Last month, at “PAT for Pharma and Biotech,” a conference hosted by the Institute for International Research, Richard Poska, director of global pharmaceutical regulatory affairs, CMC, for Illinois-based Abbott Laboratories, discussed how the cost of implementing PAT technologies might be minimized using existing in-house technologies. Among the points he stressed is the need for “continuous learning” about pharmaceutical processes.To gain a stable foundation from which to build a strong PAT program, companies need to understand their processes. In order to do this, Poska said, they need to generate the data required through modified design of experiment techniques. They must then demonstrate that the process controls all variability using statistical tools such as statistical process control and process capability analysis. Poska stressed the dangers of embracing technology for its own sake, interspersing a slide from “2001: A Space Odyssey” to make his point, and concluded with the warning that “good science should never be set to autopilot.” He also pondered what PAT’s final goal should be.PharmaManufacturing.com spoke with Poska to get his perspectives:PM — Is PAT per se always needed if a company has strong DoE, SPC, process capability and other programs established?RP — I believe that design of experiments, statistical process control and other programs actually complement the PAT initiative, as PAT provides a better understanding (and, it is hoped, control!) of our processes. If we are able to attain a robust process through carefully designed experiments, the need for PAT technologies becomes less critical as the process will have its variability minimized.PM — What are the biggest mistakes that pharma companies make when considering PAT?RP — In my opinion, companies are too willing to jump into a new technology before they really understand their processes. I usually use the example of the automation craze of the 1990s — I believe you really need to do your homework on your product and processes to understand them before you implement a PAT technology in production. To be clear, that is not to say that PAT technologies shouldn't be an integral part of the development process. PM — Cost data indicate that PAT can be quite expensive . . . does it always have to be this way? Are there less expensive options using existing instrumentation at-line?RP — Implementing new PAT technologies can be quite expensive as was pointed out earlier this year by Joep Timmermans of Pfizer [who estimated $100,000 for basic requirements and $200,000 in implementation costs such as networking and process control]. I would suggest that companies take a hard look at what they're already measuring and how.Tablet-monitoring systems are a good example of current technology that has widespread use. These systems meet most PAT technology criteria in that they sample throughout the batch and provide control and feedback mechanisms to control weight, thickness and hardness, which are inter-related by measuring the force used to compress a tablet. If the correlations haven't been made, it should only take a minimal amount of effort to establish them and have a package to negotiate with regulators as to the need to perform these measurements as a regulatory specification.PM — Many people from pharma companies complain that they can't get management support for PAT. What will be needed? What should they use to make a business case?RP — I believe a good business case is needed to convince management that process understanding and PAT technologies will be a good investment. I asked an audience at a recent PAT workshop how many were wearing watches, and then how many wore analog (dial), and how many wore digital watches. I then asked why those using analog watches weren't using digital watches as they were a newer technology.The answer lies in understanding how change will make things better, because new technology doesn't always give you "better" data. It was interesting to see these same folks’ faces when I asked why any of them use watches at all, as their cell phones provide the time, too!PM — FDA has suggested that PAT function as a cradle-to-grave tool for product and process optimization, from R&D on. Where do you see PAT fitting in?RP — As part of FDA's 21st Century Initiatives, both the PAT Guidance and the white paper on Innovation and Improvement offer suggestions for improving industry’s process understanding and control. I believe many companies are using PAT techniques in development in a way to better understand their new products. I believe that this is the right approach as it becomes more difficult to change formulations or processes once they are approved by regulatory agencies. The hurdles to change not only come from the FDA (which is committed to minimizing the requirements), but also from management, which will be looking at the bottom-line benefits of implementing these technologies on a production level. Industry and FDA both need to come to agreement what our long term goal should be—a variable process that is well characterized and measured or a robust process, or both.