Factoring fu Variability Into Estimates of Unbound Drug Concentrations Negatively Biases the MIC Versus % Probability...

The clinical breakpoint for a drug-pathogen combination reflects the drug susceptibility of the pathogen wild-type population, the location of the infection, the integrity of the host immune response, and the drug-pathogen pharmacokinetic (PK)/pharmacodynamic (PD) relationship. That PK/PD relationship, along with the population variability in drug exposure, is used to determine the probability of target attainment (PTA) of the PK/PD index at a specified minimum inhibitory concentration (MIC) for a selected target value. The PTA is used to identify the pharmacodynamic cutoff value (CO_PD), which is one of the three components used to establish the clinical breakpoint. A challenge encountered when defining the CO_PD is that the available PK information typically reflects total (free plus protein-bound) plasma concentrations. However, it is the unbound drug concentrations that exert the therapeutic effects and how the population fraction unbound (fu) incorporated into the CO_PD assessments can markedly influence the CO_PD. Factors examined included the estimated population fu mean (risk of bias) and the incorporation of estimated fu population variability into the Monte Carlo simulations when converting total to unbound plasma concentrations (risk of inflating variability). In this in silico study, the drug fu, systemic clearance, and the variability of both were altered so that the relative impact of each could be explored. We demonstrate that incorporating fu variability into the estimation of fAUCback can bias the CO_PD assessment and that the magnitude of bias reflects the relative variability in systemic clearance and fu.