NONMEM Analysis Research Articles

A three-step approach combining bayesian regression and NONMEM population analysis: Application to midazolam

NONMEM, the only available supported program for population pharmacokinetic analysis, does not provide the analyst with individual subject parameter estimates. As a result, the relationship between pharmacokinetic parameters and demographic factors such as age, gender, and body weight cannot be sought by plotting demographic factors vs. kinetic parameters. To overcome this problem, we devised a three-step approach. In step 1, an initial NONMEM analysis provides the population pharmacokinetic parameters without taking into account the demographic factors. Step 2 consists of individual bayesian regressions using the measured drug concentrations for each subject and the population pharmacokinetic parameters obtained in step 1. The bayesian parameter estimates of the individual subject can be plotted against the demographic factors of interest. From the scatter plots, it can be seen which are the demographic factors that appear to affect the pharmacokinetic parameters. In step 3, the NONMEM analysis is resumed, and the demographic factors found in step 2 are entered into the NONMEM regression model in a stepwise manner. This method was used to analyze the pharmacokinetics of midazolam in 64 subjects from 714 plasma concentrations and 11 demographic factors. CL (elimination clearance) and V1 were found to be a function of body weight. Age and liver disease were found to decrease CL. Of the 11 demographic factors recorded for each patient, none was found to influence VSS or intercompartmental clearance.

POPULATION PHARMACOKINETICS OF PHENOBARBITAL IN NEWBORNS DETERMINED BY NONMEM ANALYSIS

Routine clinical pharmacokinetic (PK) data from 59 newborns, gestational ages 24-42 weeks (mean ± SD = 31±4.1) and birthweights 600-3620 gm (mean ± SD = 1520±700) who received anticonvulsant doses of phenobarbital (PB) were evaluated using NONMEM analysis. Fifty of these infants were less than 1800 gm and were treated for one week to prevent cerebral hemorrhage. Infants initially received 20 mg/kg of PB IV followed by maintenance doses of 2.5 mg/kg every 12 hours. Serum PB was measured 2 hours after loading and on days 3 and 6.Mean serum PB concentration post-loading was 19.8 ± 3.7 (SD) μg/ml. The mean apparent volume of distribution (AVd) was 0.96±0.024 (SEM) L/kg with a coefficient of variation (CV) of 16%. Mean clearance was 0.08 ± 0.003 (SEM) ml/min/kg, with a CV of 19%. Calculated mean serum half-life was 140 hrs.Within this population, neither gestational age nor birthweight had an effect on PK parameters. Asphyxia (5 min Apgar score < 5), on the other hand, had an effect on AVd, with a 13.7% increase noted in asphyxiated versus non-asphyxiated infants, p < 0.05.We conclude: (1) NONMEM is an accurate and easily implemented method of population PK analysis in newborns, (2) inter-individual variability in AVd and clearance of PB is minimal in the first week of life, and (3) alterations in PK parameters of phenobarbital are seen in asphyxia, where increases in AVd may result in lower than expected serum concentrations.