Targeting Vancomycin AUC in Neonates − A Model‐Based Bayesian Approach for Personalized Therapeutic Drug Monitoring

BACKGROUND: When treating methicillin-resistant Staphylococcus aureus (MRSA) infections with vancomycin, national guidelines recommend targeting an AUC24/MIC ≥400 to ensure adequate drug exposure. To optimize neonatal vancomycin dosing, accurate AUC24 estimates are needed. The objective of this study was to assess the utility of a model based Bayesian approach for estimating vancomycin AUC24 in neonates.

METHODS: Neonates who received vancomycin and had ≥1 ‘peak’ and ≥1 ‘trough’ concentrations at two healthcare systems (2006-2013) were studied. Bayesian estimates of clearance were calculated for each neonate using a published, externally validated population pharmacokinetic model in NONMEM (external validation was not completed until October 2014). AUC24 was calculated as the daily dose Û clearance. The percent prediction error (PE) and the percent absolute prediction error (APE) of the AUC24 estimates were compared for: 1) the full dataset, 2) a dataset with only the first peak and trough concentrations, and 3) a dataset with only the first trough concentration.

RESULTS: A total of 427 neonates were studied (median [IQR] postmenstrual age 36 [29-41] weeks and weight 2.3 [1.0-3.4] kg). Compared with the full dataset, Bayesian estimates of AUC24 using only the first trough concentration had a median PE of -0.7% (95%CI: -1.3% to 0.0%) and a median APE of 4.1% (95% CI: 3.5% to 4.8%). AUC24 predictions were within 15% of the full dataset for 90% of neonates. The addition of a peak concentration provided no substantial predictive benefit.

CONCLUSION: A model based therapeutic monitoring strategy using only a single trough concentration can adequately predict vancomycin AUC24 in neonates. Application of this approach can help clinicians personalize vancomycin therapy and warrants further study.