September 15, 2006 (Vol. 26, No. 16)
Challenges Emanate for Technology Development and Design of Processes
Adoption of the FDA’s Process Analytical Technology (PAT) guidance and implementation of PAT throughout the pharma/biotech industry on the small molecule side has been quite good, “but progress in large molecular development has been much slower,” says Rick Cooley, manager of the process analytics center of excellence at Dionex (www.dionex.com). Cooley attributes this distinction primarily to the complexity of secondary, tertiary, and quaternary structures that complicate analysis and control strategies in the manufacturing of proteins and other large molecules.
In a sense, by accepting and encouraging the principles of PAT, the FDA “openly communicated to the industry that it recognizes it is part of the problem,” says Cooley. The FDA also recognizes the opportunities PAT has to offer, which are evident from the benefits achieved using PAT and other process optimization and efficiency strategies implemented in the past by, for example, the petrochemical industry.
Sandy Weinberg, senior director, fast-track biodefense, GE Healthcare (www.gehealthcare.com), anticipates that within three to five years more companies will be assessing where they are in terms of PAT, and he envisions the potential for dramatic changes in five to ten years.
The main challenge, Weinberg suggests, is how to apply process monitoring, “to go from looking at the product to looking at the process,” because improving the process will improve the product.
Ajaz Hussain, Ph.D., vp and global head of biopharmaceutical development, and former deputy director for science in the U.S. FDA’s Center for Drug Evaluation and Research, sees PAT as an important opportunity for the pharma/biotech industry.
“For biological processing, people have agreed that the process is the product,” says Dr. Hussain. “To some extent that is true, but unless you design your manufacturing process and control it effectively, you are locking your process in to 20-year-old technology.”
Historically, biopharma had a view of PAT as a sensor- or instrument-driven approach to process control with limited application to bioprocesses involving living organisms. However, PAT includes more than putting more sensors in a fermentation tank or process stream, explains Lou Bellafiore, president of TechniKrom (www.technikrom.com). It encompasses the implementation of manufacturing science to large-scale biological manufacturing and involves identifying and controlling sources of variability in these processes.
John Walker, vp of engineering at TechniKrom, emphasizes three key demands of PAT: “clear process understanding, competent instrumentation, and an appropriate control system that knows when and how to intervene to guarantee product quality.”
Investing in PAT
One of the key hurdles to broad adoption of PAT concepts and applications is the change in thinking and business strategy it requires. Companies must be convinced that PAT makes good business sense before they will adopt organization-wide changes that, from a process and regulatory perspective, will change the way they do business.
Although the potential benefits of implementing PAT in the long-term are generally recognized, the short-term uncertainty and risk, at least in these early days of the FDA guidance, boil down to how much time and resources a company should invest in a pharmaceutical product that faces an uncertain future in terms of clinical efficacy, regulatory approval, and commercial success.
“You need to strike a balance,” says Dr. Hussain. “At what point do you invest in process development and when do you have to say it does not make business sense to do so?”
In response to customer concerns about having to buy new equipment to benefit from PAT, Bellafiore explains that minimizing variability in a process has a direct effect on increasing yield, thereby reducing the overall cost of production. A full return on investment can be achieved in a few days of processing. Too often, “companies continue to use post-process QC to maintain quality, segregating out-of-spec product by fractionating the process stream and using QC to determine where the acceptable product is.” This approach results in variable, unpredictable yields. In addition, “the FDA wants QC to be used to confirm quality not to determine quality,” Bellafiore states.
Process Monitoring and Control
Dr. Hussain emphasizes the importance of having a set of “exquisite analytical technologies” ranging from protein chemistry to rapid assessment of microbiological control. Appropriate implementation of PAT offers a “broad-ranging set of opportunities that can reduce development risk and cost,” he concludes. Once a company learns how to apply the key fundamental types of analytical control technologies effectively to one process and product, “you can then translate the knowledge to others.”
Whereas the pharma industry has been accepting of new technology on the R&D side, “it lags behind on the manufacturing side for fear of delaying approval,” says Cooley. Companies are reluctant to introduce new technologies that far along in the product development process could be misinterpreted or misunderstood in the submission or during the pre-approval inspection.
Early applications of PAT in small molecule production borrowed technologies from the petrochemical industry and applied them mainly to final dosage formulation processes, such as solids-handling operations, including grinding, mixing, and compaction, explains Mel Koch, Ph.D., director of the Center for Process Analytical Chemistry at the University of Washington, Seattle.
In biotech, PAT interest has been focused more on understanding fermentation and recombinant protein production and has involved sampling the product stream and acquiring new kinds of data and looking at side reactions and byproducts. At present, companies are mostly considering utilizing PAT to show that their products are okay using existing processes, rather than redesigning and optimizing their processes, contends Dr. Koch.
Realizing the broader benefits of PAT, which will be achieved by analyzing concentration and composition of active products and unwanted side products and looking at uptake of nutrients, will take some time. We are “still waiting for breakthrough technologies that will be applicable in bioreactors,” Dr. Koch adds.
When and how to apply PAT should be determined on an application by application basis, in the view of Anurag Rathore, associate director of process development at Amgen (www.amgen.com). “A PAT-like scheme is not applicable in all cases—only when real-time data offers benefits,” he says.
If, for example, a process step is not under any time constraints, does not represent a potential bottleneck, does not consume costly reagents and resources, and does not pose a risk of contamination or introduction of impurities, then there may be little justification for investing in the monitoring, control, and optimization of that particular process step.
In contrast, consider the manufacturing of a protein that, if allowed to sit in solution for too long, becomes unstable or begins to aggregate, or can lead to the formation of other impurities. In this scenario, it could be beneficial to monitor the process and stop it as soon as the desired product has been formed and before significant impurity formation can occur.
TechniKrom has focused on implementing PAT in critical liquid handling/blending steps, such as buffer supply, pH adjustment, LC gradient elution, and solvent feeds. The company helps clients monitor process operations and demonstrate that variability in feeds leads to variability in production yields.
TechniKrom presented to the FDA its Adaptive PAT™ methods for real-time process monitoring and control that allow for process adjustments to be made, based on real-time analyses of liquid feedstocks.
TechniKrom recently received a U.S. patent protecting its approach for applying adaptive PAT to improve virtually any liquid handling process. The company isolated the blending component of its instrumentation and made an adaptive PAT version available as an upgrade module, enabling companies to continue using their existing equipment. The module functions between the buffer feed and the main system. “We are able to achieve a reduction of variability that is up to 50 times better than current approaches,” says Bellafiore.
Rathore describes two potential applications of PAT in biotech unit operations: diafiltration and chromatography. In both cases it is feasible to design control schemes and analytical tools that enable real-time testing and real-time (versus retrospective) decision-making. These control schemes require higher level competency from manufacturing operators, more robust analytical methods, and an improved understanding of the process.
One example of a promising application for PAT would be in online or at-line analysis of eluent from chromatography steps to determine when to start/stop pooling fractions. PAT can allow for more consistent product quality and maximum product recovery. Optimizing this type of process, however, will require the development of better online HPLC tools capable of more rapid analysis and data acquisition.
Process Analytical Tools
The optimal approach, in Dr. Hussain’s view, is to introduce new technologies—process analytical and control technologies—as they become available and validated, and to build the concepts of PAT into the design of manufacturing processes for new products upfront, as they are being developed.
Many process analytical tools have been available for some time, such as basic online HPLC technology with automated fraction collection and rapid microbial detection methods. Dr. Hussain describes “tremendous improvement” during the past three to four years in research on biosensors, “including genomics-based biosensors.” He expects many new types of sensor technology to become available within the next five years.
Process analyzers are only one part of the picture, according to Cooley, who emphasizes the need for a systems approach that incorporates data management and feedback and feed-forward control measures. It is important “to monitor (or model and predict) and control critical quality attributes in real-time,” he says.
Many technologies are commercially available, including online HPLC and mass spectrometry, and the issue is mainly on the design end—building the currently available tools and techniques into the process.
Weinberg emphasizes the need to design flexible versus fixed processes, with flexible parameters. For example, rather than designating a static temperature for a particular process, identify an optimal temperature range that allows for adjustments depending on the fluctuation of other process parameters. Weinberg also stresses the importance of building robustness into instruments and designing them to monitor and self-control their own operations.
Another challenge in implementing PAT is the need for organization-wide cooperation and integration. Teams from different parts of a company have to work together to design and adopt systems that will work across the organization.
Based on his involvement in developing the FDA’s PAT guidance, Dr. Hussain explains that the agency’s intention was not to dictate how companies should implement PAT, but rather to create a flexible regulatory process that would involve regular meetings with regulators, at which time companies could present and discuss individual strategies and innovative approaches.
“In some sense, regulatory uncertainty can only be reduced through consultation with the FDA,” says Dr. Hussain. That is how the FDA avoided a more prescriptive guidance, which would have hampered innovation.
“Ultimately, the cost associated with manufacturing is reflected in how efficient the manufacturing process is, and the efficiency of the manufacturing process depends on how well you understand the process and are able to control it,” says Dr. Hussain. Whenever process variables deviate from the specifications, the result will be the loss of one or several batches and an extensive investigation process, which all adds up to very long cycle times. Designing in PAT upfront can offer substantial, long-lasting benefits for optimizing productivity across an organization.