ENBIS: European Network for Business and Industrial Statistics

Application for showing similarity of dissolution profiles with method based on multivariate statistical distance

23 September 2009, 10:45 – 11:05

Abstract

Submitted by

Heli Rita

Authors

Anni Liimatainen and Heli Rita

Affiliation

Orion Pharma

Abstract

There’s a frequent need for showing similarity of the dissolution profiles of two pharmaceutical products or formulations. It is essential in the biowaiver approach, replacing the in vivo bioequivalence study in a regulatory submission, and whenever remarkable changes are made to the manufactured products or their manufacturing processes.

The methods to be used for showing similarity of the dissolution profiles are presented in several guidelines by the regulatory authorities. The topic has also been widely discussed in pharmaceutical papers. The most commonly used method which is openly supported by most regulatory agencies, the similarity factor f2, is solely based on the mean profiles. It thus ignores the possibly notable effect of variability in dissolution between drug units. This approach has been strongly criticized in several papers, and more adequate methods have been proposed.

A method, which utilizes information about the variability between individual dissolution profiles, is based on the multivariate statistical distance (MSD). This method is well described in the guideline of the FDA and more accurately in the paper by Tsong et al. (Tsong, Y, Hammerstrom, T, Sathe, P, Shah, V.P. (1996) “Statistical assessment of mean differences between two dissolution data sets”, Drug Information Journal, Vol. 30, 1105-1112). Surprisingly, it appeared extremely difficult to apply in practice due to complicated calculations. We managed to program the calculations in SAS interactive matrix language (IML) and apply the method successfully in regulatory applications. Even an easy-to-use end-user interface was developed in the SAS environment, by utilizing the SAS Add-in for Microsoft Office. The end-user may start the developed system from within the MS-Word and get the resulting output, containing graphical visualisations, descriptive statistics and the results of similarity comparisons, in an MS-Word document. The system is illustrated with some examples.

View paper

Return to programme