PI-067 - EXPOSURE-EFFICACY ANALYSIS OF CAMRELIZUMAB PLUS RIVOCERANIB IN PATIENTS WITH ADVANCED HEPATOCELLULAR CARCINOMA.

N. DJEBLI¹, X. Wei², S. Xie³, S. Boesmans³, B. Wang⁴, M. Zhou⁵, S. Li⁶, C. Yang⁶, K. Shen⁶, S. Jang²; ¹Luzsana / Hengrui, Luzsana Biotechnology (subsidiary of Jiangsu Hengrui Pharmaceutical), Basel, Switzerland, ²Elevar Therapeutics, Fort Lee, NJ, USA, Elevar Therapeutics, Fort Lee, NJ, USA, ³Amador Bioscience Inc, Amador Bioscience Inc, Pleasanton, CA, USA, ⁴Amador Bioscience Inc, Pleasanton, Amador Bioscience Inc, Pleasanton, CA, USA, ⁵Luzsana Biotechnology (Hengrui Pharmaceuticals), 1Luzsana Biotechnology Princeton (subsidiary of Jiangsu Hengrui Pharmaceutical), NJ, USA, ⁶Jiangsu Hengrui Pharmaceuticals, Shanghai, China, Jiangsu Hengrui Pharmaceuticals, Shanghai, China.

Background: Camrelizumab (cam) is a monoclonal antibody that specifically binds to programmed cell death receptor 1 (PD-1). Rivoceranib (rivo) is a vascular endothelial growth factor receptor tyrosine kinase inhibitor. The combination of camrelizumab 200 mg once every 2 weeks (Q2W) and rivoceranib 250 mg once daily (QD) has shown favorable clinical efficacy for the treatment of patients with advanced hepatocellular carcinoma (HCC) in a pivotal Phase 3 study (CARES-310). The purpose of this analysis was to characterize exposure-response (ER) relationships of cam and rivo exposures with efficacy endpoints including overall survival (OS), progress free survival (PFS), and objective response rate (ORR) in patients with advanced HCC.

Methods: The ER population included 272 patients with HCC from CARES-310. Exposure metrics (AUC, Cavg, Cmax, and Cmin) were generated using population pharmacokinetic models for cam and rivo. Kaplan-Meier (KM) analysis (OS and PFS) or boxplots (ORR) were conducted to explore ER potential relationships. ER models were developed using either logistic regression (LR) or time-to-event (TTE) as appropriate. Covariate analysis was performed in the base model that was stable and provided reasonably precise parameter estimates.

Results: Parametric TTE models or LR analysis did not identify a significant relationship of OS, PFS and ORR with cam or rivo exposures. Exploratory KM analysis and covariate analysis identified no significant effects of baseline body weight, race (Asian vs. Caucasian) and geographic region (Asia vs. non-Asia), or viral etiology (HBV+ vs. HCV+ vs. non-viral) on OS and PFS. Univariate analysis of drug exposure and covariates revealed that the estimated tumor growth rate (KG) from a previously developed tumor growth inhibition model was the best predictor of OS. Compared to the survival rate in patients with median KG (0.00143 day-1), OS was predicted to be 31.1% lower after 1 year in patients with faster KG (95th percentile of KG, 0.0102 day-1) (Figure 1).

Conclusion: The ER analysis demonstrated that there was no significant association between efficacy (OS, PFS, and ORR) and cam or rivo exposure in the 272 patients with advanced HCC from Study CARES-310, indicating an achievement of efficacy plateau at dose regimen of cam 200 mg Q2W and rivo 250 mg QD. The estimated KG was identified as the best predictor of OS.