Gain Statistical Control of Your Processes

One area in which the pharmaceutical industry has been lacking in its approach to PAT and process improvement is in statistical control. In this exclusive interview, Philippe Cini, managing consultant for Tunnell Consulting, offers his thoughts on how the industry can achieve effective quantitative process characterization.

The first step toward a successful PAT program is quantitative process understanding, says Tunnell Consulting’s Philippe Cini, one of that firm’s leading PAT gurus. spoke with Cini recently to ask about how the pharmaceutical industry can follow in the footsteps of the chemical industry and achieve greater statistical control.

We refer to Cini’s fascinating presentation from this year’s Interphex show in New York. To access Cini’s in-depth and self-explanatory PowerPoint slides, click the “Download Now” button below.

P.M.: Why does pharma lag behind in understanding its processes and PAT?.

Cini: There are several reasons. First, there is a historical reason. In the pharmaceutical industry, typically R&D and development have been staffed by pharmacists and what we call pharmaceutic scientists that have a good knowledge of material science and empirical knowledge of solid processing, but that body of knowledge has not been cross-fertilized with other bodies of knowledge, such as chemical engineering.

Also, we have not had a handshake yet with polymer science, a sister discipline to chemical engineering. And we haven’t had a handshake with statistics—far from it. That handshake has existed in the chemical industry from 20 years.

Another reason is the regulatory burden. We have to lock in the process at the time we submit the NDA. When we do the scale-up, we do those infamous three validation batches and have to do them consistent with the parameters and constraints of the NDA. We file the NDA at a time when the process is typically ill-defined and ill-understood.

Then when we launch the product and identify ways to improve the process, we are reluctant to refile. The FDA could reject the new filing. We’ve been denied the freedom to do continuous improvement. . . Now, FDA wants to be part of the solution.

The other reason is that sales and marketing, and clinical R&D, have driven the industry. Manufacturing and process development have been looked at as a necessary evil. They have never been looked at as an enabler or strategic partner. So executives have been innovation- and risk-averse.

P.M.: What’s the first step towards better process characterization?

Cini: To characterize the variation in the process with basic tools, such as control charts, Cpk, and bivariate analysis. These are the simple ones. They are very intuitive.

P.M.: Those simple tools are part of three levels, or “screens,” that you discussed for process understanding, with the ultimate screen being advanced quantitative tools such as regression analysis. Why do you use the term screens?

Cini: :The way we gain understanding of a process is by identifying, or having a brainstorm if you will, on the different ways of having variation in a process. We may start with 80 or 100 possible drivers of variation, and we need to distill it down to just a few. That’s why I talk about screens. We need to be able to sort the likely and unlikely sources of variation by taking them through those screens.

P.M.: How does the industry approach advanced quantitative control?

Cini: These statistics, the basic and advanced tools, are used to varying degrees by pharmaceutical companies, but not systematically. Also, statistics should not be in isolation. When you combine statistics with formulation science and engineering knowledge, you start really seeing the whole process.

If you look at the chemical industry, companies have entire departments of statisticians. When I started to work for Shell, I was a young scientist fresh out of school. They told me, you’re not going to do experiments without a statistician. That’s a true commitment to statistics.

P.M.: Is that why you say that teams are an integral part of process understanding?

Cini: Absolutely. When we work with our pharmaceutical clients, we create cross-functional teams which contain a statistician, a process engineer, and a formulation scientist. They work together in a collaborative fashion every day. That’s when you start seeing the sparks. That’s when it really happens.

Click the Download Now button below to access Cini's PowerPoint presentation from Interphex 2005.

Show Comments
Hide Comments

Join the discussion

We welcome your thoughtful comments.
All comments will display your user name.

Want to participate in the discussion?

Register for free

Log in for complete access.


No one has commented on this page yet.

RSS feed for comments on this page | RSS feed for all comments