(T-014) Longitudinal Population Pharmacokinetic/Pharmacodynamic Modeling of Continuous Efficacy Endpoint and Biomarker for a Novel Compound

Yuan Xiong, PhD – Director, Janssen R&D, USA; Arun Kannan, PhD – Associate Director, Janssen R&D, USA; Wangda Zhou, PhD – Associate Director, Janssen R&D, USA; Mahesh Samtani, PhD – Sr Director, Janssen R&D, USA

Objectives: The objective of this work was to demonstrate the utility of including and modeling biomarker data in a longitudinal population PK/PD model for a novel compound.

Methods: To determine whether a relationship existed between a biomarker and a continuous efficacy endpoint for a novel compound, a longitudinal population PK/PD model was fitted sequentially, with population PK first and then simultaneous modeling of efficacy endpoint and the biomarker with PK fixed. The population PK was described using a one-compartment model with first-order absorption, linear elimination, and standard allometric relationship for the effect of body weight on apparent volume and clearance. Indirect response models involving a simultaneous fit of the biomarker and efficacy data were used to describe the longitudinal population PD of the efficacy and biomarker time courses, having a common EC50. The same drug concentration was shared for both indirect response models. The model for the drug effect on the continuous efficacy endpoint had a placebo effect taken into account as placebo data were included in the dataset. Each indirect response model had its own baseline and maximum effects, as well as different values for kout. Each response had its own residual error term. Random effects for both baseline responses and the maximum effect for the biomarker were included. The population under consideration consisted of healthy adults and patients. Modeling and simulation of n=1000 replicates for the visual predictive check (VPC) were performed using NONMEM (version 7.4) and Pirana (Certara). Model diagnostics (including binning and plotting of VPC) were generated using an internal R package.

Results: Sequential population PK/PD modeling with indirect response models for the continuous efficacy endpoint and the biomarker with the common EC50 and the same drug concentrations as a driver fit the data well. The precision of the parameter estimate for EC50 was slightly improved by inclusion of biomarker data in the model, compared with a similar model for efficacy alone without the biomarker. Both the efficacy endpoint and the biomarker appeared to respond similarly to drug concentrations. The efficacy endpoint and the biomarker followed a similar time course.

Conclusions: The model with the common EC50 and the same drug concentrations as a driver for the drug effect on the efficacy endpoint and the biomarker fit the data well and hence demonstrated that the biomarker may in part explain the amelioration of the disease by the novel compound.

Citations: NA