Regression To The Rescue: The Use Of Statistical Doe To Derive A Novel Manufacturing Control Strategy For Ensuring The Quality Of An Antibody-Drug Conjugate
Lisa Bernstein, Genentech Inc  *Daniel A. Coleman, Genentech Inc. 

Keywords: antibody-drug conjugate, control strategy

ADC-X is an antibody-drug conjugate therapeutic candidate consisting of a monoclonal antibody linked to a potent cytotoxin by a chemical linker. The Drug-to-Antibody Ratio (DAR), the average number of toxin molecules per antibody, is a critical quality attribute related to both safety and efficacy; hence it must be controlled in manufacturing. This talk describes how a sequential experimentation strategy characterized the ADC-X manufacturing process. Multiple laboratory constraints on experimentation were managed by incorporating blocking and split-plotting into fractional-factorial designs. These characterization studies identified the important factors in each process step and provided a regression model relating those factors to the DAR and other attributes. The presentation will also show how the designed experiments uncovered a serious challenge to commercial ADC-X manufacturing and then provided a tool that met the challenge: the regression model formed the basis for a novel process control strategy involving an adjustable setpoint for the amount of linker in each batch. This control strategy has been successfully implemented in large-scale manufacturing.