Final results of adjuvant nivolumab for hepatocellular carcinoma (HCC) after surgical resection (SR) or radiofrequency ablation (RFA) (NIVOLVE): A phase 2 prospective multicenter single-arm trial and exploratory biomarker analysis.

416 Background: The NIVOLVE trial was designed to assess the efficacy and safety of nivolumab as an adjuvant therapy for HCC, and to identify biomarkers predictive of recurrence in patients after SR or RFA (Registration # UMIN 000026648). Methods: The trial involved 11 sites and was conducted in patients with HCC who showed a complete response after SR (n = 33) or RFA (n = 22) (ITT). Overall, 53 of 55 patients with Child-Pugh A received nivolumab (240 mg/body every 2 weeks (8 cycles)), followed by nivolumab (480 mg/body every 4 weeks [8 cycles]) within 6 weeks after SR or RFA. The primary endpoint was the 1-year recurrence-free survival rate (RFSR). The key secondary endpoint was RFS. Exploratory biomarker analysis included mutations, copy number alterations, and positivity of immune cells. Techniques included next generation sequencing, immunohistochemical staining (IHC) of formalin-fixed paraffin-embedded tissues (n = 31 [13 with recurrence and 18 without]) for CD8, PD-1, PD-L1, Foxp3, and β-catenin, and ctDNA analysis using pre-nivolumab whole blood by deep sequencing (CAPP-seq; Avenio). Results: The 1-year RFSR and median RFS were 78.6% and 26.3 months (95% confidence interval (CI), 16.8-NR), respectively, with no difference between SR and RFA. Copy number gains (CNGs) in WNT/β-catenin related genes ( APC, CTNNB1, TCF7L1, TCF7L2) (n = 30) correlated with shorter RFS (positive: 10.6 months vs. negative: not reached [NR]; p < 0.0001). IHC revealed that negative staining for PD-1 (p < 0.0001), a low combined positive score for PD-L1 (p = 0.005), a low number of CD8+ tumor infiltrating lymphocytes (TILs) (p < 0.001), and positivity for Foxp3+ cells (p = 0.015) correlated with recurrence. HCC cases with low numbers of CD8+ TILs or cases positive for Foxp3+ cells (n = 16) showed a significantly shorter RFS (16.8 months [95% CI, 8.7-25.2]) than those with high numbers of CD8+ TILs and those positive for PD-1/PD-L1 expression (n = 15) (NR [95% CI,26.2months–NR]) (p < 0.0001). HCC cases with activation of the WNT/β-catenin pathway assessed by IHC (n = 9) showed shorter RFS (17.0 months [95% CI,1.1–26.2]) than those without activation (n = 22) (NR [95% CI,23.0 months–NR]) (p = 0.014). Patients positive for ctDNA (n = 15) before nivolumab tended to have shorter RFS than those without ctDNA (n = 9) (26.3 vs. NR). There was no correlation between TMB and RFS. Treatment-emergent adverse events (AEs) (n = 53) were as follows: all grades, 93%; grades 3–4 (18.9%); and immune related AEs, 25%. Conclusions: The 1-year RFSR and RFS in the NIVOLVE trial were 78.6% and 26.3 months, respectively. No new safety signal was observed. CNG in WNT/β-catenin-related genes, activation of the WNT/β-catenin pathway, the presence of Foxp3+ cells, and low numbers of CD8+ TILs may predict recurrence after SR or RFA with adjuvant nivolumab. Clinical trial information: UMIN 000026648.