Abstract
Three immunotherapy-based regimens are now approved for first-line treatment of advanced hepatocellular carcinoma: atezolizumab plus bevacizumab, durvalumab plus tremelimumab (STRIDE), and nivolumab plus ipilimumab. Although each has demonstrated an overall survival benefit in a pivotal phase III trial, no randomized study has directly compared these regimens and no validated clinical or molecular biomarker identifies which patients derive differential benefit from one strategy over another. Current selection therefore remains driven primarily by hepatic and functional reserve, regimen-specific safety constraints, and treatment feasibility. In this Current Opinion, we propose a practical selection framework, examine emerging evidence in patients with Child-Pugh B liver function, and emphasize that current sequencing evidence does not establish a preferred first-line immunotherapy combination. We further discuss why predictive biomarker development in advanced hepatocellular carcinoma has repeatedly failed to translate, including limited tissue availability, tumor and liver-disease heterogeneity, prognostic confounding, and reliance on single-treatment cohorts rather than treatment-by-biomarker interaction analyses. We propose that the most promising future selection tools will integrate tumor immune state with angiogenic, myeloid, and circulating molecular features and be prospectively tested across competing regimens and treatment sequences.