Simulation of an optimal pharmacokinetic and pharmacodynamic amikacin dosing regimen for neonates

Objectives: Evaluation of the currently used amikacin dosing regimen in neonates and proposal of an optimal pharmacokinetic (PK) and pharmacodynamic (PD) amikacin dosing regimen.

Methods: A retrospective chart review was performed including all neonates treated with amikacin at Dunedin Hospital from Oct 2003 to Jan 2007. The population PK analysis included all the study subjects and was performed using NONMEM, version 5. Data was collected from 80 neonates and the final PK model analysed 358 amikacin concentrations. A clinical audit was also undertaken to review hearing tests performed on surviving neonates (n=36) from the original PK study who were considered to be at risk of hearing impairment due to prematurity, low birth weight, jaundice and/or use of aminoglycosides. Two separate PD analyses were undertaken: the first included all neonates with culture proven sepsis (n=26) and used posthoc estimates from the PK model and minimum inhibitory concentrations (MIC) data in a logistic regression model using Stata, version 8. The outcome measure was treatment failure. The second PD analysis used logistic regression to evaluate a specific outcome measure related to hearing impairment in neonates treated with amikacin. The final covariate PK model was used to develop a simulation model to evaluate the current amikacin dosing regimen and to propose a new dosing regimen using MATLAB (student version 7.1). The simulations used a nonparametric dataset of covariate distributions which, included postmenstrual age (PMA) (24.7–44.1 weeks) and current weight (CWT) (0.45-4.43 kg) from 719 individual neonates. The new dosing regimen was required to achieve treatment success under the following criteria: Cmax 24-35 mg/L and AUC24 130-590 mg•L/h.

Results: There were 46 (57.5%) extremely premature neonates (PMA <28 weeks) and 44 (55%) of these neonates had a birth weight of <1000g. The final covariate model was a one-compartment first order elimination PK model. The final covariate model estimated clearance (CL) =0.23•3.23(PMA) •0.691(CWT) and volume of distribution (V) =0.957•0.89(CWT). Results from the PD analysis were used to adjust the optimal target value for the peak concentration of amikacin in the simulations. MATLAB was used to simulate 1000 neonates and modelled the amikacin given by a 30 min infusion for two doses. Dose intervals varied from 24-48 h and doses given from 13-19 mg. Simulation and evaluation of the current dosing regimen showed after dose one, potentially 24% of neonates (PMA <27 weeks) achieved a Cmax >35 mg/L and 44% of neonates (PMA 28-30 weeks) had a Cmax >35 mg/L. Simulation of a new dosing regimen recommended a lower dose (mg/kg) for neonates <28 weeks than the currently recommended higher dose for neonates <28 weeks. There was little difference in the dose (mg/kg) recommendations for neonates ≥28 weeks PMA.

Conclusions: Evaluation of the current amikacin dosing regimen demonstrated poor achievement of optimal peak concentrations in neonates. Overall, the simulation of a new dosing regimen resulted in a suggested regimen, which appears to be clinically realistic and predicts well for neonates of all ages.

Catherine Sherwin