Feed

Primary liver cancer is a major global cause of cancer death, and hepatocellular carcinoma (HCC) is the predominant histological subtype. This literature review synthesizes current evidence on the exposome, multi-omics landscape, and artificial intelligence (AI)-based integration strategies relevant to biomarker discovery in liver cancer, with a focus on biological rationale, emerging clinical applications, and translational limitations. Key etiologic drivers include viral hepatitis, alcohol-related liver disease, and metabolic dysfunction-associated steatotic liver disease, all of which interact with environmental exposures across the life course. Biomarker discovery increasingly relies on integrated assessment of exposure-related signals together with genomic, epigenomic, transcriptomic, proteomic, metabolomic, and spatially resolved data. Hepatocarcinogenesis involves a complex interplay of chronic liver injury, environmentally patterned molecular perturbation, and dynamic tumor-host interactions. We emphasize an exposome-informed, multimodal strategy in which interpretable AI models identify clinically relevant signatures for early detection, prognostic stratification, and treatment guidance. Critical limitations of current evidence include incomplete exposure assessment, heterogeneous data platforms, retrospective study design, limited external validation, and insufficient model transparency. Emerging approaches, including proteogenomic, lipidomic, single-cell, and digital pathology-based modeling, show promise but require further validation in etiologically diverse cohorts. The purpose of this review is to critically examine how AI can integrate exposome-related information with multi-omics data for biomarker discovery in liver cancer. Here, particular attention is given to the exposure-to-biomarker sequence, immune-metabolic remodeling, liquid-biopsy translation, and the reduction of high-dimensional signatures into clinically deployable assays.