G. Graham1, I. Nestorov1, L. Aarons2
1) Centre for Applied Pharmacokinetic Research, School of Pharmacy, University of Manchester, Oxford Road, Manchester, M13 9PL, United Kingdom. 2) School of Pharmacy, University of Manchester, Oxford Road, Manchester, M13 9PL, United Kingdom.
Purpose: To apply formal optimal experimental design methodology to the design of a clinical trial where the pharmacodynamic response is measured on a categorical scale.
Methods: D-optimal design was used as the design criteria. This optimality criteria minimises the variance-covariance matrix associated with the parameter estimates thus leading to improved parameter estimation. The proportional odds model was used to describe the categorical pharmacodynamic data while NONMEM was used to implement the model fitting.
Results: A phase II clinical trial with an anti-migraine drug, naratriptan, where the categorical response measures pain severity (0=no pain, 1=mild pain, 2= moderate pain, 3=severe pain). The optimal design should supply doses and sampling times at which to take measurements. The results of the D-optimal design for a dichotomised pharmacodynamic pain severity response showed that for a 4 parameter model with 2 covariates, dose and time, there are 4 distinct design points of which 2 are for dose levels and 2 are for sampling times. Replication at these design points was found when repeated measurements above the number of parameters in the model were needed for data collection.
Conclusions: Optimal design methodology, although rarely used in clinical trial design, is a valid and useful tool. Optimal design methodology, although rarely used for clinical trial design where the pharmacodynamic response is measured on binary scale, can be useful and leads to the selection of rational trial parameters. This approach can be extended to the more general case of categorical data.
Reference: PAGE 9 () Abstr 114 [www.page-meeting.org/?abstract=114]
Poster: poster