Pharma's Process Analytical Technology

The Food & Drug Administration's recent approval of process analytical technology (PAT) for use in the pharmaceutical industry is being hailed by drug manufacturers because it encourages process optimization based on process understanding. Now, analytical chemists in pharma will be able to implement advanced analytical frameworks that have resulted in improved quality in other industries. Chemists who are trained in PAT and who are able to make the transition from traditional analytical lab-based monitoring have a distinct advantage in employment. As PAT evolves in the pharmaceutical industry, the skills that are considered desirable for analytical chemists are likely to evolve as well.

PAT allows for the design and control of manufacturing through timely analysis of process parameters that are critical for ensuring an acceptable end point. FDA believes that PAT will encourage the development of new and efficient tools for pharmaceutical development, manufacturing, and quality assurance resulting in increased product quality and reduced costs for the industry.

Prior to PAT's approval by FDA, quality analysis in pharmaceutical manufacturing was carried out by testing intermediates and final products. Products that failed to meet specifications were rejected. Under PAT, the focus is on both the process and the end product. The goal is to continually generate real-time data to increase the level of certainty throughout the process.

According to data from the American Chemical Society Office of Member Information, the pharmaceutical industry is the largest employer of analytical chemists. The U.S. Bureau of Labor Statistics is projecting that overall employment of chemists is expected to grow anywhere from 10 to 20% through 2012, with most of that growth concentrated in pharmaceutical and medical manufacturing and in scientific research and development services firms. Although those forecasts are conditional, analytical chemists are nonetheless expected to play a significant role as PAT implementation evolves, according to the analytical community in general as well as people interviewed by C&EN. Given the range of tasks that analytical chemists currently perform in the pharmaceutical industry, it is likely that PAT will offer them new and expanding opportunities in the future.

PROCESS ANALYTICAL measurements can be taken one of three ways: at-line, where the sample is removed and analyzed close to the process stream; on-line, where the sample is diverted from the manufacturing process to an analyzer and possibly returned to the stream; and in-line, an invasive or noninvasive process that analyzes the sample while it's part of the process stream.

PAT has been used extensively in other industries such as petrochemicals for years, but the pharmaceutical industry has lagged behind largely because of perceived regulatory barriers.

"PAT is a formalized way of looking at things that many pharmaceutical manufacturers have been doing informally," according to Elizabeth Fowler, vice president of quality and regulatory affairs at Xcellerex, a Marlborough, Mass., provider of contract services for process development and manufacturing for biopharmaceuticals. "It's also making sure that you use the data you collect and analyze them to increase process understanding and improve process control and efficiency. PAT guidance encourages innovation to bring more control into the process and makes it easier to implement new techniques for monitoring approved processes."

What led FDA to realize that PAT was worthwhile, Fowler says, "is a desire to ensure that the agency is doing what it can to protect public health through ensuring product quality and to bring down the cost of drugs. The manufacturers' perception has been that FDA approvals were based on paperwork, not on science, and all the paper in the world doesn't make a high-quality product. Product failures--especially ones that might have been caught early on--are more expensive to fix afterward than they are during the manufacturing process."

Prior to FDA's support, PAT in the pharmaceutical industry was conducted on a "don't ask, don't tell" basis because of the regulated environment. Many analytical chemists currently using PAT have either picked it up on the job or transitioned from another industry that uses it.

Andrew J. Lange, senior principal scientist at Pfizer in San Diego, says: "The people I know are either engineers or they started as analytical chemists who morphed into process analytical scientists. There is also a subset of people working in pharmaceuticals who have been recruited from the chemical industry. The majority of analytical chemists have never set foot inside a manufacturing facility; what they do is get samples from the factory to analyze, but the sample is then out of its native environment. In process analysis, you're making measurements in situ, and there are a host of considerations specific to that environment that you don't consider in the laboratory. The advantage is you're looking at the material as the process is going on."

Lange began as an analytical chemist in the technical operations area, as opposed to the quality operations area, which measures samples in the lab. "Technical operations involved process support, process start-up, and troubleshooting," he says. "I was the sole analytical chemist in technical operations, and the rest were mostly organic chemists and engineers. My transition into the process analytical realm was straightforward as a result."

John M. Wasylyk, principal scientist in analytical R&D at Bristol-Myers Squibb, New Brunswick, N.J., has been working in PAT over the past decade. "We've been doing in-line analysis for years, simply in terms of pressure monitoring and turbidity--simple techniques that have been around for years," he says. "The analytical chemist either works with people who are doing PAT or wears two hats running both traditional methods and new technology. There are spectroscopists who use and are versed in PAT techniques, but the techniques are only as good as the primary methods behind them, such as high-performance liquid chromatography, gas chromatography, or UV analysis."

Steve Doherty, who works in process analytical measurement technology for Eli Lilly in Indianapolis, says that 13 people in his group perform PAT. The four senior scientists in the group all came from the commodities chemicals industry. "It's fairly common," he says. "Right at the time that companies are trying to grow process analytical functions, the downturn in the pharmaceutical climate makes it hard to justify personnel increases, so most of the hiring is done by word of mouth."

"By bringing pharmaceutical manufacturing to a state in which we understand the processes very well, we can control the critical process parameters, which must be within a particular range for the final product to meet its quality attributes," Fowler says. "By bringing everything in the process to its desired state, you can improve product uniformity and quality by making the process much more consistent."

According to Rick E. Cooley, a team leader in the manufacturing science and technology division of Eli Lilly: "Think of process analytics as taking the lab and moving it next to where the sample originates. The traditional way is to take the samples to a remote lab for processing and analysis. In PAT, you take the analysis and connect it directly to the process, eliminating the transmission of the sample from the process to the lab.

"People have been doing feedback process control for years in pharmaceuticals," Cooley says. "But what they've been measuring isn't based on the critical quality attributes of the product. If I want to mill an API [active pharmaceutical ingredient] to a certain particle size, then I need to run the mill at a certain rpm. The control may be on maintaining the mill at that speed. Under PAT, I measure the critical quality attribute and change the process to achieve that. You validate a process under a certain set of operating parameters, then try to control the operations parameters with the assumption that you'll get a quality product out of the process. If there is a variation in the equipment or the raw materials, the variability shows up in the end product. Now, FDA is asking us to measure and predict the attribute and to change the process to keep it under control."

Bristol-Myers Squibb and other companies use multiple probes in reaction vessels in the lab prior to pilot plant and manufacturing, Wasylyk says. "With PAT we can get closer to the reaction and the processes in the manufacturing, pilot plant, and developing stage," he adds. "Are there side products, by-products, or degradents that form? How do we improve the quality of our product? You use an FTIR probe, a Raman probe, an NIR probe, and a probe to monitor crystal growth. Hopefully, one of the tools or all four will provide some knowledge. In the reactions we run, a spectroscopy tool won't provide a big advantage for rapid robust reactions that go correctly and work quickly. However, the next reaction or process may be long, arduous, and hazardous, so those are the ones that you want to put PAT to the test."

The relationship between FDA and the pharmaceutical industry is changing under PAT. For example, FDA has set up "safe harbor" allowances for industry to implement PAT on existing processes and avoid regulatory repercussions. Also, to increase the likelihood that New Drug Applications are accepted, "FDA is bending over backward to help us," Wasylyk says. "They toured various pharmaceutical manufacturing and pilot plants in order to learn about these techniques. We are encouraged to contact them prior to filing about the tools we're using or the data we have collected in order to formulate and communicate our implementation plans."

CHEMOMETRICS is another essential tool in PAT implementation; specifically, statistical knowledge and an understanding of how the chemometric packages work are essential. "We push for the process chemists to learn experimental design, how to perform multivariate analysis, and how to change multiple parameters to shorten the experimental time you have in order to come up with the best chemical process," Wasylyk says.

Says Fowler: "The interactions in even a single step in a process can be very complex, and there are any number of different variables that could be adjusted. Multivariate analysis will tell you which variable has a major effect on the inconsistency you're seeing and which factors interact. You need to explore the effect of various conditions on obtaining the desired result, and from those data you can derive the critical process parameters. If a critical parameter doesn't stay within the range, it alters the desired product quality."

Understanding scientifically the multivariate relationships and applying this knowledge in different scenarios is beneficial. "Being able to do pattern recognition is more powerful in terms of speed as opposed to having a perfect analytical assay," Cooley says. "In manufacturing, being precise is more important than being accurate. You can adjust for a bias as long as you understand it. You should be able to answer the question, Is this the same thing I produced yesterday, and is it within the limits of what I need to make? This is where chemometrics and those tools come in to tell you whether what you are making is within the limits of what you define as good."

There are questions over how to handle the vast amounts of data generated by the process analyzers. The amount requires not only a data management system that can handle the volume and types of data generated but also one that raises questions of procedures, questions that FDA is looking to industry to solve. "Most large companies have automation functions," Doherty says. "In the regulated manufacturing industries where you keep the raw data for the process, how much do you archive? Do you need two wavelengths to demonstrate the completion of the run? Do you need to save all the spectra?"

A few academic centers focus specifically on training in process analytical chemistry, such as the Center for Process Analytical Chemistry (CPAC) at the University of Washington, Seattle, and the Measurement & Control Engineering Center at the University of Tennessee, Knoxville. But in general, during their academic training, students have little exposure to industry's requirements and needs compared with the time that's devoted to teaching theory.

"The challenge is for academics to train people to develop these skill sets that are going to be needed," CPAC Director Melvin V. Koch says. "The analytical chemist is going to have to learn more than traditional analytical science because PAT is going to force people to develop technologies that provide the answers."

Koch, who had an industrial career before becoming director of CPAC (C&EN, May 17, 2004, page 40), emphasizes a multidisciplinary and multi-industrial approach. "We're working across 10 academic departments and with 30 industrial partners. If you have graduate students who are trained to work on a team with engineers and data-handling experts, they are absorbed [into the workforce] and become productive quickly. Other industries have been doing PAT for a long time, and they have been able to train people internally. The brain drain from traditional industry into pharmaceuticals is a short-term solution. Some universities I've dealt with are asking whether they even need an analytical department, rather than how to broaden its impact to affect programs like PAT."

Analytical chemists are but one component of a large PAT team. "Most people I know in PAT have exceptional people skills," Doherty says. "To solve problems with process analyzers, you need management support and team skills because you can't be the lone specialist in the ivory tower. You need to be articulate and be able to influence people. We're starting to hear more about ROI [return on investment], long-term cost of ownership, and economic efficiencies, and we need to be able to talk in those terms with production personnel to sell our technology."

Cooley recommends other skills, such as "manufacturing experience, an understanding of process unit operations, and a basic understanding of chemical engineering principles and processing equipment. Multiple areas of expertise are useful: engineering, automation, information technology, finance, and regulatory experience; all those areas in manufacturing are driven by ROI."

The most successful chemists, however, are those who subscribe to continuous learning. "We got into science because we like to learn and try new technologies and applications even though we have advanced degrees," Wasylyk says. "We have not given up on learning. A number of us have taken extra courses on chemometrics, for example." He adds that he's also learned those jack-of-all-trade skills that come in handy, such as splicing fiber-optic cable in case a cable breaks at 3 AM.

What will help future recruitment is FDA's support for PAT. "PAT is sexy now. Historically, I had difficulty recruiting people before FDA promoted PAT," Cooley says. "You're taking off the lab coat and putting on a hard hat, but people didn't see the job as being very glamorous because it wasn't getting any recognition. FDA has legitimized process analytical, and it's considered a good career move because now senior management is paying attention."