Hepatocellular carcinoma (HCC) is one of the most prevalent and lethal causes of cancer-related death worldwide. The treatment of HCC remains challenging and is largely predicated on early diagnosis. Surveillance of high-risk groups using abdominal ultrasonography, with or without serum analysis of α-fetoprotein (AFP), can permit detection of early, potentially curable tumours, but is limited by its insensitivity. Reviewed here are two current approaches that aim to address this limitation. The first is to use old re-emerged empirically derived biomarkers such as AFP, now applied within statistical models. The second is to use circulating nucleic acid biomarkers, which include cell-free DNA (for example, circulating tumour DNA, cell-free mitochondrial DNA and cell-free viral DNA) and cell-free RNA, applying modern molecular biology-based technologies and machine learning techniques closely allied to the underlying biology of cancer. Taken together, these approaches are likely to be complementary. Both hold considerable promise for achieving earlier diagnosis as well as offering additional functionalities including improved monitoring of therapy and prediction of response thereto.

肝细胞癌 (HCC) 是全球癌症相关死亡最普遍和最致命的原因之一。HCC 的治疗仍然具有挑战性，主要依赖于早期诊断。使用腹部超声检查对高危人群进行监测，无论是否进行甲胎蛋白 (AFP) 血清分析，都可以发现早期可能治愈的肿瘤，但受限于其不敏感性。此处回顾了旨在解决此限制的两种当前方法。第一个是使用旧的重新出现的经验衍生的生物标志物，例如 AFP，现在应用于统计模型中。二是使用循环核酸生物标志物，包括游离DNA（例如循环肿瘤DNA、游离线粒体DNA和游离病毒DNA）和游离RNA，应用与癌症的潜在生物学密切相关的现代分子生物学技术和机器学习技术。总之，这些方法可能是互补的。两者都对实现早期诊断以及提供额外的功能（包括改进的治疗监测和对其反应的预测）具有相当大的希望。