Speciality Chemicals Magazine MAY / JUN 2025 | Page 33

PHARMACEUTICALS

Identifying stable forms

A similar, integrated approach has been applied to the identification of the thermodynamically stable, or the preferred form of a molecule. The approach combines experimental data, informatics and energetic methods to de-risk candidate selection making use of the data contained within the CSD to perform a‘ health check’. 11

This approach is particularly useful and I have used it in a CDMO setting to enable candidate selection between lead API candidates that were almost identical in structure. In this instance a robust set of experimental polymorphism data and single crystal structural data was in place.

One molecule demonstrated a particularly stable hydrogen bonding network and melting point some 25 ° C higher than the identified stable form of lead 2. Structural interrogation via the health check demonstrated that whilst experimentally a more stable form had not been found, the risk of this appearing for lead 2 was significant against the stable form identified for the other. Given that the in vitro and PK profiles for each were almost identical, lead 1 was selected for progression.

One of the critical benefits of an integrated development path vs. purely in silico is access to material.

As a synthetic process is optimised the impurity profile changes, and not always subtly. Early batches can be profiled with only a few mg cost in terms of spent API. This work is best accomplished using material of very high purity. Understanding form change throughout development reduces the risk of failure at a later stage when more is at stake from a production perspective.

The impact impurity can play on the form fate of a molecule is well documented and an important consideration when progressing from early phase development toward commercial launch. The case of Ritonavor is a well-known example of the commercial damage a change in form can have upon a drug product. This is an ideal example of what the integration of traditional and AI based development should help to mitigate against.

Future of drug development

Small molecules continue to play a pivotal role in the supply of effective medicines to an ageing population. Their complexity in terms of structure and material behaviour makes development a continual challenge.

Despite advances in the application of AI and ML during the discovery and hit-to-lead process, challenging molecules are still emerging. The trend for increases in molecular weight and the number of hydrogen bond donors and acceptors in the search for increased specificity and potency remains, increasing the likelihood of poor physicochemical properties.

Those molecules that sit within DCS Classes II and IV are of particular significance. However, integrating development teams and deriving a phase appropriate strategy, plus making use in silico and AI technology can provide a streamlined and risk-mitigating journey from the early phases to the clinic.

Of particular interest is the increase in the accuracy of AI and ML in the pre-formulation development space. Over 20 years ago, predicting a polymorphic landscape from 2D and 3D structural data seemed out of reach. Today, this is a reality that is being used across various industries.

ML and AI will continue to evolve and integrate in all areas of pharmaceutical development as technology and the quality and volume of data improves. Whether it evolves to a state where human interaction is limited remains unknown. ●

MAY / JUN 2025 SPECCHEMONLINE. COM

33