Real-world landscape of BRAF fusions in chinese patients with lung and colorectal cancers: Insights from 26,248 DNA-NGS and RNA-NGS tests

Background: Although rare, BRAF fusions are highly actionable alterations with strong responses to targeted therapy, particularly in lung and colorectal cancers. Their significant clinical relevance makes accurate detection essential, despite challenges posed by their low frequency and heterogeneous structures across sequencing technology. This study provides a large real-world analysis of BRAF fusion characteristics in Chinese lung and colorectal cancer patients, focusing on fusion partners, co-mutations, and differences between DNA-based and RNA-based next-generation sequencing (NGS). Methods: A total of 26,248 patients underwent NGS testing, including 11,337 analyzed by DNA-NGS and 14,911 by RNA-NGS. BRAF fusions, partner genes, breakpoint patterns, and co-occurring driver alterations were systematically evaluated. Detection performance and fusion configuration differences were compared between DNA-NGS and RNA-NGS. Results: Among all patients, 55 individuals (0.43%) carried 65 distinct BRAF fusions, with a higher proportion of female patients (58.5%). IMMP2L was the most common fusion partner (11.1%), followed by TRIM24 (7.9%), CD74 (6.3%), and MKRN1 (6.3%). Co-mutations were identified in 30 fusion-positive patients, 90% of whom had lung cancer, consistent with reports suggesting BRAF fusions often emerge as secondary events after targeted therapy.Overall BRAF fusion prevalence did not differ significantly between DNA-NGS and RNA-NGS (0.24% vs. 0.19%, p = 0.45). However, the two platforms showed marked differences in fusion structure. All RNA-NGS-detected fusions were canonical 5′-3′ events, with BRAF exon 10 and exon 2 as the most frequent breakpoints (28.6%). In contrast, only 55.6% of DNA-NGS-detected cases showed canonical 5′-3′ fusions. Six of these also carried reciprocal or non-reciprocal rearrangements, while 12 cases harbored only non-canonical reciprocal configurations, suggesting that several DNA-NGS-identified events may not be transcriptionally functional or may represent structural variants lacking RNA expression.Importantly, the most frequent subtype, IMMP2L-BRAF , was detected exclusively in the RNA-NGS cohort, suggesting potential technical limitations in resolving IMMP2L -related breakpoints at the DNA level and underscoring the methodological advantages of RNA profiling. Conclusions: This large real-world cohort outlines the BRAF fusion landscape in Chinese lung and colorectal cancers and reveals substantial discordance between DNA-NGS and RNA-NGS. The distinct partner spectrum and structural differences observed reinforce the value of RNA-based sequencing for accurate detection of clinically actionable BRAF fusions, supporting the integration of RNA analysis to optimize molecular testing and guide targeted therapeutic decisions.