In early phases of pharmaceutical process development, crystallization process design often faces challenges in limited materials and uncertain chemical, physical, and/or mechanical properties, such as organic and inorganic impurities, particle morphology, particle size distribution, hardness, flowability, etc. The probability of occurrence and severity of those challenges may not be fully understood and requires
ad hoc assessment in both drug substance and drug product manufacturing. In this work, a potential risk identified was the level of a class 2 residual solvent in an early crystallization process that uses antisolvent addition followed by cooling. Several strategies were investigated to mitigate this risk, including (1) an additional recrystallization step in alcohol-water solvent system, (2) using an alcohol as a co-solvent, (3) seed type (e.g., un-milled vs jet-milled seeds), (4) seeding point supersaturation level, (5) seed aging time, and (6) overall supersaturation control. It is important to point out that these diverse strategies also come with additional challenges. For example, the use of a co-solvent introduced a solvent-mediated crystal morphology change from rod block to hexagonal plate [1]. Using jet-milled seeds resulted in fine and agglomerated particles, and the supersaturation control with un-milled seeds could introduce secondary nucleation (causing broad particle size distribution) and lengthy antisolvent addition time. It is found that jet-milled seeds together with well-controlled supersaturation at low levels significantly reduced the residual class 2 solvent content consistently in the isolated product. Additionally, mathematical modeling and simulation tools were also explored to understand the attainable region of particle size distributions [2], probability of jet-milled seed dissolving, etc. This work highlights the importance of continuous monitoring of the processing risk throughout the development phases is vital to develop robust commercializable processes.
References
[1] Turner TD, Dawson N, Edwards M, Pickering JH, Hammond RB, Docherty R, Roberts KJ. A digital mechanistic workflow for predicting solvent-mediated crystal morphology: The α and β forms of L-glutamic acid. Crystal Growth & Design. 2022, 22: 3042-3059.
[2] Vetter T, Burcham CL, Doherty MF. Designing robust crystallization processes in the presence of parameter uncertainty using attainable regions. Industrial & Engineering Chemistry Research. 2015, 54(42): 10350-10363.
