Insights into amoxicillin pharmacokinetics using physiology-based pharmacokinetic modelling

ABSTRACT Amoxicillin is the most commonly used antibiotic globally. However, there is a relatively poor understanding of its pharmacokinetics, pharmacodynamics, and clinical pharmacology. We constructed a physiology-based pharmacokinetic (PBPK) model of amoxicillin to gain deeper insights into the adequacy of amoxicillin regimens for the treatment of infections relevant to global health. We constructed an amoxicillin PBPK model in PK-Sim using known ADME and physicochemical parameters, in vitro characterized transporter kinetics of amoxicillin with OAT3, and time-concentration data from the published literature. Unknown parameters were fitted using a subset of available clinical pharmacokinetic data for training, before final validation with a holdout data set. Population simulations were performed using the final model for a range of amoxicillin regimens and contexts. The final amoxicillin PBPK model was high performing by fold-error metrics for both plasma and urinary concentrations. Simulations demonstrated all regimens achieved the 40%T>minimum inhibitory concentration (MIC) target for Streptococcus pneumoniae up to 1 mg/L and all except p.o. 500 mg amoxicillin q8h for the wild-type Haemophilus influenzae MIC distribution. The simulations also demonstrated high urinary amoxicillin exposures and evaluated the effects of inadequate active ingredient content, missed doses, and probenecid co-administration. This PBPK model gives the following insights into amoxicillin pharmacokinetics: i) the adequacy of amoxicillin for infections caused by S. pneumoniae and H. influenzae ; ii) the impact of non-compliance and loss of active ingredients on antimicrobial coverage; iii) the effect of probenecid co-administration to improve coverage; and iv) the characterization of high urinary amoxicillin exposure with consequences for use in urinary tract infections.