December 2004

Volume 7, Issue 12

Table of Contents





	Article Indexes
	Author Indexes

PAT Pending

Government legislation is prompting drug companies to reevaluate their manufacturing processes.

Randall C. Willis

Download PDF | Table of Contents

Bayer For all of the articles published over the years highlighting the rapid pace of technology innovation in drug discovery, the pharmaceutical industry still lags most other automated industries in one critical area: manufacturing quality control.

In basic R&D, pharmaceutical discovery embraces a variety of automated and integrated technologies covering a spectrum of analytical methods, including chromatography and spectrometry, and various forms of in vitro and in vivo testing systems. By comparison, technical innovation in the industry’s manufacturing sector has moved at a snail’s pace, with many methods technicians use for process analysis remaining largely unchanged for the past three or more decades.

To address this problem, and ensure the safety and efficacy of the multitudes of medicines produced annually in the United States, the FDA assembled a subcommittee in 2001 to look into establishing guidelines for the use of quality testing methods. These methods comprise the process analytical technologies, or PATs (1). By August 2003, the committee had developed draft guidance for PAT implementation in the drug industry that it suggests “will encourage the voluntary development and implementation of innovative pharmaceutical manufacturing and quality assurance.”

PAT by number

Designed to improve efficiencies in both manufacturing and regulatory processes through an integrated approach to quality analysis, the PAT initiative hinges on four main components: data analysis, process analytical tools, process monitoring, and continuous feedback.

Quality control. A technician tests samples at Aventis’s Brindisi, Italy, manufacturing site.
Credit: Aventis

The formulation of a drug involves complex coordination of a variety of physical, chemical, and biological products and processes. Thus, any practical efficiency improvements will have to incorporate a knowledge base that contains a scientific understanding of how these variables interrelate and a means of applying this knowledge to different formulation scenarios. In other words, design-of-experiment (DOE) evaluations will be critical to the success of PAT.

By analyzing current methodologies and results in a systematic and statistical manner, drug manufacturers should be able to build a catalog or database of practical experiences that will enable them to simulate and test new processes. This informatics-based approach should allow companies to identify the variables most critical to the final desired product and its behavior, decide where they need to insert controls into the process, and discover the factors that control sample degradation.

Regular inspections. Testing is routine in Bayer’s antibiotic production.
Credit: Bayer

For example, scientists at Sigma-Aldrich Biotechnology recently used a DOE approach to develop a cell culture medium optimized for a variety of Chinese hamster ovary (CHO) cell lines, which biopharmaceutical firms use to produce protein-based biologics (2). The researchers used statistical software to identify the best-performing culture media in their arsenal and develop methods to further increase cell growth and productivity.

“By using DOE software, it is possible to find one or several best-fit media for a particular CHO clone,” the company’s scientists say. “One unique feature of this program is that specified criteria can be assigned importance values based on the user’s needs and desired outcomes. Another major benefit of this method for data analysis is that multiple criteria can be analyzed together to determine synergistic responses between the criteria.”

By reducing the need for trial and error, FDA regulators expect that this type of exercise will allow companies to more easily improve the manufacturing process and thereby reduce product development times.

Analytical PAT-ois

The PAT document defines “analysis” broadly, encompassing not only the chemical analysis of a substance but also physical, microbiological, statistical, and risk analyses. Thus, companies have been left looking for new ways to study new problems or new applications for existing analytical instruments.

In May 2003, at Purdue University’s Advancing Manufacturing Summit, Eli Lilly researchers presented some ideas about the analytical methods that could be adapted for real-time analysis of steps in the pharmaceutical manufacturing process (3). These methods included:

Fourier transform infrared (FTIR) spectroscopy for reaction analysis;
near-IR (NIR) spectroscopy to measure product dryness and uniformity;
HPLC, GC, NMR spectrometry, and MS for reaction analysis and product identity; and
ultrasound to measure sample granularity.

Drawing a manufacturing line

Analytical pharmaceutical testing has traditionally occurred either off-line, in a test laboratory, or at-line, in the production area. Both methods, however, suffer significant drawbacks that can seriously disrupt the testing process and raise questions about the results.

With the need to collect and move samples from one area of the manufacturing plant to another, off-line testing can result in significant delays between sample taking and results analysis. Thus, a company runs the risk that by the time a quality control manager has determined there is something wrong on the production line, large quantities of the errant product have been produced.

A simple solution is to move the analytical process closer to the production line. Thus, in at-line testing, a technician removes samples from the production stream using an instrument, such as a sample thief (a multiport collection device), and quickly tests the samples at a nearby workstation. Although the time lag is minimized using at-line sampling and analysis, this process can still introduce physical artifacts and increases the potential for product contamination.

Recognizing these problems, the PAT guidance recommends that companies supplement their processes with other sampling methods. For example, by adding on-line techniques, the line technician can divert small samples from the main product stream for testing and then return the samples to the production line. Similarly, with in-line testing, rather than bringing a sample to an instrument, the technician brings the instrument to the sample, by inserting a probe into the process stream. This method reduces the likelihood of contamination, but it relies on using remote probes, which may not always be technically feasible or available.

Perhaps the ultimate solution for testing along a production line, however, is the use of noninvasive methods, where a sensor (e.g., a Raman detector) never physically contacts the process stream and thus leaves it undisturbed. Because these methods largely eliminate the chance of contamination and minimize the potential response times between detecting a problem and acting upon it, PAT legislators are heavily promoting the development and implementation of noninvasive techniques, and these efforts are being recognized by the analytical instrumentation industry.

A challenge for a number of these approaches, however, is the desire by PAT regulators for companies to move much of their analytical methodologies closer to the production line, relying less on off-line or at-line testing (see box, “Drawing a manufacturing line”). This means moving the analytical instrumentation ever closer to the samples as they are produced.

For example, it is difficult, if not impossible, for technicians to test products by HPLC without extensive sample preparation, which often means moving the samples out of the production area and into a lab. Recently, however, scientist Jonas Johansson and Uppsala University Associate Professor Staffan Folestad, both working at AstraZeneca’s PAT Centre of Excellence, described how new Raman spectroscopy instrumentation is making extensive inroads into the pharmaceutical manufacturing sector (4).

The main advantage of Raman, they argue, is that it is a highly selective method that allows researchers to easily and accurately determine the active pharmaceutical ingredient (API) content of a formulation while largely ignoring the physical parameters of the samples or sampling conditions. As an example, they cite a study in which aspirin tablets were assayed for both the API and the main degradation product, salicylic acid. The study compared the results obtained using Raman spectroscopy with those achieved using more standard HPLC protocols, and found a good correlation between the two methods.

Similarly, as predicted by the Lilly researchers, NIR use has been increasing in PAT-style testing and has been particularly useful in on-line analysis. Because technicians can take multiple measurements with a single NIR instrument, they are able to follow a variety of processes in a chemical reaction simultaneously—such as changes in reactant concentration, byproduct formation, and product generation—which further allows them to make minor adjustments to the process on the fly.

Likewise, strong differences in the NIR spectra of various polymorphs of a single API allow scientists to quantify polymorph formation during both preformulation and formulation processes. This ability is critical to a drug’s success, because subtle changes in form can result in significant differences in drug behavior in vivo.

A good real-time

Of course, methods to monitor and analyze the production stream are of limited use if companies can’t respond rapidly to problems as they arise. To address this issue, the PAT guidance recommends that companies develop process-measurement systems with real- or near-real-time analysis of each critical product attribute. Furthermore, the document suggests that manufacturers establish process control mechanisms to allow them to make production adjustments in response to the analytical results.

In using real-time monitoring, the process end points no longer have to be fixed time periods (e.g., mix for 10 min), but rather can be the time required to reach a specific state or condition (e.g., mix until sample humidity drops below 5%). At the same time, even when companies achieve real-time monitoring, FDA regulators still expect them to maintain established processing times, although a time window or spread might be more appropriate when using these methods. The thinking behind this is that even if a sample is dry within 5 min when it usually takes 20–25 min, there might still be a problem with the process.

The advantages to real-time monitoring are best explained in a recent article by Pfizer Central Research scientist Perry Hailey, who describes applying NIR to pharmaceutical blending analysis (5). He suggests that one could argue “that stable blending processes do not require real-time monitoring, but taking this viewpoint would be to miss the overall objective of real-time measurements: that of parametric release.”

“If, for example, the blend was monitored in real time and shown to be within specification, then the measurement taken during the blending stage could contribute toward the release of the final drug product on a weight basis only,” he explains.

Standing PAT

But perhaps the most critical variables to the success of the PAT initiative will be continuous learning and information management by the companies. It will be important for manufacturers to understand the significance of each data point and how best to use this information to justify proposals for postapproval process changes. Similarly, the assembly and dissemination of this information will be critical both within the company and with the FDA as part of the drug approval process.

“The lack of timely cost-effective data availability with connectivity to the point of use may be the single largest hurdle to compliance and operating efficiencies in pharmaceutical manufacturing today,” believes Justin Neway, founder, executive vice president, and chief scientific officer of Aegis Analytical Corp (6). “PAT must be part of a comprehensive manufacturing enterprise-wide solution that ensures relevant time data availability, and must provide a validated environment for data-intensive decision-making.”

One industry concern about integrating PAT into manufacturing processes is the projected costs associated with restructuring production lines to accommodate new product and data analysis tools. Although this is of less concern for the larger international manufacturers, who are nonetheless feeling the pinch of accommodating many countries’ manufacturing regulations, PAT will have a significant impact on small pharmaceutical and biotechnology firms that are already struggling with tight or nonexistent margins.

The FDA argues, however, that corporate managers should take a longer-term view toward implementation. They suggest that the savings gained from more efficient resource usage, optimal machinery operation, reduced waste, and a greatly reduced risk of product recall more than outweigh the direct costs of implementing PAT. Aegis Analytical’s Neway agrees.

“Pharmaceutical manufacturers have continued to generously allocate funds to discovery and marketing while allocating insufficient funds to manufacturing,” he argues. “Without strong manufacturing operations, many of the new drugs will produce less revenue than their full potential as a result of longer-than-necessary process start-up and scale-up times, too many lost batches, process instability and quality problems, and fines and recalls.”

Perhaps the most persuasive argument for introducing PAT to the production process is a sense of inevitability. Numerous analysts have commented that regulatory guidelines have a tendency to become laws, and it is probably better for companies to try to stay ahead of the curve by taking a proactive stance.

References

PAT—A Framework for Innovative Pharmaceutical Development, Manufacturing, and Quality Assurance; www.fda.gov/cder/.
An Efficient Approach to Cell Culture Medium Optimization—A Statistical Method to Medium Mixing; www.statease.com/pubs/cellculture.pdf.
Technology Tools for Developing Improved Pharmaceutical Manufacturing Processes. Presented at the Advancing Manufacturing Summit, West Lafayette, IN, May 20, 2004; www.purdue.edu/amap/ppt_03/HUFF.ppt.
Folestad, S.; Johansson, J. Eur. Pharm. Rev. 2003, 8 (4), 36–42.
The Role of NIR Spectroscopy in the Measurement of Pharmaceutical Manufacture; www.brimrose.com/.
Neway, J. O. Pharm. Technol. Oct 2003, pp 46–52.

About the Author

Randall C. Willis is a freelance writer from Toronto and former senior associate editor of Modern Drug Discovery. He can be reached at rnlwillis@rogers.com.