Wanchana Ungphakorn (1), Kayode Ogungbenro (2), Alison H. Thomson (1,3)
(1) Strathclyde Institute of Pharmacy & Biomedical Sciences, University of Strathclyde, Glasgow, (2) Centre for Applied Pharmacokinetics Research, The University of Manchester, Manchester, (3) Pharmacy Dept, Western Infirmary, Glasgow
Objectives: The population approach is a powerful tool to estimate pharmacokinetic (PK) parameters and to identify inter- and intra-individual variability from sparse sampling data but requires careful consideration of design issues to maximise the information that can be extracted from the data. The aims of this study were to (1) use optimal design methods to develop study designs and sampling windows for population PK studies of oral ciprofloxacin in malnourished children and (2) to investigate the influence of prior information on the results of optimal design methods.
Methods: The optimal design was developed using the population Fisher information matrix which is implemented in the PopDes program [1]. A modified Fedorov exchange algorithm with a grid size of 0.25 was used for the optimisation. The structural model, PK parameters and variability obtained from different patient populations were used as input information [2-4]. The maximum number of elementary designs was fixed at 3. Different proportions of subjects and different numbers of samples were examined and both the sample size and sampling windows were determined. The sampling time was limited to between 0 and 12 hours after the dose for most designs but the effect of sampling after a second dose was also investigated.
Results: For 3- and 4-sample designs, the optimal number of groups was 3 and 2, respectively. When using 2 groups, the number of subjects in each group could be varied. If up to 5 samples were allowed to be taken from each patient, one group of subjects would be adequate. The first sampling time point was dependent on the input variable which was related to the lag time specified in the model; however, the last sampling time was always 12 hours. Samples taken only after the first dose gave sufficient information. The expected CV of all parameters was less than 10% with sample sizes of 25 and 40 for 5- and 4-sample designs, respectively. For 3 samples, the CV for Ka remained above 10% despite increasing the sample size to 100. Thus, a higher cut-off point of 20% was used for Ka. It was found that a total of 40 subjects would be enough for a 3-sample study design.
Conclusion: Optimal study designs and sampling windows have been developed for future PK studies in malnourished children. Since the optimal designs were dependent on the prior information, prior knowledge of drug concentration-time profiles should be used with optimal design methods when designing population PK studies.
References:
[1] Gueorguieva, I. et al. A program for individual and population optimal design for univariate and multivariate response pharmacokinetic-pharmacodynamic models.” Comput Methods Programs Biomed 2007;86: 51-61.
[2] Thuo, N. et al. Dosing regimens of oral ciprofloxacin for children with severe malnutrition: a population pharmacokinetic study with Monte Carlo simulation. J Antimicrob Chemother 2011;66: 2336-2345.
[3] Rajagopalan, P. and M. R. Gastonguay . Population pharmacokinetics of ciprofloxacin in pediatric patients. J Clin Pharmacol 2003;43: 698-710.
[4] Schaefer, H. G.et al. Pharmacokinetics of ciprofloxacin in pediatric cystic fibrosis patients. Antimicrob Agents Chemother 1996;40: 29-34.
Reference: PAGE 21 () Abstr 2344 [www.page-meeting.org/?abstract=2344]
Poster: Study Design