Abstract
BACKGROUND & AIMS: Hepatocellular carcinoma (HCC) is associated with high cancer-specific mortality. While immune-checkpoint inhibitors (ICIs) improved overall survival (OS) compared to tyrosine kinase inhibitors, biomarkers predicting response to ICI in HCC are lacking. This study investigates the prognostic value of serum-based prognostic scores in patients with HCC receiving atezolizumab and bevacizumab. METHODS: This post-hoc study analysis evaluates data from the phase 3 IMbrave150 trial, comparing atezolizumab plus bevacizumab to sorafenib in patients with unresectable HCC. 212 patients were included in the analysis. The prognostic value of imaging was compared to albumin, C-reactive protein (CRP) and interleukin-6 (IL-6), as well as composite scores, including the modified Glasgow Prognostic Score (mGPS) after three cycles of therapy. For further analysis, patients were classified in three risk groups according to each scoring system (low-risk, intermediate-risk and high-risk). RESULTS: The on-treatment mGPS, assessed 9 weeks post-treatment initiation, predicted OS with hazard ratios of 2.31 (95% CI 1.39-3.83, p < 0.001) for intermediate-risk and 3.40 (95% CI 3.07-5.59, p < 0.001) for high-risk, compared to low-risk groups, showing greater accuracy than RECIST imaging. Albumin, CRP, and IL-6 were individually good prognostic indicators, with albumin/CRP (ACR) and albumin/IL-6 (AIR) ratios having the highest prognostic power (c-index: ACR 0.66 (95% CI 0.61-0.71), AIR 0.67 (0.62-0.72), mGPS 0.62 (0.57-0.66)). Multivariable analysis confirmed serum-based scores' prognostic value independent of imaging. Serum-based scores significantly correlated with survival in patients with stable disease (SD, 79% of patients) or progressive disease (PD, 12% of patients). CONCLUSIONS: The on-treatment mGPS, as well as ACR and AIR, predicted outcomes in patients with HCC independent of and more accurately than radiological staging. Since, the mGPS is the most cost effective and widely validated score, we consider it best suited for clinical use. Prospective validation is needed to confirm these findings.