Abstract
The aim of this study was to compare the performance of parametric and nonparametric approaches in population pharmacokinetic analysis for cystic fibrosis patients.
The study involved a retrospective analysis of a database of aminoglycoside concentration measurements in patients with cystic fibrosis from Glasgow and The Hague. The data had previously been analysed using a traditional parametric population modelling approach using NONMEM (version 7) (1). The present analysis focuses on population analysis using a non-parametric approach with the software Pmetrics (2). One and two compartment models and the influence of covariates, using both simple and mechanistic approaches, were compared for parametric and non-parametric approaches.
The combined dataset included 331 patients (166 from Glasgow) with 1490 courses of therapy and 3690 aminoglycoside concentration measurements. The NONMEM analysis found that a two compartment model was superior to a one compartment model. This was confirmed using the non-parametric approach where the two compartment model was again superior (-2 likelihood value 9004 vs 9618). The population estimates of clearance (CL) and V1 generated from NONMEM and Pmetrics were very similar. In addition, both NONMEM and Pmetrics identified creatinine clearance and height as factors influencing CL and height influencing V1.
Similar results were obtained when parametric and nonparametric approaches were used to analyse aminoglycoside data from patients with cystic fibrosis
The study involved a retrospective analysis of a database of aminoglycoside concentration measurements in patients with cystic fibrosis from Glasgow and The Hague. The data had previously been analysed using a traditional parametric population modelling approach using NONMEM (version 7) (1). The present analysis focuses on population analysis using a non-parametric approach with the software Pmetrics (2). One and two compartment models and the influence of covariates, using both simple and mechanistic approaches, were compared for parametric and non-parametric approaches.
The combined dataset included 331 patients (166 from Glasgow) with 1490 courses of therapy and 3690 aminoglycoside concentration measurements. The NONMEM analysis found that a two compartment model was superior to a one compartment model. This was confirmed using the non-parametric approach where the two compartment model was again superior (-2 likelihood value 9004 vs 9618). The population estimates of clearance (CL) and V1 generated from NONMEM and Pmetrics were very similar. In addition, both NONMEM and Pmetrics identified creatinine clearance and height as factors influencing CL and height influencing V1.
Similar results were obtained when parametric and nonparametric approaches were used to analyse aminoglycoside data from patients with cystic fibrosis
| Original language | English |
|---|---|
| Pages (from-to) | 2931-2931 |
| Number of pages | 1 |
| Journal | PAGE. Abstracts of the Annual Meeting of the Population Approach Group in Europe. |
| Volume | 22 |
| Publication status | Published - 11 Jun 2013 |
| Event | PAGE 2013 meeting - Glasgow Royal Concert Hall, Glasgow, United Kingdom Duration: 11 Jun 2013 → 14 Jun 2013 |
Keywords
- NONMEM
- pmetrics analysis
- aminoglycosides
- adult patients
- cystic fibrosis
- population pharmacokinetic analysis