lPrediction of Vancomycin Area-under-the-curve using Trough Concentrations Only: Performance Evaluation of Pediatric Population Pharmacokinetic Models

Abstract

Objectives In pediatric patients, vancomycin plays a pivotal role in combating infections, necessitating precise therapeutic drug monitoring to ensure efficacy and safety. The adoption of model-informed precision dosing (MIPD) has demonstrated potential in optimizing dosing strategies based on the area under the concentration-time curve (AUC24h). However, predictive performance of population pharmacokinetic models using only trough concentrations to estimate AUC24h has not been evaluated. Methods Predictive performance of 23 vancomycin population pharmacokinetic models was retrospectively evaluated in two cohorts; (A) 21 subjects with PNA<50 days, (B) 124 subjects with PNA≥50 days. Multiple scenarios were investigated using; (a) peak and trough concentration, (b) using either peak or (c) trough concentration solely, (d) covariate information (a priori). The median AUC24h per subject across all models was used as ‘true’ AUC24h. Results For both cohorts, relative root mean square error (rRMSE) for the AUC24h precision using only trough concentrations was similar to the rRMSE using both a peak and trough sample. For cohort A, the model by Chen, Colin, and Mehrotra showed best trough-based performance with the lowest relative Bias (rBias) (-3.3% and -2.6%) and rRMSE (6.8% and 7.3%). For cohort B, the models from Alsultan and Lv illustrated the lowest rBias (1.75% and -5.4%) and rRMSE (16.6% and 15.1%). Conclusions This study illustrates that trough concentration-based AUC24h estimation is a feasible approach in vancomycin MIPD. These findings endorse the selected models for advanced MIPD vancomycin therapy in pediatrics, though further investigation into clinical outcomes is recommended.