On Monday, November 09, Abbott’s Director of QA, Stephen Tyler, presented on the challenges that Abbott has faced in applying PAT tools in a holistic process throughout the API product quality lifecycle. Tyler’s presentation was part of a session on using Process Analytical Technologies for PQLI (Product Quality Lifecycle Implementation) and Quality by Design efforts.
Tyler summarized Abbott’s efforts starting in the early formulation stages of an API. API development is never easy, Tyler said. “You start off with a path and you think you know where you’re going, but things change and you end up going off in a different direction.”
Given the unpredictable nature of API process development, applying PAT as early as possible, as part of a “holistic” approach, is critical to being able to handle changes throughout the lifecycle. There must be a focus on “fundamental understanding” of the API and its processes, Tyler emphasized, not just for process control but in order to enable later experimentation and modeling to anticipate and deal with process changes along the way.
Tyler detailed some of the work of Abbott formulation scientists to apply PAT to obtain optimum crystal morphology, perform concentration testing, etc., using Mid-IR and other basic PAT technologies. “We used very simple science through PAT to address concerns of our formulation scientists,” he said.
It wasn’t long that these early studies paid off as the API’s path diverged. “After we had done all this nice work, during product strategy review it was decided that this was going to go to a plant that could not use the original solvent.” The formulation and process data that had already been collected were instrumental in tweaking the process to account for the new solvent, which, Tyler said, “turned out to be far more robust than the original solvent.”
As the lifecycle progressed, Abbott opted to rely more on NIR, such as for drying optimization, or mass spec and Raman for monitoring de-solvation, as the team’s focus turned more towards optimization, yield, and cost.
By the time the API had moved on to the pilot plant, Abbott was able to not only robustly monitor processes, but also to mirror those processes by performing high-level simulation. “ For monitoring heat transfer in a filter dryer, for example, “we took our Raman data and our model data and compared them,” Tyler noted. “The actual correlation was quite good.”
“This states to us that if we have a fundamental understanding and have these models, in the long term we don’t necessarily have to make large investments in PAT. We have to invest in understanding our processes.”
Tyler also discussed work in API particle size feedback control. “Ultimately. . .it gives you the ability to fine-tune your process and deliver an API particle size distribution that is very consistent . . .”
Using PAT to optimize lifecycle development does not end once a product reaches the market, Tyler emphasized. “Once you pat yourself on the back, you’ve got to monitor . . . and make adjustments as you need. Your models need to be continuously assessed and tweaked to become more robust.”
The benefits of these lifecycle efforts include: no IP samples, RPM feedback control, online testing, continuous PSD data, and so on. But these benefits have not come without a strain on resources. “All of these things require a tremendous number of people . . . someone who pushes the concept but also people in manufacturing, R&D and so on. It can’t be done in silos.”
“It’s all about control strategy,” Tyler added. “It is a lifecycle and it starts in R&D and beginning to understand how you’re going to control your product. . . it has to be iterative. . . this continues on and on throughout the product lifecycle, all the way until you discontinue the product.”