Rick Cooley Puts PAT in Perspective

Feb. 24, 2005
A pioneer offers his perspectives on PAT and where it’s heading and reminisces about his 23+ years leading Eli Lilly’s process analytics program. His advice: Remember that PAT won’t magically lead to headcount reduction; it isn’t always easy to do, and you can (and should) start small! Before starting a new position as North American manager of process analytics with the chromatography vendor Dionex, he shared his views with Editor in Chief Agnes Shanley.
By Agnes Shanley, Editor in ChiefAfter 32 years with Eli Lilly and over 23 years heading up that company’s process analytics effort, Rick Cooley will start a new position on March 1, as manager of the Process Analytics Center of Excellence for the chromatography equipment vendor, Dionex Corp. (Sunnyvale, Calif.).Mr. Cooley has been a key advisor to FDA and numerous industry and government groups, and founded the Pharmaceuticals Process Analytics Roundtable, to encourage the exchange of information on the subject. PharmaManufacturing.com asked him about his perspectives on process analytics and PAT, and where it’s heading.PM – How did you first get involved in applying process analytics to biopharmaceutical manufacturing?RC – It all began in 1981. Dr. Leroy Baker, who was then manager of Lilly’s bulk biosynthetic insulin manufacturing plant, realized that it would be extremely difficult to meet market demand for the product given the overall cycle time dictated mainly by off-line testing constraints. At that time, collecting fractions from process scale purification columns, sampling the fractions, and assaying the samples using off-line HPLC in the QC lab to decide which fractions were pure enough was a lengthy procedure.Management realized that it would be hard for us to meet demand given these lengthy cycle times.Dr. Baker, who is now retired, read avidly and kept up with what was going on in other industries, and how they were using process analytics. He’d read about work in the petrochemicals industry, applying online gas chromatography to distillation columns, and asked why we couldn’t do that with HPLC for pharmaceuticals.At that time, the process instrumentation community had just developed an HPLC system that could be used online. One vendor was developing a system for polymers, but it was still fairly crude and wouldn’t work in our system, so we chose equipment we were familiar with and packaged it in a way that would survive on the factory floor.After testing prototypes in our pilot plant, our team had success fairly quickly. Management wanted to move it to the manufacturing plant. They supported the effort as they saw it as a “major hope” for reducing cycle time.The original systems didn’t have a computer interface, so we taught operators to interpret the data from  a strip-chart recorder. We later automated the system, including the data integration, and interfaced the systems directly with the distributed control system (DCS) to provide automatic control of the purification columns.Dr. Baker moved from manufacturing management back to the lab and led a team that made some changes to the process that made it work much more efficiently. At this point, process intensification became the focus of our work. We saw an order of magnitude increase in throughput as a result of a combination of process changes that were enabled by the use of process analytics and process automation.PM – How did you organize staff to support the project?RC – We’d considered different approaches; whether to have the staff involved in this project work under the responsibility of QC, plant engineering, or whether to set up a whole new function.Since the most likely problems would be with the assay itself rather than the equipment, we decided to recruit individuals with HPLC experience from the lab, and make them “process analytics technicians.” We also redesigned the equipment to be modular, so that defective units could be rapidly replaced. After focusing on the problem, we launched a dedicated team. Today, there are approximately 16 people working on this team at Lilly that supports the biotech manufacturing areas in Indianapolis. The long-term success of process analytics in this area is totally due to the efforts of our process analytics technicians.Having a dedicated team was important. I’m an analytical chemist by training, and, as I’ve seen the field develop, process analytics tends to become something of a “stepchild” in pharmaceutical companies -- it could be part of development, technical services, engineering or automation.PM – Do you see engineers becoming more critical to the QC function in the future?RC – Success with process analytics requires a joint effort between analytical chemists and engineers. At Lilly, from the start, it was critical for the process analytical group to work very closely with process engineers on existing applications and to find new potential applications for process analytics that could help the engineers solve process problems. PM – What is holding pharmaceutical PAT back?RC – Pharmaceutical manufacturers are reluctant to make significant changes in existing processes. I can only comment about the process analytics portion of PAT, which has become a much broader initiative, but a lot of people tend to take a theoretical approach and talk about what PAT can do without having the actual hands-on experience. They often imply that PAT is simple to implement.This isn’t always the case. There are huge benefits to process analytics, but you need to have an adequate infrastructure in place, and the right people involved, in order to realize them.Some look at PAT as a headcount-reduction tool, and it’s not. And it’s not a panacea. Its goal is simply to reduce variability, cycle time and risks of producing off-spec product. We also wanted to increase production efficiency.You need the right people to support it and you need to have quality control involved much more closely with manufacturing, and to develop a good working relationship and ongoing dialog in order to support it.PM – Any advice for those starting to implement a pharmaceutical process analytics program?RC – Start small. Look for a very specific project with a very high probability of success. Use a simple technology. You hear a lot of people talk about online spectroscopy and chemometrics, but simple applications can be done with a fixed-wavelength instrument and an ND filter. For example, in protein purification, a process analytical system could be set up to improve diafiltration, using a fixed-wavelength UV sensor to measure protein concentration and a simple feedback control loop. It costs less than $10,000 for the sensor, the system is rugged, and it takes the variability out of that process step. We did that at Lilly. So, you don’t need a $100,000 multi-wavelength spectrometer to start a process analytics project.PM – How can technical staff get management’s support for PAT projects?RC – Management needs to see an ROI. They’re influenced by cost, and applications can gain their support if you can show that they will eliminate or delay a capital expenditure.Arguing that the project will increase process understanding is unlikely to sway them, since they’re not usually clear on what that means or convinced there is a significant need to increase it.PM – Where is PAT heading?RC – In 20 years, there may be significant use of process analytics for “real-time release,” but there are a lot of things we need to be able to measure better, such as trace components, before off-line testing can be totally eliminated. Today, the technology doesn’t really exist to do this extremely rapidly, although advances in MEMS (micro electro-mechanical systems) are likely to change the picture. I’m not sure that spectroscopy, by itself, will get down to that level.