Can predictive modeling replace years of drug stability testing? This case study explores how researchers of Janssen Pharmaceuticals, TU Dortmund University, and amofor teamed up to tackle one of the most persistent challenges in ASD formulation: accurately predicting shelf life without waiting years for empirical data.
The team developed a physics-based approach that combines long-term experimental data with crystallization kinetics to deliver reliable stability forecasts. Below, we guide you through the entire approach… step-by-step.
Why Standard Shelf Life Predictions Fail
Pharmaceutical companies face significant financial and time constraints when conducting multi-year stability studies. To predict shelf life, they traditionally rely on chemical degradation models like the Arrhenius equation. While these models are effective for tracking degradation of molecules, they don’t address the major failure mechanism associated with Amorphous Solid Dispersions (ASDs): crystallization.
- Chemical degradation occurs due to unwanted breaking of covalent bonds, molecular changes that always occur and follow predictable kinetics. It’s similar to nuclear decay in a power plant: atoms degrade at defined rates, governed by Arrhenius-type behavior. This makes chemical degradation relatively straightforward to model.
- ASD crystallization, in contrast, is a physical phase transition that may or may not occur. At some conditions, the system might be thermodynamically stable and never crystallize, yielding an effectively infinite shelf life. At other conditions, nucleation starts, and crystallization progresses. This process is influenced by many interrelated factors – molecular mobility, nucleation rate, crystal growth, temperature, humidity, and drug-polymer interactions – all of which are highly sensitive to formulation and environment. Crystallization doesn’t follow simple kinetic laws, making it far less predictable than chemical degradation.
Faced with this clear need for a more accurate and specific predictive approach, our interdisciplinary research team from amofor, Janssen, and TU Dortmund set out to address the problem through physics-based modeling. We designed a comprehensive two-year study with nifedipine and celecoxib – two well known model compounds known for their rapid crystallization behavior. By storing formulations under tightly controlled humidity and temperature conditions, we precisely tracked crystallization over time.
What Makes This Study Different
What sets this study apart is the depth of the stability data. Formulations were stored for up to 1000 days – far longer than the typical 3, 6 or 12 month intervals used in standard stability testing. Unlike conventional studies, this one tracked how quickly crystallization progressed over time, not just whether it occurred.
This granularity allowed the team to develop a uniquely powerful predictive model. Key takeaways:
- We developed a predictive model that accurately forecasts long-term shelf life. The model integrates key parameters that can affect crystallization: Temperature, humidity, glass transition, drug load, and nucleation propensity. The combination of all these parameters cannot be captured that easily with traditional degradation models.
- Temperature alone is not a reliable predictor. Phase diagrams and molecular mobility must also be considered to accurately forecast crystallization.
- While standard stability tests give only pass/fail data at fixed time points, this model enables real-time, continuous prediction of crystallization behavior.
- Water sorption might significantly accelerate crystallization, particularly at high humidity levels.
- Nifedipine ASDs crystallized much faster than celecoxib ASDs, highlighting the need for drug-specific modeling.
Inside the Model: Step-by-Step Process
-
Data Collection: This is the experimental input that links the real-world data to our simulations
– Solubilities in five organic solvents
– 1 DSC measurement of the API
– Crystallization information about the API (either qualitatively via the glass-forming ability class or quantitative re-crystallization data) -
Model Development
– Incorporated core parameters: temperature, humidity, drug load, glass transition, and nucleation propensity.
– Construct phase diagrams to map the stability regions of ASDs, identifying conditions where formulations remain stable or undergo crystallization. This step was crucial in refining predictive accuracy and guiding formulation adjustments.
– Predict molecular mobility, influence of thermal history of the glass, Tg, water sorption
– Option: Validation of the model with the first stability data. -
Predictive Simulations:
– Run simulations across a range of drug loads and environmental conditions.
– Identify conditions where ASDs remain stable, allowing formulators to extend shelf life without lengthy storage studies.
– A-priori Identification of formulations that will survive a certain stress test
– Screen the role of different polymers on stability
Impact and Advantages
This research proves that long-term stability studies can be dramatically accelerated using robust physical simulations. Instead of waiting months or years, scientists can now simulate ASD stability behavior across a wide range of conditions – including temperature, humidity, and drug load.
The model developed goes well beyond traditional degradation models. It accounts for crucial variables like drug load, nucleation propensity, and glass transition – factors that conventional models overlook. This makes it especially powerful for modern ASD development.
Try It with Your Own Data
Pharma teams often start by validating innovative techniques with their own compounds. We offer trial access to our software Solcalc so that teams can evaluate predictive performance using internal data. This real-world testing helps build a strong business case for adoption.
Whether you’re exploring a new formulation or looking to solve an old stability challenge, this case-study-based trial effectively demonstrates the model’s value.
Shift from reactive testing to proactive modeling!
Contact us today to explore how this model empowers formulation scientists to design smarter, faster, and with better outcomes.