Liver resection after atezolizumab and bevacizumab versus maintenance therapy for locally advanced hepatocellular carcinoma (TALENTOP): a multicentre, open-label, randomised, phase 3 trial.

Resection might offer additional benefit in patients with advanced-stage hepatocellular carcinoma treated with systemic therapy. However, high-quality evidence supporting the procedure is absent. We aimed to establish whether resection can provide survival benefit in patients with advanced hepatocellular carcinoma who respond to systemic therapy. In this randomised, open-label, multicentre, phase 3 trial, we recruited treatment-naive patients with hepatocellular carcinoma, macrovascular invasion, and no extrahepatic metastasis from 24 hospitals in China. These patients were treated in the induction phase with three cycles of intravenous atezolizumab (1200 mg every 3 weeks) plus intravenous bevacizumab (15 mg/kg bodyweight every 3 weeks) and one cycle of atezolizumab monotherapy. Patients who completed the induction phase, had a partial response or stable disease according to Response Evaluation Criteria in Solid Tumors (RECIST) version 1.1, and were considered feasible for resection were randomly assigned (1:1) to undergo surgical resection followed by 12 months of atezolizumab plus bevacizumab initiated 4-6 weeks after surgery (ie, the surgery group), or to maintenance atezolizumab plus bevacizumab until loss of clinical benefit or intolerable toxicity (the maintenance therapy group). Randomisation was by computer-generated sequence with permuted blocks via a central interactive web response system, stratified by tumour response and Eastern Cooperative Oncology Group performance status. Treatment administered to patients was not masked. The primary endpoint was time to treatment failure, assessed by an independent review facility and analysed by intention to treat. Time to treatment failure was defined as the time from randomisation to the first documented treatment failure (ie, local recurrence or disease progression according to RECIST 1.1, emergence of extrahepatic spread, or death). This study is registered with ClinicalTrials.gov (NCT04649489) and is ongoing. Between April 4, 2021, and July 18, 2024, a total of 489 patients were enrolled in the induction phase. Of them, 201 were randomly assigned to the surgery group (n=101; 93 male, eight female) or the maintenance therapy group (n=100; 87 male, 13 female). After a median follow-up of 18·4 months, the median time to treatment failure was 20·4 months in the surgery group and 11·8 months in the maintenance therapy group (hazard ratio 0·60, 95% CI 0·39-0·91; p=0·015). Grade 3 or 4 treatment-related adverse events occurred in 32 (39%) of 83 patients in the surgery group and 21 (21%) of 100 in the maintenance therapy group, the most common of which were increased alanine aminotransferase (seven patients [8%] in the surgery group vs one [1%] in the maintenance therapy group), reduced platelet count (seven [8%] vs three [3%]), and proteinuria (three [4%] vs seven [7%]). Two treatment-related deaths occurred in the surgery group due to abnormal liver function (considered related to atezolizumab) and liver failure (considered related to atezolizumab, bevacizumab, or surgery). In patients with advanced hepatocellular carcinoma with macrovascular invasion after systemic therapy, time to treatment failure was longer in those who had liver resection than in those who received maintenance therapy. Shanghai Roche Pharmaceuticals and Ministry of Science and Technology of China.