Design of experiment (DOE) is a systematic experimental approach recommended by regulatory agencies for optimization/characterization of unit operations used in the manufacturing of therapeutics product. It leverages the power of statistics (empirical modelling) to evaluate a broad design space and determine the critical process parameters and their impact on critical product quality attributes while staying within the experimental budget. For a typical process development/characterization, often a two tiered sequential experimental approach is preferred. The first tier utilizes a screening design to assess the relative impact of a large number of factors and their interactions on a response of interest. Statistically significant process parameters identified from screening designs are then further studied using more comprehensive designs (e.g Response Surface [RSD]) in the development of a better predictive model.<br/>Fractional factorial (FF) design is a classical screening design which is used to determine the impact of main and second order interaction effects. This 2-level design lacks the ability to independently estimate quadratic effects (curvature in factor to response relationship) and thus often require additional experiments to better understand relationships of key process parameters. Definitive screening design (DSD) has emerged as an effective screening strategy due to its capability to estimate quadratic effects in addition to main effects and second order interactions. Although the statistical power of DSD design does not match that of RSD, it does provide the opportunity to develop a 2<sup>nd</sup> degree model with only a single screening study.<br/>This study focuses on the comparison of DSD with classical FF screening design for the characterization of flow through chromatography operation in purification of mAbs. DSD and FF screening designs were generated using the similar process parameters and experimental budget. Key responses were modelled in both designs. DSD and FF designs were compared for a) model diagnostic statistics, b) identification of significant process parameters and c) model predictions for key responses.

Process characterization of flow-through chromatography: Fractional factorial vs definitive screening designs

ACS Spring 2022

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.