基于扩增子的DNA与RNA联合检测相较于杂交捕获DNA-NGS改善中国实体瘤中ROS1融合的识别

Background: ROS1 fusions are actionable drivers across many solid tumors, with multiple targeted therapies approved, yet conventional DNA assays show limited sensitivity. Although RNA-NGS improves fusion detection, the relative performance of amplicon-based DNA+RNA NGS co-detection versus hybrid-capture DNA-NGS for ROS1 fusion identification across diverse tumors remains unclear. Methods: We analyzed a total of 16,424 solid tumor samples using a 35-gene amplicon-based DNA+RNA NGS co-detection assay, including BTC (n=310), COAD (n=3,839), ESCA (n=83), LIHC (n=123), NSCLC (n=11,242), SARC (n=61), and STAD (n=766). In parallel, 4,624 tumors were assessed using a hybrid-capture DNA-NGS assay, including BTC (n=398), COAD (n=857), ESCA (n=83), LIHC (n=269), NSCLC (n=2,504), SARC (n=67), and STAD (n=446). ROS1-positive cases were identified for each tumor type, and fusion orientation and partner genes were characterized to evaluate differences in fusion detection performance between the two sequencing strategies. Results: Amplicon-based DNA+RNA NGS co-detection identified 217 ROS1-positive tumors, yielding detection rates of 0.32% in BTC (1/310), 0.10% in COAD (4/3,839), 2.41% in ESCA (2/83), 0.81% in LIHC (1/123), 1.82% in NSCLC (205/11,242), 3.28% in SARC (2/61), and 0.26% in STAD (2/766). All detected ROS1 fusions were canonical 5′-3′ events. Across all fusion-positive tumors, the predominant partner genes were CD74-ROS1 (46.08%), EZR-ROS1 (21.66%), SDC4-ROS1 (16.13%), SLC34A2-ROS1 (3.23%), CCDC6-ROS1 (3.23%), GOPC-ROS1 (2.30%), TPR-ROS1 (0.92%), EML4-ROS1 (0.46%), LRIG3-ROS1 (0.46%), PPFIBP1-ROS1 (0.46%), and ZCCHC8-ROS1 (0.46%). In contrast, hybrid-capture DNA-NGS identified only 43 ROS1-positive tumors, with detection rates of 0.75% in BTC (3/398), 0% in COAD (0/857), 0% in ESCA (0/83), 0.37% in LIHC (1/269), 1.56% in NSCLC (39/2,504), 0% in SARC (0/67), and 0% in STAD (0/446). All fusions detected by DNA-NGS were classical 5′-3′ rearrangements, and the dominant partner genes were CD74-ROS1 (53.49%), SDC4-ROS1 (18.60%), EZR-ROS1 (11.63%), GOPC-ROS1 (6.98%), TPM3-ROS1 (2.33%), ROS1-NT5DC1 (2.33%), ROS1-SLC34A2 (2.33%), and ROS1-TBC1D32 (2.33%). Overall, the co-detection strategy demonstrated substantially higher fusion detection rates and broader partner diversity, particularly in ESCA, NSCLC, and SARC. Conclusion: Amplicon-based DNA+RNA NGS showed higher sensitivity and broader fusion-partner detection than hybrid-capture DNA-NGS for ROS1 rearrangements, underscoring the value of integrating RNA-based methods into routine testing to improve ROS1 fusion identification and support precision oncology.