Atezolizumab plus bevacizumab versus active surveillance in patients with resected or ablated high-risk hepatocellular carcinoma (IMbrave050): a randomised, open-label, multicentre, phase 3 trial

Adults with a first diagnosis of HCC who had undergone surgical resection or ablation (radiofrequency or microwave only) with curative intent 4 to 12 weeks before randomisation and were at high risk of recurrence (exact definition of high-risk features in the full text of the paper); without major macrovascular invasion (except Vp1/Vp2) and without extrahepatic spread; Child-Pugh A and ECOG 0 or 1. Up to one cycle of adjuvant TACE after resection was allowed. Exclusion criteria (ClinicalTrials.gov): any HCC treatment before resection or ablation, clinically significant ascites, history of hepatic encephalopathy, bleeding from untreated or incompletely treated varices in the past 6 months, autoimmune disease, HBV and HCV (or HBV and HDV) co-infection, uncontrolled hypertension and use of aspirin or full-dose anticoagulants.

Primary endpoint: recurrence-free survival (RFS) assessed by an independent review facility in the intention-to-treat population.

Secondary endpoints: overall survival and safety.

Grade 3 or 4 adverse events occurred in 136 (41%) of 332 patients receiving atezolizumab + bevacizumab and 44 (13%) of 330 patients in the active surveillance group. Grade 5 adverse events occurred in 6 patients (2%; two treatment-related) in the treatment group and one patient (less than 1%) in the surveillance group. Both drugs were discontinued because of adverse events in 29 patients (9%).

IMbrave050 was the first positive phase 3 adjuvant trial in HCC, and the AASLD 2023 guideline, based on its initial results, had recommended adjuvant immune checkpoint inhibitor-based therapy in high-risk patients (Level 2, Strong). However, in the updated analysis (Yopp et al., J Hepatol 2026), the recurrence-free survival benefit was not sustained (HR 0.90; 95% CI 0.72–1.12), overall survival was immature (HR 1.26), and the authors concluded that the benefit–risk ratio does not support adjuvant use of this combination. There is currently no standard adjuvant systemic therapy for HCC, and this combination is not FDA-approved for this use.

Open-label design with a surveillance control group (no placebo); the initial result was based on an interim analysis with short follow-up and was not confirmed with longer follow-up; the updated RFS analysis is post hoc; the possibility of crossover of the control group to treatment after recurrence makes overall survival difficult to interpret.