Sathe P, Tsong Y, Shah V P
Office of Pharmaceutical Sciences, U.S. Food and Drug Administration, Rockville, Maryland 20855, USA.
Adv Exp Med Biol. 1997;423:31-42. doi: 10.1007/978-1-4684-6036-0_3.
Dissolution data for the immediate or modified release drug products are usually collected as percent dissolved at multiple time points. Once an in-vitro/in-vivo relationship is established on a drug product, the dissolution profile becomes meaningful and important. In that context, if a firm desires to modify its formulation on which the in-vitro/in-vivo association has been established, a meaningful insight into the pharmacokinetics may be obtained by comparing the dissolution profiles of the two lots. In this presentation, we demonstrated a model dependent dissolution profile comparison approach using example of carbamazepine tablet dissolution data. Once a mathematical function was selected to describe the dissolution data coming from various standard lots, a similarity region could be constructed using the model parameter variances. To compare the test and reference lot dissolution profiles, a statistical distance was calculated between the mean parameters. A confidence region generated around the normalized mean statistical distance could then be compared with the similarity region to assess the similarity or dissimilarity of the dissolution profiles.