Impact of sample characteristics on RNA-based next-generation sequencing (NGS) for fusion gene detection in non-small cell lung cancer (NSCLC).

3132 Background: RNA-based next-generation sequencing (NGS) has been widely employed for detecting fusion genes in NSCLC, due to its superior sensitivity and simplified design compared to DNA-based NGS. However, the impact of sample quality on fusion variant detection using RNA-based NGS remains unclear. Methods: The study analyzed 5,386 and 5,538 NSCLC samples using DNA- or RNA-based NGS to detect common fusion genes (ALK, RET, ROS1, NTRK, NRG1, MET exon 14 skipping, and FGFR). NGS libraries were constructed using capture-based or amplicon-based methods for DNA and RNA samples, respectively, focusing on pathogenic mutations. Results: RNA-based NGS detected 2.44% more fusions than DNA-based NGS [9.50% (526/5538) vs. 7.06% (380/5386)], with notable advantages for NTRK (0.13% vs. 0.02%), NRG1 (0.25% vs. 0.06%), MET exon 14 skipping (2.15% vs. 1.36%), and FGFR fusions (0.40% vs. 0.02%). Tumor cell content analysis showed no significant impact on fusion detection rates within the 20%-90% range for either method. However, higher tumor cell content (≥80%) significantly increased RNA-based NGS detection rates compared to DNA-based NGS, nearly doubling the total detection rate (17.3% vs. 8.88%), primarily due to increased ALK fusion detection (8.97% vs. 5.02%). The type of sampling (surgical, biopsy, or others) did not significantly affect overall fusion detection rates for either method (p > 0.05). However, gene-specific analyses showed significantly higher detection rates for ROS1, MET, and RET using RNA-based NGS in biopsy samples compared to DNA-based methods (ROS1: 11.83% vs. 1.18%, MET exon 14 skipping: 2.87% vs. 1.62%, RET: 1.24% vs. 0.79%). Conversely, RNA-based detection of ALK and NRG1 fusions was higher in surgical samples (ALK: 4.00% vs. 3.25%, NRG1: 0.34% vs. 0.08%) compared to DNA-based methods. Regarding sample types, pleural/peritoneal effusions showed higher detection rates than FFPE samples, though not statistically significant. RNA-based NGS consistently showed superior detection rates for ALK and MET exon 14 skipping in all sample types compared to DNA-based methods, with the most substantial increase for MET exon 14 skipping in pleural/peritoneal effusions (2.14% vs. 0.98%). Conversely, RNA-based NGS for NRG1 and ROS1 fusions showed a greater relative increase in detection rate in 10% neutral formalin-fixed tissue/FFPE sections/unstained slides compared to pleural/peritoneal effusions. Conclusions: Sample characteristics did not significantly impact the overall detection rate of RNA-based fusion assays. However, detection rates for specific fusions like ALK, NRG1, and MET exon 14 skipping varied with sample type, sampling method, and tumor cell content. Optimizing testing strategies and sample handling is crucial to improving diagnostic accuracy in NSCLC.