ARTE Score Predicts Decompensation on Atezo-Bev : J Hepatol | May 2026

Introduction:

Introduction: Atezolizumab plus bevacizumab (Atezo-Bev) is the standard first-line systemic therapy for many patients with unresectable hepatocellular carcinoma (uHCC). However, treatment success depends not only on tumor control but also on preservation of liver function. Hepatic decompensation remains a major cause of treatment interruption, reduced survival, and impaired quality of life. Identifying patients at greatest risk before therapy initiation is therefore crucial for optimal treatment selection and monitoring.

Problem Statement:

Problem Statement: Existing prognostic models primarily focus on survival outcomes and do not specifically predict hepatic decompensation during Atezo-Bev therapy. Clinicians lack a simple, validated tool that can identify patients at increased risk of liver failure despite having Child-Pugh A liver function at treatment initiation.

Summary:

Summary: This multicenter study analyzed 453 Child-Pugh A patients with unresectable HCC receiving first-line atezolizumab plus bevacizumab from the ARTE database and externally validated findings in an independent cohort of 292 patients from the AB-real database.

During a median follow-up of 14 months, hepatic decompensation occurred in 16.3% of patients. Multivariable Cox regression identified three independent predictors of decompensation: neoplastic portal vein thrombosis, elevated bilirubin levels, and thrombocytopenia.

These variables were integrated into a simple point-based risk stratification model termed the ARTE Score.

Based on the cumulative score, patients were classified into three risk categories:

Low risk (0–1 points)

Intermediate risk (2 points)

High risk (3–4 points)

The model demonstrated clear separation of decompensation risk across all categories.

Compared with low-risk patients, intermediate-risk patients experienced nearly a twofold increase in decompensation risk, while high-risk patients had more than a fourfold increased risk.

Decompensation-free survival showed marked differences between groups. At 12 months, decompensation-free survival remained high in low-risk patients but progressively declined in intermediate- and high-risk groups. These differences persisted at 24 months, highlighting the score’s ability to identify clinically meaningful long-term risk.

Importantly, the model maintained its predictive performance in the independent external validation cohort, supporting its reproducibility and generalizability across different clinical settings.

The biological rationale behind the score is intuitive. Portal vein tumor thrombosis reflects advanced tumor burden and impaired hepatic perfusion. Elevated bilirubin indicates reduced hepatic reserve even within Child-Pugh A patients. Thrombocytopenia serves as a surrogate marker of clinically significant portal hypertension and advanced underlying cirrhosis. Together, these variables capture the combined impact of tumor aggressiveness and baseline liver vulnerability.

Clinically, the ARTE Score may assist in treatment planning before initiating Atezo-Bev. High-risk patients may benefit from closer laboratory surveillance, earlier recognition of liver deterioration, optimization of portal hypertension management, and consideration of alternative therapeutic strategies when appropriate.

The score is particularly valuable because it relies exclusively on routinely available clinical variables, making implementation feasible in everyday practice without specialized testing or biomarkers.

Overall, this study introduces the ARTE Score as a simple, externally validated tool for predicting hepatic decompensation in patients with unresectable HCC receiving atezolizumab plus bevacizumab, providing an important step toward more personalized risk assessment and treatment decision-making in advanced liver cancer.