Baseline Predictors, Prognostic Impact, and Risk Stratification of Early Progressive Disease on Atezolizumab-Bevacizumab in Unresectable Hepatocellular Carcinoma: A Multicenter Real-World Study

Introduction: Early progressive disease (early PD) at the first imaging assessment during atezolizumab plus bevacizumab (Atezo/Bev) therapy may represent primary resistance and is associated with poor outcomes in unresectable hepatocellular carcinoma (uHCC). This study aimed to quantify the frequency, prognostic impact, and baseline predictors of early PD in a large real-world cohort. Methods: Consecutive patients with uHCC who initiated Atezo/Bev therapy at 28 institutions in Japan (September 2020–August 2025) were retrospectively analyzed. Early PD was defined as PD at the first on-treatment imaging assessment, according to the Response Evaluation Criteria in Solid Tumors (RECIST) version 1.1. Overall survival (OS) was compared using Kaplan-Meier analysis. Baseline predictors were examined using prespecified multivariable logistic regression with multiple imputation: the primary multivariable model included Barcelona Clinic Liver Cancer (BCLC) stage C and alpha-fetoprotein (AFP) ≥400 ng/mL, while an alternative tumor-feature model including extrahepatic spread (EHS) and macrovascular invasion (MVI) was evaluated as a sensitivity analysis. Results: Among 685 treated patients, 666 were evaluable at the first imaging assessment, of whom 148 (22.2%) had early PD. The median OS was 333 days in the early-PD group versus 768 days in the non-early-PD group (log-rank p < 0.001). In the primary multivariable model, AFP ≥ 400 ng/mL (OR 2.33, 95% confidence interval [CI]: 1.58–3.43) and BCLC-C (OR 1.76, 95% CI: 1.15–2.68) independently predicted early PD. In the alternative tumor-feature model, AFP ≥ 400 ng/mL (OR 2.50, 95% CI: 1.69–3.69) and EHS (OR 1.52, 95% CI: 1.03–2.23) remained significant predictors, whereas MVI did not. Two-point baseline scores (AFP ≥ 400 plus BCLC-C or EHS) stratified the risk of early PD in a stepwise manner, reaching approximately 40% in the highest-risk group (p for trend < 0.001). Conclusion: Early PD occurs in approximately one-fifth of patients with uHCC treated with Atezo/Bev and is associated with markedly poorer OS. AFP ≥ 400 ng/mL and BCLC-C provide a pragmatic baseline risk stratification approach that may help guide early monitoring and treatment sequencing.