Variable Selection in Multivariate Modeling of Drug Product Formula and Manufacturing Process

Multivariate data analysis methods such as partial least square (PLS) modeling have been increasingly applied to pharmaceutical product development. This study applied the PLS modeling to analyze a product development dataset generated from a design of experiment and historical batch data. Attention was paid in particular to the assessment of the importance of predictor variables, and subsequently the variable selection in the PLS modeling. The assessment indicated that irrelevant and collinear predictors could be extensively present in the initial PLS model. Therefore, variable selection is an important step in the optimization of the pharmaceutical product process model. The variable importance for projections (VIP) and coefficient values can be employed to rank the importance of predictors. On the basis of this ranking, the irrelevant predictors can be removed. To further reduce collinear predictors, multiple rounds of PLS modeling on different combinations of predictors may be necessary. To this end, stepwise reduction of predictors based on their VIP/coefficient ranking was introduced and was proven to be an effective approach to identify and remove redundant collinear predictors. Overall, the study demonstrated that the variable selection procedure implemented herein can effectively evaluate the importance of variables and optimize models of drug product processes.