Process development of API crystallization: a model-driven approach

About

We are excited to spotlight a recent study titled "Integration of a model-driven workflow into an industrial pharmaceutical facility: supporting process development of API crystallisation". The study was authored by a research team from CMAC Future Manufacturing Research Hub (Glasgow, UK),  Pfizer (Groton, Connecticut, USA) and Pfizer (Sandwich, UK) [1].

In this study, the Crystalline instrument was used to optimize the crystallization process for (3S,5R)-3-(aminomethyl)-5-methyl-octanoic acid (PD-299685), a potential treatment for menopause-related hot flashes. Small-scale experiments were conducted to efficiently collect critical data on solubility, kinetic parameters, and nucleation processes. This optimized approach addresses the resource-intensive nature of crystallization in pharmaceutical manufacturing, ensuring effective process development and yielding high-quality API suitable for large-scale production.

Experiment Procedure

Small-scale crystallization experiments were conducted using Crystalline, which includes process analytical technology (PAT) for real-time monitoring, illustrated in Figure 1.

• Solubility Studies: Using the Crystalline platform, solubility measurements were performed across various solvent systems with the polythermal method. The optimal solvent mixture, 55:45 water/1-propanol, provided a steep solubility curve, ideal for achieving high yields through cooling crystallization.
• Kinetic Parameter Study: Nucleation and growth rates were optimized by varying supersaturation and temperature using Crystalline with a 3-blade pitched impeller. This setup produced well-formed crystals and allowed for precise tuning of conditions to achieve optimal crystallization.
• Seeded and Antisolvent Experiments: Water was added as an antisolvent, significantly reducing solubility and increasing yield. The optimized process produced crystals with a d(v,90) of 234 µm and an aspect ratio of 0.766, ideal for pharmaceutical use.
• Process Design: The data from Crystalline experiments informed the design of a scalable crystallization process. The final method achieved a 99% yield, a d90 of 759 µm, and an aspect ratio of 0.718, demonstrating the process's scalability and suitability for industrial production.

Results

The Crystalline platform proved highly effective in collecting kinetic and thermodynamic data, which informed the solvent selection and crystallization parameters. The chosen system, a mixture of 55:45 water/1-propanol, provided optimal solubility and growth characteristics, with subsequent antisolvent addition enhancing yield and reducing particle size. In the kinetic studies, the images revealed the differences in particle formation when using various stirring mechanisms, such as the 3-blade pitched impeller, which produced well-formed crystals with consistent shapes and sizes, as shown in Figure 2. The workflow enabled the design of a hybrid crystallization process that achieved high yields and desired particle properties (aspect ratio and d(v,90) values).

Conclusion

Using Crystalline in a model-driven workflow reduced material and time requirements while providing robust process data for PD-299685 crystallization. The workflow facilitated successful scale-up, improving process efficiency and product quality, underscoring the value of advanced PAT tools like Crystalline in pharmaceutical crystallization development​.

References

[1] The experiments presented in this article were carried by the following team of researchers:

• Thomas Pickles, Cameron J. Brown and Alastair J. Florence (CMAC Future Manufacturing Research Hub, Technology and Innovation Centre, The University of Strathclyde, Glasgow, G1 1RD, UK)
• Vaclav Svoboda (Chemical Research and Development, Worldwide Research and Development, Pfizer, Groton, Connecticut, USA)
• Ivan Marziano (Pfizer R&D UK Limited, Ramsgate Road, Sandwich CT13 9NJ, UK)

The full paper is available here: https://pubs.rsc.org/en/content/articlehtml/2024/ce/d4ce00358f