Population Pharmacokinetics
How population PK models identify sources of variability in drug response across patient groups and inform individualized dosing.
What Is Population Pharmacokinetics?
Population pharmacokinetics (PopPK) is a quantitative approach that characterizes the typical PK parameters of a drug in a target population while simultaneously identifying and quantifying the sources of variability among individuals. Unlike traditional PK studies that require intensive sampling from a few subjects, PopPK can use sparse samples from many patients in real clinical settings.
The Mixed-Effects Model
PopPK uses nonlinear mixed-effects (NLME) modeling. The "mixed" refers to two types of effects:
- Fixed effects: population-typical parameters (mean CL, mean Vd) and the influence of measurable covariates (age, weight, renal function, genotype)
- Random effects: unexplained variability, divided into:
- Between-subject variability (BSV): PK differences among individuals not explained by covariates
- Residual variability: measurement error, model misspecification, within-subject variation
The general model structure:
Individual parameter = Population typical value x exp(eta)
Where eta represents the random deviation of an individual from the population mean.
Covariates
Covariates are patient characteristics that explain part of the between-subject variability:
| Covariate | Typical Effect |
|---|---|
| Body weight | Scales Vd and CL (allometric) |
| Creatinine clearance | Predicts renal CL |
| Age | Affects Vd (body composition) and CL (organ function) |
| Sex | Influences body composition and some CYP activity |
| Genotype | CYP2D6, CYP2C19 polymorphisms alter metabolic CL |
| Disease state | Hepatic/renal impairment, critical illness |
| Co-medications | CYP inhibitors/inducers as binary covariates |
Covariate identification follows a systematic process: graphical exploration, forward inclusion based on objective function improvement, and backward elimination to confirm significance.
Software and Methods
The standard tool is NONMEM (Nonlinear Mixed-Effects Modeling), developed by Lewis Sheiner and Stuart Beal. Other tools include Monolix, Phoenix NLME, and the open-source nlmixr2 (R package). Estimation methods include FOCE-I (First-Order Conditional Estimation with Interaction) and Bayesian MCMC.
Applications
Drug Development
PopPK is integral to regulatory submissions. The FDA expects PopPK analysis for most new drug applications to characterize PK across demographics and identify populations needing dose adjustment.
Model-Informed Precision Dosing
PopPK models combined with Bayesian estimation enable individualized dosing. A patient's sparse drug levels are combined with the population model to generate a posterior estimate of their individual PK parameters, which then guides personalized dose selection.
Special Populations
PopPK can characterize PK in populations excluded from traditional trials: neonates (ethical sampling limitations), pregnant women, and critically ill patients. The sparse sampling design makes it feasible to study these groups.
Key Takeaways
- PopPK quantifies typical PK parameters and sources of variability in patient populations
- Mixed-effects models separate fixed effects (covariates) from random variability
- Covariates like weight, renal function, and genotype explain inter-individual differences
- NONMEM is the standard software; Bayesian methods enable individualized dosing
- PopPK data are increasingly required by regulatory agencies for drug approval