Prediction of response to atezolizumab/bevacizumab in advanced hepatocellular carcinoma through radiomic features in pretreatment MRI

The advent of Atezolizumab/Bevacizumab as the first-line systemic therapy for advanced HCC marks a significant advancement. However, the need for biomarkers to prognosticate drug response persists. Radiomics, a method involving the conversion of medical images into high-dimensional quantitative features, holds promise for yielding novel biomarkers. This study aims to identify new biomarkers, derived not only from clinical data but also from pre-treatment contrast-enhanced MRI via radiomics feature extraction, aiming to forecast therapeutic response.

A retrospective analysis was conducted on patients with advanced HCC treated with Atezolizumab/Bevacizumab across six centers in Germany and Austria. Response assessment mandated an initial MRI and a subsequent radiological evaluation (CT or MRI) after three months of treatment. The cohort was stratified into two groups based on mRECIST criteria: controlled disease (encompassing complete response, partial response, or stable disease) and disease progression. The liver and all tumoral lesions were segmented in the initial MRI, from which radiomics features were extracted. The relevant radiomics features were then employed to predict treatment response and overall survival.

Of the 104 patients, 70 exhibited controlled disease while 34 faced disease progression. Factors such as hepatitis B virus etiology, ascitis presence, prior systemic therapy, elevated C-reactive protein levels, and metastatic disease correlated with progression. The progressive disease group demonstrated significantly diminished progression-free survival (81.5 days vs 298.5 days, p=0.001) and overall survival (150.0 days vs 568.5 days, p=0.001). By combining clinical parameters with relevant radiomics features, a predictive model for disease progression was developed.

Publication History

Article published online:
23 January 2024