Biomarker-Based Responder Selection and Early Prediction of Treatment Response in Hepatocellular Carcinoma: Dynamic Changes in Alpha-Fetoprotein and Des-Gamma-Carboxy Prothrombin During Atezolizumab Plus Bevacizumab Therapy.

Immune checkpoint inhibitor (ICI)-based combinations are the standard first-line therapy for unresectable hepatocellular carcinoma (HCC). A major challenge is the early identification of patients with primary progression (1st-PD) and those who experience early progression despite initial disease control (2nd-PD). This study evaluated whether very early treatment changes in alpha-fetoprotein (AFP) and des-gamma-carboxy prothrombin (DCP) could serve as predictors of treatment response during atezolizumab plus bevacizumab (Atz + Bev) therapy. A total of 147 patients treated with Atz + Bev were retrospectively analyzed. Serum tumor markers were measured approximately every 3 weeks, and radiologic responses were assessed using the Response Evaluation Criteria in Solid Tumors version 1.1 at week 6 (first evaluation) and again at a median of 14.8 weeks (second evaluation). At the first evaluation, 32 patients achieved a partial response, 81 showed stable disease, and 25 had progression. In the week 3 landmark analysis, early increases in AFP (ratio ≥ 1.4) or DCP (ratio ≥ 1.0) identified patients who would experience primary radiologic progression, with a clear separation in landmark progression-free survival (PFS) (3.4 vs. 13.1 months; < 0.001). Among the 109 patients with disease control at week 6, 92 maintained control and 17 progressed at the second evaluation. In the week 9 landmark cohort, modest rises in AFP (ratio ≥ 1.1) or DCP (ratio ≥ 1.5) identified individuals at risk for early secondary progression, again showing marked differences in landmark PFS (3.8 vs. 14.0 months; < 0.001). The dynamic monitoring of AFP and DCP provides a simple framework for biomarker-based responder selection and adaptive treatment optimization during Atz + Bev therapy. Clinically actionable thresholds at weeks 3 and 9 may support timely treatment switching and the integration of locoregional strategies, enabling personalized, biomarker-guided management to improve outcomes in unresectable HCC.