Data-driven personalization of a physiologically based pharmacokinetic model for caffeine: A systematic assessment

Area of Science:

• Pharmacokinetics and Drug Metabolism
• Computational Biology and Bioinformatics
• Precision Medicine

Background:

• Physiologically based pharmacokinetic (PBPK) models offer potential for precise individual drug dosing.
• Current clinical application of PBPK models for precision dosing remains limited.
• Systematic assessment of individual patient data integration into PBPK models is needed.

Purpose of the Study:

• To evaluate the impact of individual patient data on pharmacokinetic (PK) prediction accuracy using PBPK models.
• To determine the incremental benefit of incorporating demography, physiology, and metabolic phenotype into PBPK models.
• To assess the potential of personalized PBPK models for model-informed precision dosing.

Main Methods:

• A PBPK model for caffeine was developed and stepwise personalized using data from 48 healthy volunteers.
• Individual data included demography (age, sex, height, weight), physiology (liver blood flow, GFR, hematocrit), and CYP1A2 phenotype.
• Model performance was compared against a baseline model using average individual parameters, with predictions assessed against observed caffeine concentrations.

Main Results:

• Personalization based on demography alone improved prediction accuracy from 45.8% to 57.8% within a 0.8- to 1.25-fold range.
• Adding measured physiological parameters did not significantly enhance prediction accuracy (59.1%).
• Incorporating individual demography, physiology, and CYP1A2 phenotype resulted in the highest accuracy, with 66.15% of predictions within the 1.25-fold range.

Conclusions:

• Individual patient attributes significantly improve the accuracy of pharmacokinetic predictions.
• Personalized PBPK models demonstrate considerable potential for enhancing model-informed precision dosing strategies.
• Further application of personalized PBPK models in clinical practice could optimize therapeutic outcomes.