Population pharmacokinetic modeling has been employed as a central part of the delineation of exposure response relationships for anti-infective agents. We have prospectively developed a relationship between the patient-individual exposure relative to the MIC of the infecting pathogen to the fluoroquinolone antibiotic levofloxacin and the probability of that patient having a successful clinical or microbiological outcome (the patient did well/the organism was eradicated from the primary infection site) (1). These relationships were developed in 22 separate centers simultaneously. In order to accomplish this, a population optimal sampling strategy was developed and the data base was developed. Plasma concentration-time profiles were obtained in 272 patients, of whom 134 had a microbiologically-documented infection. Population pharmacokinetic modeling was followed by MAP-Bayesian estimation and patient simulation produced the pharmacodynamic variables of Peak/MIC ratio, AUC/MIC ratio a nd the Time Plasma Concentrations Exceed the MIC. These (among other covariates) were examined in a separate analysis in a logistic regression analysis, so that a measure of exposure could be linked to the probability of outcome. While this demonstrates that a number of relatively sophisticated techniques can be linked to produce a relationship between drug exposure and outcome in a prospective fashion (2-4), it would be desirable to perform similar analyses in a fully integrated, single step manner. However, the ability to build such sophisticated models and run them on relatively large data bases would stretch the utility of the currently extant population pharmacokinetic modeling programs past the breaking point. Antiviral chemotherapy provides a particularly attractive opportunity to demonstrate the power of the modeling process. The importance of the intervention for clinical outcome is clear (5,6). Virtually no -guidance is available for dosing these agents in order to achieve the desired effect. For many agents, surrogate markers are available which have been demonstrably linked to clinical outcome. Examples include RNA PCR for HIV and, more recently, RNA PCR for Hepatitis C and DNA PCR (as well as a pp65 antigen assay) for Cytomegalovirus (7,8). Each of these assays provides quantitative measures of drug effect on the replication of the pathogen in question. This allows important questions to be asked and answered using population kinetic/dynamic modeling as the central tool.
Showing the most recent 10 out of 77 publications
