Background: Substantial progress of mechanism-based and empirical modelling for anti-infectives over the past decade has enabled these models to rationally translate the time-course of killing and emergence of resistance from in vitro to animal infection models and ultimately to patients.
Objectives: To illustrate applications of translational mechanism-based models for anti-infective mono- and combination therapy and point out gaps and experimental data needed to improve the translational predictions by these models.
Methods: We used data from static time-kill experiments (SCTK) and dynamic 1-compartment and hollow fibre infection models. Bacterial growth and killing by anti-infective mono- and combination therapy was described by mechanism-based, life-cycle growth models that accounted for different antibiotics affecting specific target sites. Mechanism-based models were developed in NONMEM® and S-ADAPT using pooled fitting and importance sampling algorithms. Deterministic and Monte Carlo simulations to predict killing and emergence of resistance in patients were performed in Berkeley Madonna.
Results: SCTK studies provided a wealth of exposure-response information over usually 48 h and were crucial for model development, but do not describe the change of antibiotic concentrations and receptor occupancy over time. As expected, mathematical models based on 48-h SCTK data had limitations for predicting emergence of resistance on days ~4 to 10 observed in the hollow fibre system. Life-cycle growth models enabled us to predict the rate of killing and emergence of resistance for slowly and rapidly replicating bacteria assuming that the rate of bacterial growth and the rate of killing are correlated for many antibiotics. Our mechanism-based model for ceftazidime successfully predicted the exposure targets required for bacteriostasis and near-maximal killing at 24 h in mice and for cure of patients.
Conclusions: Mechanism-based models for anti-infective mono- and combination therapy have enabled a rational translation from in vitro to animal infection models and ultimately to patients. Accounting for the rate of bacterial growth, killing, and resistance in vivo as well as for the function of the immune system was critical. More experimental data on the time-course of killing and resistance in vivo as well as on the function and between patient variability of the immune system will be critical to prospectively optimise antibiotic regimens for patients.