(470f) Defining the Optimal Operating Window for Pharmaceutical Reactions Using the Dynamic Response Surface Methodology for All Measured Species

Authors: 
Dong, Y., Tufts University
Georgakis, C., Tufts University
Mustakis, J., Pfizer Inc.
Hawkins, J. M., Pfizer Inc.
McMullen, J. P., Merck & Co. Inc.
Grosser, S. T., Merck & Co. Inc.

For pharmaceutical reactions, it is of interest to identify the end-of-reaction conditions that lead to a final mixture in which the product is above a certain benchmark concentration and the concentrations of the by-products or intermediates are below target values. For robust manufacturing processes, the end of reaction conditions must be stable to ensure that impurity growth does not occur during the in-process control sample procedure or during other events that may require extended cycle times. Numerically, this robustness can be defined as the conditions that maximize the time window where the batch concentrations remain within the product quality constraints. This is achieved here by utilizing the Dynamic Response Surface Methodology (DRSM) Model1,2 a recently proposed data-driven method that models time-resolved outputs.

Here, we apply a constrained variation of the DRSM-22 model in a simulated pharmaceutical reaction network of eight reactions involving ten species, as well as experimental data from the two collaborating pharmaceutical companies. The DRSM models for the time evolution of the measured species concentrations under different operating conditions play a crucial role in the solution of the above optimization problem. The optimal operating widows is initially estimated from the normal predictions of the DRSM model. This can be updated by accounting for the prediction interval of the DRSM. We examine an additional robustness issue related to the sensitivity of the calculated operating window with respect to a possible variation of the operating conditions from the calculated optimal ones.

References:

  1. Klebanov N, Georgakis C. Dynamic Response Surface Models: A Data-Driven Approach for the Analysis of Time-Varying Process Outputs. Industrial & Engineering Chemistry Research. 2016;55(14):4022-4034.
  2. Wang ZY, Georgakis C. New Dynamic Response Surface Methodology for Modeling Nonlinear Processes over Semi-infinite Time Horizons. Industrial & Engineering Chemistry Research. 2017;56(38):10770-10782.