ALK融合断点与亚型的图谱：DNA-NGS与RNA-NGS的比较分析

Background: ALK fusions are important therapeutic biomarkers across solid tumors, and accumulating evidence shows that fusion breakpoints, isoforms, and fusion orientations influence response to ALK tyrosine kinase inhibitors (TKIs). Non-canonical or reciprocal fusions and rare breakpoints may yield nonfunctional transcripts and lead to inferior outcomes. Understanding the differences in ALK fusion profiles detected by DNA-based and RNA-based next-generation sequencing (NGS) is therefore critical for optimizing molecular testing and guiding targeted therapy. Methods: A total of 34,085 solid tumor patients were analyzed, including 15,957 tested by DNA-NGS and 18,128 by RNA-NGS. ALK fusion positivity, fusion orientations, and characteristics of subtypes were compared between DNA- and RNA-based sequencing approaches. Results: DNA-NGS identified 358 ALK-positive patients carrying 547 fusions, while RNA-NGS identified 469 patients with 482 fusions. RNA-NGS showed a significantly higher positivity rate than DNA-NGS (2.6% vs. 2.2%, p = 0.043). The two approaches differed markedly in fusion multiplicity: 57.5% of DNA-positive patients harbored multiple fusions, including one patient with eight partners, whereas only 2.5% of RNA-positive patients showed multiple events. Additionally, 27 DNA-NGS cases carried exclusively non-canonical reciprocal or non-reciprocal fusions, suggesting potential nonfunctional variants that complicate therapeutic interpretation.Fusion orientation and partner diversity also varied substantially. All RNA-detected fusions were canonical 5′→3′ events, involving 11 partner genes; EML4 accounted for 97%, followed by KLC1, DCTN1, and KIF5B (each <1%). Conversely, the DNA-NGS cohort included 3.8% 3'→3' fusions, 5.6% 5'→5' fusions, and among the 5'→3' events, only 70% represented canonical forward fusions. In total, 38 partner genes were identified in the DNA group, with EML4 remaining most prevalent (85.5%), followed by KIF5B (2.0%) and STRN (1.2%). Only five fusion partners overlapped between the two detection modalities.Breakpoint diversity was greater in DNA-NGS (21 breakpoint types vs. 17 in RNA-NGS). Although EML4-ALK isoform distributions were similar ( p = 0.55), DNA-NGS identified more atypical short-variant breakpoints (3.1% vs. 2.1%), which are associated with lower protein expression and poorer TKI response. Conclusions: RNA-NGS achieved a higher detection rate and a cleaner, more biologically meaningful ALK fusion profile, with fewer multi-partner or non-canonical events and predominately functional fusion isoforms. DNA-NGS frequently identified complex or potentially nonfunctional structures that may mislead therapeutic decisions. These findings suggest that RNA-NGS provides a more accurate representation of transcriptionally active ALK drivers and may be better suited to guide ALK-targeted therapy.