Target Attainment Analysis to Evaluate Dosing Regimens of an Oral β-Lactam Antibioic
Elisabet I. Nielsen, Lena E. Friberg, Mats O. Karlsson
Department of Pharmaceutical Biosciences, Uppsala University, Uppsala, Sweden
Objectives: In lack of data supporting mechanism-based PKPD modeling, the selection of dosing regimens within the infectious disease area is commonly based on probability of target attainment (PTA) analysis with the targets relating the drug exposure to the MIC. For β-lactam antibiotics the efficacy has been shown to be related to the percentage of the dosing interval that the free drug concentration exceeds the MIC. The aim of the current study was to (1) evaluate the sensitivity in the PTA for expected PK differences between healthy volunteers and patients (lower CL and/or higher IIV for the latter group) and (2) evaluate a method that acknowledge parameter uncertainty in the PTA analysis.
Methods: A population model was developed in NONMEM 7 based on data from 15 healthy volunteers receiving oral single doses of a β-lactam antibiotic in a randomized cross-over study with intensive PK sampling. Concentration-time profiles following different dosing regimens were simulated and PTA was determined. Simulations were performed with perturbed typical and variability parameters to represent expected parameter distributions in patients. Uncertainty in parameter estimates was acknowledged by performing simulations from a parameter distribution obtained from a non-parametric bootstrap using the sse functionality in PsN.
Results: The PK was well described by a two compartment model with first order elimination with the absorption described by a transit compartment model. Based on the data from healthy volunteers, the IIV was estimated to be between 12-26% for disposition parameters and between 12-46% for absorption parameters. As expected, the β-lactam antibiotic was more effectively dosed 3 times daily than 2 times daily in all simulation settings. When increasing the IIV in the parameters, the PTA curve becomes less steep, with lower target attainment in the high (most interesting) PTA region.
Conclusions: Differences in parameter estimates between healthy volunteers and patients will influence the shape of the PTA curve and thereby also the choice of optimal dosing regimens. Such differences should therefore be taken into consideration in the dose optimization. Simulating from a bootstrap is a simple method to acknowledge and visualize the impact of parameter uncertainty in the PTA analysis.