Pascal Anger

Methodology for transfer of liquid chromatography methods based on statistical considerations

Abstract An important task in the pharmaceutical industry today is analytical transfer. However, no actual guidelines are available today. It is for this reason that we decided to devise a rigorous method using statistic exploitation of results. The statistical technique used is ANOVA (analysis of variance). We chose to treat the case of quantitative analysis in LC but the methodology could easily be adapted to other analytical techniques. The criteria of the transfer validation could be formulated thus: “for each response of interest, the new laboratory must produce results that are not significantly different from those of the reference entity”, or more explicitly by: “the new laboratory must have dispersion characteristics compatible with those of the reference entity and must exhibit no bias”. While compatibility of precision can easily be assessed, the test of absence of bias requires that certified materials be available. Since certified materials can only be obtained through an inter-laboratory study, it means that the reference entity is necessarily a pool of laboratories. Using a single laboratory instead of a pool would not allow a distinction to be drawn between a bias and an inter-laboratory dispersion, which would lead to abnormal transfer failure. The methodology developed was then used on an example. The last part deals with the situation where certified materials are suspected to be slightly degraded. It is explained how such a case, likely to be encountered in pharmaceutical products stored over a long period, can be handled without re-starting the study from scratch.