There are few drug companies who haven’t experienced the pain of dumping millions of dollars and years of research into a compound that ultimately isn’t viable. Just because something works in a test tube or a test animal doesn’t mean it can be produced in thousand-ton quantities in 12 locations around the world. It’s clear that the design-test-manufacture pipeline needs greater streamlining. Considerations such as quality, safety, stability and sourcing should be built into the discovery and development cycle from day one. This is where in-silico techniques such as modeling and simulation can play an important role.
Today’s pharmaceutical and biotech companies are continually challenged to find new substances that have novel, targeted therapeutic effects in a very crowded and competitive market. The requirement for efficacy without adverse interactions is difficult to satisfy, and many of the simplest solutions have already been identified. This leaves very large molecules—with their inherent difficulties of synthesis, process-ability and solubility—or combination therapies—where dosage, metering, and delivery demand a finely tuned balance of life, chemical and materials science.
In such a complex environment, computational models are useful in helping organizations better understand all the variables that can impact a potential drug’s viability as a marketable and profitable product. What effect, for example, does crystallization have on stability? Which salts are chemically compatible with a new drug substance? Will a compound be most effectively delivered by tablet, gel, or injectable? Modeling and simulation can be deployed to predict both the upstream and downstream behavior of compounds and formulations, providing several significant advantages.
First, project teams can virtually design and screen lead drug candidates before actual and expensive live experimentation. This results in less time and expense in the lab and a greater success of the project. Second, fewer non-viable candidates will be pushed forward into development, potentially saving millions. Finally, greater insight into how pharmaceutical materials might behave on a large scale in the plant, allows for a more streamlined, less error-prone production process. What’s key is that in-silico techniques can be deployed to more closely integrate the discovery processes surrounding target and lead candidate selection with testing, scale-up and manufacturing.
Let’s look at the example of a multi-national pharmaceutical company. This organization used modeling and molecular simulation of the crystal morphology and crystal surface interactions of both the active pharmaceutical ingredient (API) and the excipients (e.g. binder, filler and disintegrant) to investigate their ability to compound, mill and form their product into direct compression tablets.1 Once they established models of the critical process attributes and effects, they were then able to set-up and quickly optimize their batch milling and compounding production processes.
Producing consistent powder that works for direct compression tablet manufacturing from intrinsically variable organic crystalline material for drugs involves many complex batch manufacturing processes. Machinery set-up, configuration, operating parameters and designing or determining the processes to achieve consistent particle size, flow and morphology requires major investments in research and manufacturing technology. The company developed then deployed several complex models to better predict how the crystal they were investigating might break, how much energy would be required, how the milling process parameters related to that energy and finally what milling operation technique would be best. This type of sophisticated integrated molecular modeling and multi-scale simulation offered great potential in reducing the time and effort spent in experiments, and in linking bulk processability more closely with early product design and development, helped the overall throughput of the development and scale up process.2
Several critical conditions need to be met for modeling and simulation to add value for most organizations. Central to this is that developing useful models requires good data. Excellent technologies exist for modeling atoms, molecules, polymers and crystals but it is the data that goes in to them that will ground the findings in reality. The key, therefore, is to be able to leverage data from multiple sources—chemical and biological databases, electronic lab notebooks, production systems, even publicly available information—in order to build the most accurate and complete predictive models of the system. But in a large, globally-distributed enterprise, data access and integration can prove challenging. This is because of the wide diversity of data formats (structural, outcomes, text, numeric, image-based, etc.), the number of instruments, their formats and data hierarchies and variety of equipment that needs to be accommodated (everything from microscopes to customized high throughput rigs), and the many different locations within the organization where data may be “hidden.”
Beyond data integration, it is equally important to be able to capture “best practice” development processes that are consistent, predictable and repeatable. Modeling can help organizations build, test and optimize the processes, but they also should be automated, simple to deploy and easy to adjust when the need arises.
Fortunately, technologies are now available that facilitate the “plug-and-play” web services-based integration of both processes and multiple sources of scientific data. An underlying, enterprise-level scientific informatics platform with these capabilities is highly useful in providing the data access, integration and process control that facilitates the broad use of in-silico techniques. Assisted by the right informatics and modeling and simulation technologies, pharmaceutical companies can increase efficiency and cost savings, as well as significantly optimize both discovery and development.
About the Author
Michael Doyle, PhD, is Director of Product Marketing and Principal Scientist at Accelrys. Michael received his PhD, MA and BA from University of Cambridge, U.K. After working with ICI Agrochemicals, BP Central Research, and BP Exploration, he joined Accelrys in 2001. His background encompasses modeling, simulation, statistics and informatics. He has special interests in pharmaceutical development, chemical catalysis, exploration informatics, and formulation tools.
1. Lee I, et al. Rotor-stator Milling of APIs - Empirical Scale-up Parameters and Theoretical Relationships Between the Morphology and Breakage of Crystals. APR. 2004; 7( 5):120-123.
2. Haware RV, et al. Anisotropic crystal deformation measurements determined using powder X-ray diffraction and a new in situ compression stage. Int J Pharm. 2011; Jun 17. [Epub ahead of print].