Perioperative nivolumab ± relatlimab (PD-1/LAG-3) is surgery-safe and drives ~20%–27% major path responses in high-risk resectable HCC

Nivolumab is an anti–PD-1 monoclonal antibody, and relatlimab is an anti–LAG-3 monoclonal antibody (dual checkpoint blockade); both were tested as perioperative immunotherapy in an open-label, randomized, noncomparative Phase II trial in potentially resectable, high-risk HCC (NCT04658147), using 2 neoadjuvant doses over 8 weeks followed by resection and up to 10 months of post-op therapy (n=30; 15/arm). Feasibility was strong: 0 patients had treatment-related AEs that prevented surgery; grade 3+ TRAEs occurred in 26.7% in each arm, with no grade 5 events; R0 resection rates were 86.7% with nivolumab alone vs 73.3% with nivo+relatlimab (one aborted surgery due to advanced cirrhosis). Efficacy signals were modest by RECIST (ORR 13.3% Arm A; 20.0% Arm B) but richer pathologically: complete/major pathologic (or pathologic/radiologic) response was 20.0% vs 26.7%, with 2-year DFS 73.3% vs 88.9% and 2-year OS 68.5% vs 84.8%—directionally encouraging vs historical “surgery/ablation → surveillance” where recurrence is common, and relative to the adjuvant IMbrave050 benchmark where atezolizumab+bevacizumab improved RFS vs surveillance (HR 0.72) but did not eliminate high relapse risk. Peripheral pharmacodynamics showed near-100% LAG-3 receptor occupancy on CD8+ T-cells at C2D1, but only ~25% LAG-3 occupancy in the tumor microenvironment at C2D15, hinting at potential intratumoral target-engagement limits.