From Theory to Practice: Which Biomarkers Are Ready for Predicting Response in Advanced HCC?

Systemic therapies for advanced hepatocellular carcinoma (HCC) have expanded considerably with the advent of tyrosine kinase inhibitors, immune checkpoint inhibitors and immunotherapy-anti-angiogenic combinations. However, despite this therapeutic diversification, first-line treatment selection remains largely empirical, as few biomarkers are available at diagnosis to inform therapeutic choice. This review provides a practice-oriented synthesis of predictive biomarkers evaluated in advanced HCC, distinguishing those that are clinically transposable from those that remain exploratory. Among currently available tools, only a limited set of biomarkers demonstrates sufficient robustness and feasibility for real-world use, and most provide information after treatment initiation rather than guiding upfront selection. Liver function assessment (ALBI score), dynamic Alpha-fetoprotein (AFP) kinetics and anti-drug antibodies (ADA) emerge as the most actionable parameters. ALBI retains predominantly prognostic value, whereas early AFP decline and ADA formation offer treatment-specific, on-treatment insights that may support clinical decision-making. Imaging biomarkers such as mRECIST remain essential for early efficacy assessment but lack predictive value at baseline. Routine histological markers-including PD-L1, mismatch repair proteins and β-catenin immunostaining-do not reliably predict response prior to therapy initiation. In contrast, a broad range of circulating, tissue-based and imaging-derived biomarkers-including liquid biopsy analytes (cytokines, CTCs, ctDNA, miRNAs), transcriptomic immune signatures, artificial intelligence AI-assisted histopathology, radiomics and multi-omic tumour profiling-provide substantial mechanistic insight but remain investigational. Their limited clinical applicability reflects methodological heterogeneity, insufficient standardisation and the absence of prospective validation in biomarker-driven trials. Among emerging approaches, proteomics stands out as a particularly promising strategy. By directly capturing the protein expression landscape of the tumour and its microenvironment, proteomics may overcome the inferential limitations of genomic and transcriptomic biomarkers and contribute to the development of biologically grounded, pre-treatment stratification tools in advanced HCC. Ultimately, this review underscores a critical unmet need: the development of integrated, multi-omic predictive strategies that combine baseline tissue characteristics with dynamic liquid biopsy markers and advanced imaging. Such approaches, strengthened by AI and prospective biomarker-driven trials, are essential to move beyond empirical therapy selection and toward true precision medicine in advanced HCC.