Development of a Population Pharmacokinetic (PopPK) Model with Covariate Analysis Using NONMEM: An Educational Drug Development Case Study

This project presents the development of a Population Pharmacokinetic (PopPK) model using dataset D_3 as part of the course CDD 205: Modelling for Drug Development at the UCSF School of Pharmacy. The objective was to construct a structural, statistical, and covariate model that minimizes the Objective Function Value (OFV) while accurately describing inter-individual variability and predicting individualized dosing. The dataset consists of 78 subjects who received repeated oral dosing (1000 mg, 1500 mg, or 2000 mg) every 24 hours for 20 days. Intensive concentration measurements were collected over a 24-hour period after steady state was reached. Key variables include dosing history (AMT, ADDL, II), event identifiers (EVID), observation times (TIME), drug concentrations (DV), and multiple clinical covariates including weight (WT), liver function markers (TBIL, AST, ALT), renal function (eGFR), sex, and race. A one-compartment oral structural model was identified from the data. The model incorporated proportional and additive residual error components and estimated inter-individual variability on clearance (CL), volume of distribution (V), and absorption rate constant (KA). Extensive covariate analysis was performed to identify physiologically relevant predictors of pharmacokinetic variability. Among multiple tested covariates (WT, TBIL, AST, ALT, eGFR, sex, race), total bilirubin (TBIL) demonstrated the most significant impact on clearance, reducing the OFV by 22 points when incorporated into the model. This finding supports a predominantly hepatic clearance pathway for the drug. The final model integrates TBIL as a covariate on clearance and demonstrates improved goodness-of-fit, as shown by individual and population prediction plots. This work illustrates a complete PopPK modeling pipeline including: • Dataset exploration and visualization• Structural model selection• Statistical model refinement• Covariate screening and selection• NONMEM model implementation• Goodness-of-fit evaluation• Clinical interpretation of covariate effects The project serves as an educational and reproducible example of PopPK model development and covariate analysis for drug development applications.