered with TP53 mutations (22%), TERT mutations (18%), CDKN2A deletions (14%), and CDKN2B deletions (12%).” (Results)
  • “The median time on combination MEK and BRAF inhibitors (N=11) therapies was 1 month (range 0-18 months) and for MEK inhibitors (N=9) was 8 months (range 1-26 months).” (Results)
  • “Nine patients remained on treatment for >6 months, including all the patients with pilocytic astrocytomas (6/6).” (Results)
  • “Fifteen patients acquired BRAF fusions following targeted therapy [EGFR-mutant lung adenocarcinoma (N=10)…]” (Results)
  • “The median time from start of targeted therapy to BRAF fusion detection was 23 months (range 9-37 months).” (Results)

This page was processed by crosslinker on 2026-05-04.