Barry Weatherley (1), Lynn McFadyen (1), Daniela Conrado (2), Brinda Tammara (3)
(1) Pharmacometrics, Pfizer, Sandwich, UK (2) Clinical Pharmacology, Pfizer, Groton, USA (3) Clinical Pharmacology, Pfizer, Collegeville, USA
Objectives: Develop a PK model for PF-00251802 (parent) and its active metabolite PF-04015475 after oral administration of the prodrug PF-0417132.
Methods: PK data were obtained from a phase 2, randomized, double-blind efficacy and safety study of PF-04171327 (1, 5, 10, 15 mg dose daily) compared to 5 and 10 mg of prednisone and placebo over an 8 week period with a 4 week tapering period thereafter. PK trough samples were obtained at weeks 2, 4, 6, 10, 12 and at week 8 at 0, 1, 2, 3, 4 hours after dose.
179 patients had 1607 parent and metabolite concentrations (including samples below the quantitation limit (BLQ) and taper concentrations). 43 subjects were males (24.0%) and 136 were females (76.0%).
NONMEM 7.2 was used for a simultaneous fit to all parent and metabolite data in ordinary space. Inter-individual variability (IIV) (CL, V2 and Ka for parent; CLm and metabolite central volume V3 for metabolite) were modeled exponentially. Visual predictive checks were performed for parent and metabolite with and without taper and BLQ values to select the final model.
Results: A 2-compartment model for parent, with assumed 100% conversion to metabolite (1-compartment) was acceptable on VPC and concentration profiles, after removal of BLQ and taper concentrations. Inter-occasion variability on F1 reduced from 51% to 23.8% with BLQ and taper concentrations removed. IIV on CL (33%) and CLm (44%) were moderate but high on KA (249%). Residual variability for parent was 19.9 % and 0.305 ng/mL and for metabolite 7.8% and 0.10 ng/mL, for proportional and additive components. Important covariate effects, after allometric scaling of parent drug, were age and sex on parent clearance (CL) and sex and weight on metabolite clearance (CLm). Compared to a reference male of age 40 y and weight 70 kg with CL=7.29 and CLm =17.2 L/h, the reference female had a 27% drop in parent CL (5.4 L/h) and a 34% drop in CLm (11.4 L/h). Age effect on CL was a 6% decrease for every 10 years above 40 years. The weight effect on parent CL was fixed at 0.75 (allometric) with the power weight effect on CLm estimated at 0.45 (SE 30%).
Conclusions: Covariates age, weight and sex, in combination, predict AUC differences > 2- fold at their extremes. Post hoc graphical analysis showed that the sex covariate on PK did not translate to efficacy differences.
Authors are contractors to (BW) or employees and shareholders of Pfizer Inc (LM, BT, DC).
Reference: PAGE 24 (2015) Abstr 3381 [www.page-meeting.org/?abstract=3381]
Poster: Methodology - Covariate/Variability Models