DNA-level complex fusions show high discordance with RNA expression: A validation study of 115 Chinese cases

Background: Gene fusions identified by DNA-based next-generation sequencing (DNA-NGS) play a crucial role in guiding targeted therapy for solid tumors. However, complex fusion structures may not always transcribe into functional chimeric RNAs or proteins, due to disrupted open reading frames, opposing transcriptional directions, or absence of promoter elements. Such discrepancies can lead to false-positive DNA-NGS findings and subsequent inappropriate treatment decisions. Methods: We analyzed 115 tumor samples from the Chinese population with DNA-NGS-identified complex fusions using simultaneous DNA and RNA-based NGS (DR-NGS) for validation. Seven samples with inadequate pathology quality control were excluded. Results: Lung cancer accounted for the majority of the 115 samples (89, 77.4%). Complex fusion patterns included 5′-5′ fusions, 3′-3′ fusions, primary/reciprocal fusions, non-canonical breakpoints or partners, co-occurrence of multiple fusion drivers, UTR-region fusions, and intergenic fusions. The involved genes were ALK (32), ROS1 (10), NTRK (6), EGFR (9), RET (28), HER2/3 (2), MET (4), FGFR1-4 (15), NRG1 (1), BRAF (4), and multiple co-existing fusion genes (3). Among the 108 successfully tested cases, 45 showed no fusions by DR-NGS testing but revealed other driver mutations in 57.78% of cases, including EGFR (22, 48.89%), FGFR2 (1), BRAF (1), and KRAS (2). In the remaining 63 cases, DR-NGS testing confirmed the presence of fusions, with only one case showing a co-existing EGFR mutation (1.59%, p < 0.001). Among these, 55 exhibited classic single fusions at the RNA level after transcriptional splicing, while 8 cases with rare fusion partners in DNA-level were validated as true partners by DR-NGS testing. RNA fusion-positive and -negative groups showed male-to-female ratios of 21:42 and 23:22 (p = 0.064), and age distributions (≥65 vs. <65 years) of 9:54 and 20:25 (p < 0.001), respectively. Conclusion: This study reveals a substantial discordance between DNA-level complex fusions and their RNA expression. Nearly half of the DNA-identified complex fusions were not transcribed, with many instead harboring other driver mutations-most commonly EGFR. By contrast, RNA-verified fusions predominantly presented as classic single transcripts, with rare co-occurrence of other driver mutations. RNA fusion status was also significantly associated with patient age. These findings underscore the necessity of RNA-based confirmation for DNA-detected complex fusions, especially those coexisting with other driver gene mutations, to prevent false-positive interpretations and to optimize targeted therapy selection.