An evaluation of point and interval estimates in population pharmacokinetics using Nonmem analysis

Abstract

In a simulation study of the estimation of population pharmacokinetic parameters, including fixed and random effects, the estimates and confidence intervals produced by NONMEM were evaluated. Data were simulated according to a monoexponential model with a wide range of design and statistical parameters, under both steady state (SS) and non-SS conditions. Within the range of values for population parameters commonly encountered in research and clinical settings, NONMEM produced parameter estimates for CL, V, sigma CL, and sigma epsilon which exhibit relatively small biases. As the range of variability increases, these biases became larger and more variable. An important exception was bias in the estimate for sigma V which was large even when the underlying variability was small. NONMEM standard error estimates are appropriate as estimates of standard deviation when the underlying variability is small. Except in the case of CL, standard error estimates tend to deteriorate as underlying variability increases. An examination of confidence interval coverage indicates that caution should be exercised when the usual 95% confidence intervals are used for hypothesis testing. Finally, simulation-based corrections of point and interval estimates are possible but corrections must be performed on a case-by-case basis.

Pharmacokinetic Study of the Digoxin-acenocoumarol Interaction in Rabbits

“…The difficulty in estimating intersubject variability in the Vd with precision as encountered in the present study has already been reported in the literature for drugs with a long halflife. 17,18) Although the modifications observed in the partition coefficient of digoxin in heart both in vitro and in vivo suggest a decrease-and not an increase-in the digoxin Vd, this is not necessarily in conflict with the results obtained in the population kinetic analysis. However, such a decrease would only occur if the decrease in the partition coefficient were to occur in all tissues and not only in the heart, the site of action of the drug.…”

The back-step method—Method for obtaining unbiased population parameter estimates for ordered categorical data

“…To assess the bias in the parameter estimates, 100 original data sets were simulated for each of the conditions (1), (2), and (3) using the nominal parameter values listed in Table 1. Each original data set was then analyzed once with the standard mixed effects modeling approach, resulting in one Φ STD for each simulated data set, and once with the Back-Step Method, resulting in one Φ BSM for each simulated data set.…”

“…It has previously been shown that both the first order (FO) method and the first order conditional estimation (FOCE) method in NONMEM can produce biased pharmacokinetic (PK) parameter estimates if the variability of the data are high. 2,3 The FOCE method has also been shown to produce considerable bias in pharmacodynamic (PD) parameters. 4 More recently, Jonsson et al 5 showed that the Laplacian method in NONMEM produces biased parameter estimates when using a nonlinear pharmacodynamic model to describe the data.…”

Population Pharmacokinetics of Therapeutic Monoclonal Antibodies: Examples and Estimation Method Performance Differences

“…The challenge with detecting and characterizing nonlinear PK is not only dependent on the study design, but also on the estimation method used for the population PK analysis (163). However, little work has been done evaluating population estimation methods using PK models that are representative of the typical disposition characteristics of therapeutic mAbs, as most method comparison studies used a one-compartment model with linear elimination for comparison (142)(143)(144)(145)(146)(147). In the current study, I evaluated the FO, FOCE-I, and LAP-I methods in NONMEM and a full Bayesian MCMC method in WinBUGS with both uninformative and informative priors in population PK modeling of therapeutic mAbs with nonlinear PK.…”

“…A number of studies have evaluated the performance of different estimation methods for population PK modeling, and in most of these studies a one-compartment model with IV bolus or first-order input and first-order (linear) elimination was used for comparison (142)(143)(144)(145)(146)(147). Studies by Hashimoto et al (148) and Sheiner and Beal (149) evaluated estimation methods in NONMEM for population PK modeling of nonlinear PK data at steady-state.…”