Speaker Abstract: S-23
Fundamental Optimization of Pharmaceutical Products and Processes
Fernando Muzzio, Ph.D., Rutgers University
"Optimization", in its current usage in the pharmaceutical industry, is an empirical process; it typically consists of performing a small number of experiments within a narrow range of choices, and then selecting the value of the parameter (the blending time, the supplier of an excipient, etc) that gives "the best" results. Oftentimes, obeying to time constraints and a paucity of data, the decision is made without rigorous statistical analysis. "Engineering optimization" is a very different exercise, typically involving the development of a model (either ab initio or statistical) relating process outcome to critical parameters, and then performing a systematic search of the parameter space to find the absolute best possible outcome consistent with process constraints. While it is clear that it would be highly desirable to bring "pharmaceutical optimization" closer to "engineering optimization", it is important to recognize that a model is required, which is often unavailable in pharmaceutical product and process development. Some tools are readily available to breach this gap at least in part. In almost any case, rigorous design of experiments (DOE) methods can be used to minimize experimental effort while maximizing significance of conclusions. In many processes, models can indeed be developed or even imported from other industries (example, fluidized bed operations). Some processes are indeed quite challenging from a modeling point of view.
However, if the goal is to develop rigorous optimization strategies leading to highly robust and controllable processes, then development of accurate models is a must. However, for this to be possible, the FDA also needs to examine its own role and regulatory practices. Models are, in essence, amenable to continuous refinement. Optimization is a dynamic, moving target. Thus, process changes must be not only facilitated, but encouraged.
In this talk, the use of different modeling approaches in an integrated PAT strategy will be presented. Typical "pharmaceutical optimization" methods will be discussed first. Subsequently, advantages and limitations of DOE methods will be presented. Finally, the development of fundamental models and their use for process optimization and automated control (presumably the final goal of the PAT initiative) will be discussed in detail.
2004 FDA Science Forum | FDA Chapter, Sigma Xi | CFSAN | FDA
Last updated on 2004-MAY-28 by frf