Pharmacometric Models for Antibacterial Agents to Improve Dosing Strategies

Antibiotics are among the most commonly prescribed drugs. Although the majority of these drugs were developed several decades ago, optimal dosage (dose, dosing interval and treatment duration) have still not been well defined. This thesis focuses on the development and evaluation of pharmacometric models that can be used as tools in the establishment of improved dosing strategies for novel and already clinically available antibacterial drugs. Infectious diseases are common causes of death in preterm and term newborn infants. A population pharmacokinetic (PK) model for gentamicin was developed based on data from a prospective study. Body-weight and age (gestational and post-natal age) were found to be major factors contributing to variability in gentamicin clearance and therefore important patient characteristics to consider for improved dosing regimens. A semi-mechanistic pharmacokinetic-pharmacodynamic (PKPD) model was also developed, to characterize in vitro bacterial growth and killing kinetics following exposure to six antibacterial drugs, representing a broad selection of mechanisms of action and PK as well as PD characteristics. The model performed well in describing a wide range of static and dynamic drug exposures and was easily applied to other bacterial strains and antibiotics. It is, therefore, likely to find application in early drug development programs. Dosing of antibiotics is usually based on summary endpoints such as the PK/PD indices. Predictions based on the PKPD model showed that the commonly used PK/PD indices were well identified for all investigated drugs, supporting that models based on in vitro data can be predictive of antibacterial effects observed in vivo. However, the PK/PD indices were sensitive to the study conditions and were not always consistent between patient populations. The PK/PD indices may therefore extrapolate poorly across sub-populations. A semi-mechanistic modeling approach, utilizing the type of models described here, may thus have higher predictive value in a dose optimization tailored to specific patient populations