优化精准肿瘤学：对37,898例患者中WES与靶向panel的大规模临床基因组学分析

Background: Next-generation sequencing (NGS) is now routine in oncology practice, but clinicians still face uncertainty about how whole-exome sequencing (WES) compares with large targeted panels for identifying actionable alterations and guiding immunotherapy decisions. We addressed this by benchmarking these platforms within a massive clinico-genomic cohort to define their specific clinical utility. Methods: We retrospectively analyzed 37,898 tumor-normal pairs tested in a CAP/CLIA-certified laboratory: 3,360 WES tests (covering ~20,000 genes) and 34,538 hybrid capture targeted panels (covering ~500 cancer genes). We compared detection rates for single nucleotide variants (SNVs), copy number variants (CNVs), and rearrangements. Clinical variables, including stage and treatment history, were integrated to evaluate molecular patterns. Results: WES cases were enriched for lung adenocarcinoma, gastric, and colorectal cancers, whereas panel testing was dominated by lung adenocarcinoma, colorectal cancer, and hepatocellular carcinoma. Across all tumors, 701,119 somatic alterations were identified. WES yielded a mean of 76.7 alterations per patient and panels 12.8 per patient. Although WES covered the entire exome and detected clinically expected events such as >20% ERBB2 amplification in breast cancer and >4% ALK fusions in lung adenocarcinoma, its overall CNV and fusion detection remained clearly lower than panel, which showed much higher sensitivity due to deeper, specially targeted coverage. Mean TMBs were 6.5 muts/Mb (WES) and 8.8 muts/Mb (panel), with good agreement between platforms; overall, 17.0% of patients were TMB-High (≥10 muts/Mb). On both platforms, TMB-H was most frequent in endometrial carcinoma, lung squamous cell carcinoma, and urothelial carcinoma, while pancreatic and clear cell renal carcinomas showed very low TMB. Importantly for clinical decision making, early stage patients also harbored TMB-H tumors, indicating that potential candidates for immune checkpoint inhibitors are not restricted to advanced disease. Treatment-naïve patients had higher TMB-H rates than previously treated patients (22.5% vs 16.2%), consistent with therapy-driven clonal selection and supporting baseline TMB testing at initial diagnosis. The most frequently mutated genes were similar across platforms ( TP53 , EGFR, KRAS ). Actionable alterations were common: 68.1% of panel cases carried at least one actionable gene. Conclusions: In this largest real world clinico-genomic cohort of Chinese cancer patients, we define the complementary utility of each platform. Whereas targeted panels excel at detecting clinically actionable alterations, WES delivers a comprehensive genomic blueprint that is well suited as a baseline for future minimal residual disease monitoring and the identification of novel biomarkers.