Prediction of vancomycin exposure in patients with central nervous system infections using physiologically based pharmacokinetic modeling.

Vancomycin remains a key therapeutic option for central nervous system (CNS) infections caused by Gram-positive bacteria, yet its limited and variable penetration into the cerebrospinal fluid (CSF) poses challenges for optimal dosing. This study aimed to develop and validate a physiologically based pharmacokinetic (PBPK) model to predict vancomycin exposure in both plasma and CSF following intravenous administration in patients with CNS infections. The CNS PBPK model incorporated blood, brain mass, cranial and spinal CSF compartments, and assumed passive permeability across CNS barriers. Clinical data from healthy subjects and CNS-infected patients were used for validation. Model performance was assessed by fold error analysis. Virtual simulations were performed to compare intermittent versus continuous infusion regimens. The developed CNS PBPK model reliably predicted vancomycin concentrations in plasma and cerebrospinal fluid. For 96.51% of predicted values, deviations from observed data fell within a range of 0.5 to 2 times the measured concentration, with a mean fold difference of 1.25. Specifically, 96.28% of predicted plasma concentrations fell within 0.5 to 2 times the observed values, while all predicted CSF concentrations remained within 0.5 to 2 times the observed values. Scaling factor of 4 for permeability yielded a satisfactory fit to CSF vancomycin concentrations in infected patients. Model simulations indicated that, at the same daily dose, continuous infusion achieved more stable and higher CSF trough levels than intermittent dosing. This study presents the CNS PBPK model for vancomycin capable of accurately predicting drug disposition in CSF and plasma. The model supports individualized dosing strategies and provides a quantitative framework for optimizing vancomycin therapy in CNS infections.