Abstract
Background: In a phase II study in uHCC (Study 22, NCT02519348), a novel, priming combination regimen of tremelimumab (T; anti-CTLA-4) and durvalumab (D; anti-PD-L1) (T300+D) has shown favorable clinical activity vs each agent as monotherapy or vs another combination (T75+D). The analyses presented here assess the pharmacokinetics (PK) and relationships between tremelimumab exposure, as monotherapy or in combination, and safety, efficacy, and pharmacodynamics (PD) in this study. Methods: Overall, 216 pts were included in these analyses (T, n=65; T300+D, n=72; T75+D, n=79). Safety, antitumor activity, PK, PD, and immunogenicity were analyzed using standard pharmacometrics methods. A previously developed population PK model for T across solid tumors was validated using T monotherapy and combination therapy data from Study 22, including a post-hoc covariate analysis to assess the impact of covariates on individual PK parameters. Population PK and PD models related individual T exposures to safety parameters, PD, and efficacy (overall survival, OS; progression-free survival, PFS; and objective response rate, ORR). The E-R relationships for time-to-event variables OS and PFS were explored by Kaplan-Meier estimates and analyzed by Cox proportional-hazards models (CPHM). Results: For T monotherapy and T+D combinations, no significant E-S relationships were observed for grade 3/4 treatment-related adverse events (TRAEs), grade 3/4 TRAEs of special interest, and AEs leading to treatment discontinuation. Analyses of each quartile of T exposure suggest pts with higher exposure (3rd and 4th quartile) may have longer OS vs lower quartiles. The CPHManalysis showed that after accounting for prognostic factors (baseline albumin and neutrophil-to-lymphocyte ratio), neither AUC nor Cmin appeared to be a significant factor for OS hazard. There was no significant relationship between response and ORR, PFS and any T PK exposure metric in T-treated pts. Saturable relationships (described by Emax) were observed for maximum change from baseline for proliferating T-cell counts as functions of exposure. Conclusions: The observed PK data are generally consistent with predictions based on a historical population PK model, suggesting the PK of T in uHCC pts is consistent with pts with other solid tumors. No significant relationships were observed between E-S and E-R; therefore, PK is not a significant predictor to evaluate for T efficacy or safety. Considering the small sample size limitation, still the saturable relationships observed in proliferating T-cells appear to support a dose of T300. Future studies of pooled data from Study 22 and the larger phase III HIMALAYA study (NCT03298451) will be conducted to further the characterization of E-R relationships and the development of the T300+D regimen.