The genomic and epigenomic abnormalities of plasma cfDNA as liquid biopsy biomarkers to detect hepatocellular carcinoma: a multicenter cohort study
De‐Zhen Guo, Ao Huang, Jianlong Sun, Shiyu Zhang, Jianwen Cheng, Ruijingfang Jiang, Yupeng Wang, Huihao Chen, Fengming Zhang, Jiaxi Peng, Jie Yuan, Jiefei Wang, Hongying Guo, Zhong Chen, Yongjun Chen, Xuxiao Chen, Shida Zhu, Jia Fan, Yuying Wang, Xin-Rong Yang, Jian Zhou
Abstract
Circulating tumor DNA (ctDNA) is a promising biomarker for early cancer detection; however, the optimal biomarker approach for early detection of hepatocellular carcinoma (HCC) remains unclear. Furthermore, current next-generation sequencing-based detection methods remain costly and complex, limiting their clinical utility for high-risk population screening. Hence, this study compared ctDNA mutation and methylation for HCC detection and aimed to develop a clinically accessible assay. A total of 1965 participants, including 629 HCC and 1336 control participants, were enrolled in five centers. Parallel ultra-deep targeted sequencing and targeted bisulfite sequencing revealed that a methylation-based ctDNA model significantly outperformed the mutation model in detecting HCC (sensitivity, 92.1% vs. 63.7%; P < 0.001), with no added benefit from combining both markers. We subsequently developed a multiplex PCR-based bisulfite amplicon sequencing assay (MBA-seq) using 25 selected methylation markers, achieving high diagnostic accuracy (AUC = 0.958, sensitivity = 86.7%, specificity = 90.1%) in the testing set. Further refinement through AUC-based selection yielded a two-marker panel (OTX1 and HIST1H3G) adapted into a quantitative methylation-specific PCR assay, designated HCCtect. This optimized model showed robust performance (AUC = 0.925, sensitivity = 78.4%, specificity = 93.0%), significantly surpassing alpha-fetoprotein (P < 0.001) while comparable with MBA-seq. HCCtect also demonstrated efficacy in early-stage HCC detection (sensitivity = 69.5%) and discrimination from chronic hepatitis B or liver cirrhosis (specificity = 93.0%). These findings establish ctDNA methylation as a superior approach over mutation analysis and highlight HCCtect as a promising non-invasive tool for HCC detection and high-risk population monitoring.