Background: There is no standard of care for adjuvant therapy for patients with hepatocellular carcinoma. This trial was designed to assess the efficacy and safety of sorafenib versus placebo as adjuvant therapy in patients with hepatocellular carcinoma after surgical resection or local ablation.

Methods: We undertook this phase 3, double-blind, placebo-controlled study of patients with hepatocellular carcinoma with a complete radiological response after surgical resection (n=900) or local ablation (n=214) in 202 sites (hospitals and research centres) in 28 countries. Patients were randomly assigned (1:1) to receive 400 mg oral sorafenib or placebo twice a day, for a maximum of 4 years, according to a block randomisation scheme (block size of four) using an interactive voice-response system. Patients were stratified by curative treatment, geography, Child-Pugh status, and recurrence risk. The primary outcome was recurrence-free survival assessed after database cut-off on Nov 29, 2013. We analysed efficacy in the intention-to-treat population and safety in randomly assigned patients receiving at least one study dose. The final analysis is reported. This study is registered with ClinicalTrials.gov, number NCT00692770.

Findings: We screened 1602 patients between Aug 15, 2008, and Nov 17, 2010, and randomly assigned 1114 patients. Of 556 patients in the sorafenib group, 553 (>99%) received the study treatment and 471 (85%) terminated treatment. Of 558 patients in the placebo group, 554 (99%) received the study treatment and 447 (80%) terminated treatment. Median duration of treatment and mean daily dose were 12·5 months (IQR 2·6-35·8) and 577 mg per day (SD 212·8) for sorafenib, compared with 22·2 months (8·1-38·8) and 778·0 mg per day (79·8) for placebo. Dose modification was reported for 497 (89%) of 559 patients in the sorafenib group and 206 (38%) of 548 patients in the placebo group. At final analysis, 464 recurrence-free survival events had occurred (270 in the placebo group and 194 in the sorafenib group). Median follow-up for recurrence-free survival was 8·5 months (IQR 2·9-19·5) in the sorafenib group and 8·4 months (2·9-19·8) in the placebo group. We noted no difference in median recurrence-free survival between the two groups (33·3 months in the sorafenib group vs 33·7 months in the placebo group; hazard ratio [HR] 0·940; 95% CI 0·780-1·134; one-sided p=0·26). The most common grade 3 or 4 adverse events were hand-foot skin reaction (154 [28%] of 559 patients in the sorafenib group vs four [<1%] of 548 patients in the placebo group) and diarrhoea (36 [6%] vs five [<1%] in the placebo group). Sorafenib-related serious adverse events included hand-foot skin reaction (ten [2%]), abnormal hepatic function (four [<1%]), and fatigue (three [<1%]). There were four (<1%) drug-related deaths in the sorafenib group and two (<1%) in the placebo group.

Interpretation: Our data indicate that sorafenib is not an effective intervention in the adjuvant setting for hepatocellular carcinoma following resection or ablation.