Striking a balance between meeting regulatory standards and the need for expeditious results poses a perpetual challenge for analytical scientists, especially in extractables and leachables (E&L) testing. One possible way to overcome the cost and time efficiency hurdles is to outsource the testing to predictive computer algorithms. The integration of computational modeling into E&L testing challenges traditional approaches and offers a glimpse into the future of pharmaceutical development. To unravel the complexities of this evolving landscape, we spoke with Jason Creasy, Managing Director & Principal Consultant at Maven E&L Ltd. His insights shed light on the challenges, opportunities, and delicate balance between tradition and innovation within the realm of E&L testing.
Mathematical Modeling and Simulation in E&L Testing
There are various types of modeling and simulation used throughout E&L testing. For example, Creasy explains that empirical models, based on prior experience, are often deployed for the assessment of leachable risk. These can help analysts judge how factors such as temperature and surface area might impact the production of leachables. The models can be used to predict high-risk systems and are particularly useful during the screening stages of drug development.
But there is also potential for modeling to be highly useful at other stages of development and testing. “Once you go through the effort of developing a model, you can roll it out rapidly time and again, and get some very quick predictions,” advises Creasy. “On the other hand, if you're relying on analytical testing, you've obviously got to go and find a laboratory, set up the equipment, get hold of samples, and so on, and it all takes time and money.”He goes on to explain that the right information could greatly speed up the process of material selection. “Once you've got a model, it's seconds to produce the result instead of months. You would be able to root out unsuitable materials very quickly.”
Creasy notes that this area is currently very under-explored within E&L testing, although it has proven popular in other fields. “You see the food industry using computational models quite extensively for modeling of migration. And in the medical device world—which is closely aligned with the pharma industry—there are people like David Saylor (a member of the FDA), promoting computational models for medical device use. There's no reason why the same models don't apply to the pharma industry. It’s a perfectly acceptable approach.”
Barriers to using modeling in E&L testing
So what are the key factors stymying the progress of modeling within E&L testing? “I think the main reason for the lack of progress is that most work is performed to try to meet a regulatory expectation of some kind. And certainly, in the pharma world of extractables and leachables, the regulators seem fairly set on seeing data produced from experimentation. They are less used to seeing any computation or equivalent. And because the pharmaceutical industry is very conservative, there hasn’t been much effort put into the development of mathematical models that could replace those experimental studies.”
Creasy also points out the cost barriers to mathematical modeling. “There’s a lot of upfront work involved to develop predictions on factors such as diffusion coefficients, solubility in the extracting media, and the equilibriums that might be set up on the interface between the material and the formulation. There’s a lot of complexity that might arise from the variety of different systems that you would have to model.”
He notes that these two barriers go hand in hand. If it was likely that regulators would accept models, efforts would be put in place to develop them. “As it stands, you get the impression that regulators would push back harder on a model than they would on the analytical data, but not necessarily for any other reason than it's different to what they've seen before.”
The Future Outlook for Modeling in E&L Testing
When envisioning the future of E&L testing through modeling, Creasy's insights suggest a cautious but inevitable progression. He emphasizes the initial challenge of creating a model and securing regulatory acceptance. "The first step is coming up with a model, and then you’ve got to demonstrate that model is appropriate and get the regulator to agree with you.”
An obvious question as we move forward with modeling is the role of machine learning, but this highlights yet another challenge. According to Creasy, a significant hurdle to model development lies in the need for extensive datasets for effective and reliable machine learning. “We would need for much more sharing of data to occur in order to build up those types of machine learning algorithms." Creasy acknowledges the collaborative efforts of organizations such as the Product Quality Research Institute (PQRI) but highlights the pharmaceutical industry's inherent secrecy as a major obstacle. Indeed, he contends that there is no practical reason against data-sharing, attributing the reluctance to industry culture.
While he does not anticipate the obsolescence of analytical methods in E&L testing, Creasy does envision a scenario where testing requirements could be significantly reduced. “If we were truly able to collaborate and develop models, you could do something once and then share it, and that would be that job done.” This would drastically cut down on duplication of efforts—there is a finite set of materials in use within the industry, and the list isn’t very long.
As regulatory bodies refine their expectations, ensuring compliance becomes a dynamic task, requiring constant adaptation and vigilance. In the face of these challenges, the scientific community is tasked with not only navigating the intricacies of E&L testing but also innovating solutions that optimize cost, time, and regulatory adherence. Investment in and regulatory acceptance of the integration of computational modeling promises to challenge conventions and possibly help reshape the future of drug development.
This article is featured in our May 2024 publication, Pharmaceutical Purity and Precision. Find out what’s happening in the world of pharmaceutical impurity analysis and learn about the latest topics and techniques.
