Pharmacometric Approaches to Precision Therapeutic Management for Antimicrobials
MetadataShow full item record
AbstractAntimicrobials have been widely used for decades in the treatment of various types of bacterial infections and their properties have been thoroughly characterized in pediatric and adult patients. However, high variability and unpredictability of antimicrobials’ pharmacokinetics (PK) in patients still exist, which reinforces the value of precision dosing. The research in this thesis highlights the role of pharmacometrics in precision therapeutic management of two prototype antimicrobials, gentamicin and rifampin. The first project developed a Bayesian forecasting algorithm for precision dosing of gentamicin in pediatrics. We developed the first population PK model for gentamicin across the whole pediatric age spectrum ranging from 1-day-old newborns to 19-year-old young adults. The model utilized physiologically plausible covariate parameterization driven by principles of allometric scaling. Renal function changes manifested by glomerular filtration were described by postmenstrual age, and serum creatinine was standardized by age. The model was used as a prior in the subsequent full Bayesian analyses in pediatric patients. A full Bayesian analysis-based model-informed precision dosing (MIPD) was introduced for gentamicin dosing in pediatric patients. With a predefined probability of target attainment (PTA) criteria of 70% for both maximum and trough concentrations, the dosing regimens recommended by the empirical dosing guideline NeoFax could achieve the predefined criteria in about 5% of the 1013 patients, in comparison with 90% of the patients when the initial dosing recommendation from the MIPD approach was used. Finally, a workflow was designed for a new patient in a clinical scenario to provide MIPD for initial dosing recommendation and dosing adjustment after TDM level becomes available via a full Bayesian approach. The second project focuses on dose optimization of rifampin in adult patients with tuberculosis through dynamic positron emission tomography (PET) scans. A semi-mechanistic PK-lung-biodistribution model was developed based on plasma and intralesional drug concentration data measured by PET scans. The model could well predict the mass spectrometry data from therapeutic dose and PET data from 11C-labled micro-dose. The developed model was externally validated through exposure predictions in the therapeutic range of 10-35 mg/kg. Based on the projected drug exposure in the cavity walls at higher rifampin doses, the bacterial killing curves obtained from hollow fiber systems were used to predict the clinical cure rates in humans for higher rifampin doses (>600mg). Standard oral rifampin dosing of 10 mg/kg would achieve a 95% probability of cure in 6-9 months of treatment. Similarly, an oral rifampin dose of at least 35 mg/kg would be needed to cure patients in 4 months.
University of Maryland, Baltimore