Paul Westwood (1), Lisa O’Brien (2), Ling Gao (2), Michael Heathman (2)
(1) Eli Lilly, Windlesham, UK; (2) Eli lilly, Indianapolis, US
Objectives: Ramucirumab (RAM) is a monoclonal antibody (mAb) that binds vascular endothelial growth factor receptor 2. The aims of this analysis were to characterize the pharmacokinetics (PK) of RAM in cancer patients, characterize the inter-patient variability (IIV), and investigate patient factors that may influence RAM disposition.
Methods: The dataset consisted of 497 patients with 2782 observations from eight studies (Phase 1/1b, 2 and 3). A range of disease states were represented: gastric cancer (80.5%), non-small cell lung cancer (8.2%), colorectal cancer (1.6%), breast cancer (2.2%) and other tumour types (7.4%). RAM was administered as an IV infusion over approximately 1 hour, at either 8 mg/kg every 2 weeks or 10 mg/kg every 3 weeks. PK sampling varied between studies.
A pharmacostatistical model was first developed using the criteria of model plausibility, minimum objective function value (MOF), agreement between predicted and observed, magnitude and randomness of residuals, and precision of parameter estimates. Demographics, cancer indication, dose, renal function, hepatic status, and other laboratory values were assessed as covariates using forward selection and backward elimination. Due to the nature of other mAbs, time and exposure dependency were also investigated. Covariates were considered significant if they decreased IIV in the relevant parameter by ≥5% and reduced the MOF by ≥6.635 (p<0.01) in forward selection and ≥10.828 (p<0.001) in backward elimination. The model was evaluated using objective function mapping and visual predictive check. All analyses were performed using NONMEM (Version 7.2).
Results: The PK of RAM were well characterized by a two-compartment model with IIV estimated on all parameters and covariance between CL and V1. A combined additive/proportional error model was used. No covariates were found to satisfy the predefined criteria, and no relationship was found between dose level or time and RAM PK. Mean (%CV) PK parameters derived from post-hoc parameter estimates were: CL 0.0140 L/hr (29.8%), Vss 5.5L (14.4%), and terminal half-life 15 days (24.1%).
Conclusions: The final model adequately described the time-concentration profile of RAM in patients with a range of cancer indications. The PK parameters were consistent with those obtained from non-compartmental analyses of Phase 1 and 2 studies, with the exception of half-life. Simulation using the model confirmed the appropriateness of weight-based dosing.
Reference: PAGE 24 (2015) Abstr 3350 [www.page-meeting.org/?abstract=3350]
Poster: Drug/Disease modeling - Oncology