Magali Karagueuzian, Development Specialist, Beckman Coulter (Danaher Corporation)
Developing an assay for diagnosing human antibody is particularly difficult while using poorly soluble protein. We optimised the coupling process of the protein on the solid phase (concentration, pH, glycerol) and the measurement conditions of the antibodies from the patient (incubation time, concentration of tracers) for a total of six factors. At the time, to select the appropriate design of experiment we had the choice between the 1) old fashioned one-factor-at-a-time, which requires 13 conditions (three levels of each factor, the centre point being the same for all factors) without identifying interactions between factors, without repeat; 2) fractional factorial, which requires 32 conditions (all second-order interactions) without identifying quadratic terms; or 3) response surface design, which requires 53-54 conditions (central composite design or Box-Behnken) to get both quadratic terms and interactions. Thanks to definitive screen designs (DSDs), we identified the significant factors and their quadratic terms and interactions, running only 13 conditions and managing outliers. This makes this type of design of experiment the best-in-class method for complex development and for improving time to market and quality, two major KPIs for development.