Diagnostic Performance of cfDNA-Based Liquid Biopsy for Hepatocellular Carcinoma: A Systematic Review and Meta-Analysis.
Liu Jiwei J, Zhang Yangyu Y, Yang Shuo S, Wu Yanhua Y et al.
Conventional surveillance tools for hepatocellular carcinoma (HCC) have limited sensitivity. Circulating cell-free DNA (cfDNA)-based liquid biopsy is a promising minimally invasive approach; however, its overall diagnostic performance remains unclear. This meta-analysis aimed to comprehensively evaluate the diagnostic accuracy of cfDNA-based liquid biopsy in HCC. Following PRISMA-DTA guidelines, we searched PubMed, EMBASE, and Web of Science until July 2026. Diagnostic accuracy data were pooled using bivariate random-effects models. Methodological quality was assessed using QUADAS-2 and NOS. A total of 29 studies (4250 HCC patients, 7982 controls) were included. Pooled sensitivity and specificity of cfDNA-based testing were 0.82 (95% CI: 0.76-0.86) and 0.91 (95% CI: 0.88-0.93). The area under the summary receiver operating characteristic curve (AUC) was 0.93 (95% CI: 0.91-0.96). Subgroup analyses showed higher sensitivity in advanced HCC 0.91 (95% CI: 0.85-0.95) than early HCC 0.77 (95% CI: 0.71-0.83). Methylation profiling yielded relatively higher sensitivity 0.84 (95% CI: 0.77-0.89), whereas fragmentomics showed relatively favorable specificity 0.93 (95% CI: 0.88-0.96). Combining cfDNA-based testing with alpha-fetoprotein (AFP) significantly improved sensitivity compared with cfDNA alone (0.78 vs. 0.65). No significant publication bias was detected (p = 0.83). cfDNA-based liquid biopsy achieves high diagnostic accuracy for HCC. As exploratory observations, methylation-based assays with relatively higher sensitivity may have potential value for screening high-risk populations, while fragmentomics-based assays with relatively favorable specificity may aid differential diagnosis. Combining cfDNA-based liquid biopsy with AFP further improves diagnostic performance, supporting cfDNA as a promising noninvasive tool for HCC detection. Trial Registration: This systematic review was registered on PROSPERO: CRD420261293877 (https://www.crd.york.ac.uk/PROSPERO/).