一种利用cfDNA突变及整合基因组特征分析的靶向多癌种早期检测（MCED）检测的临床验证

Early cancer detection significantly improves survival, particularly for high-burden cancers across Asia that are often diagnosed late. Plasma-derived cell-free DNA (cfDNA) is a promising non-invasive biomarker for identifying cancer earlier than conventional methods. This study evaluated the performance of a targeted multi-cancer early detection (MCED) test that analyzes tumour-associated mutations in cfDNA to detect 20 cancer types commonly seen in Asian populations. In this retrospective, multicentre validation study, participants aged ≥20 years with either confirmed cancer or cancer-free status were recruited. Plasma cfDNA was sequenced using a targeted panel constructed from large genomic datasets. Machine learning models were trained and validated in a blinded manner to assess diagnostic performance. Among 2342 cancer cases and 1646 controls, the MCED test achieved a sensitivity of 83.1% (95% CI: 81.5-84.6%) and specificity of 95.4% (95% CI: 94.4-96.4%), with a false-positive rate of 4.6%. Performance surpassed that of traditional biomarkers such as CEA. Notably, high sensitivity was observed in cancers of major concern in Asia-liver, gastric, pancreatic, bile duct, and nasopharyngeal cancers-with early-stage detection ranging from 75.0-91.3% and late-stage detection from 95.5-100%. These findings support the potential of cfDNA-based MCED testing as a non-invasive, complementary screening approach for populations in Asia. Its strong performance across multiple high-mortality cancers underscores its value for early detection, risk stratification, and integration into regional cancer control strategies.