Sonya C. Tate(1), Amanda K. Sykes(1), Palaniappan Kulanthaivel(2), Edward M. Chan(2), Donald E. Thornton(2), P. Kellie Turner(2), Damien M. Cronier(1)
(1) Eli Lilly and Company, Erl Wood, UK, (2) Eli Lilly and Company, Indianapolis, USA
Objectives: Describe the population pharmacokinetics (PK) of abemaciclib in cancer patients enrolled on a Phase I clinical trial.
Methods: PK data were collected in a multicentre, nonrandomized, open-label Phase 1 clinical trial of abemaciclib in the treatment of approximately 300 patients with advanced cancer. The study design consisted of a dose escalation phase (50, 100, 150 or 225 mg Q24H, or 75, 100, 150, 200 and 275 mg Q12H) and 6 tumour-specific expansion phases (150 or 200 mg Q12H). PK sampling was intensive, with samples collected for 72 hours after an initial single dose, and then on days 15, 22 and 28 of the continuous dosing regimen. The population PK parameters, inter-individual (IIV) and inter-occasion variability (IOV) were estimated using NONMEM 7.2 (FOCEi). Age, weight, gender, liver and renal function were tested via step-wise covariate modelling (forwards inclusion [ΔOFV > 3.84]; backwards exclusion [ΔOFV < 10.8 and ΔIIV < 5%]). Model performance was evaluated by objective function mapping, visual predictive checks and bootstrap analysis.
Results: The popPK analysis dataset consisted of 224 patients with 4012 observations. The analysis population had a mean age of 60 (24 – 85) and mean body weight of 73.9kg (43.6 – 175); 67% were female and the majority were white (94%). The data were best described by a one compartment model with linear absorption (ka) of 0.197/h (IIV=77.6%), clearance (CL/F) of 35.9L/h (IIV=29.6%, IOV=52.1%) and volume of distribution (V/F) of 1050L; residual error was described with a combined model. Through model development, CL/F and V/F were found to be strongly correlated. Given the complex absorption observed in pre-clinical studies [1], this correlation was attributed to IIV in bioavailability. A relative bioavailability term (Frel) was therefore implemented by fixing the mean population estimate to 1 and allowing IIV to be estimated. Frel was also found to be time- and dose-dependent (initial Frel IIV=73.4%, steady state Frel IIV=123%) with 50% saturation occurring at 101 mg. Liver function markers (serum albumin and alkaline phosphatase) were found to significantly affect Frel. However, there remained a high level of variability in exposure which could not be attributed to patient characteristics or laboratory parameters.
Conclusion: Explained variability [2] in abemaciclib exposure was far lower than unexplained variability, suggesting the impact of liver function on exposure is clinically negligible.
References
[1] Tate, S.C., et al., Clin Cancer Res, 2014. 20(14): 3763-74.
[2] Hennig, S. and M.O. Karlsson, J Pharmacokinet Pharmacodyn, 2014. 41(2): 109-25.
Reference: PAGE 24 (2015) Abstr 3648 [www.page-meeting.org/?abstract=3648]
Poster: Methodology - Covariate/Variability Models