Revolutionizing drug development: AI-driven solutions for poor solubility and bioavailability

Sanjay Konagurthu, PhD, Senior Director, Science and Innovation, Thermo Fisher Scientific

April 15, 2024 (12 minute read)

Progressing poorly soluble molecules from discovery to clinical trials stands as one of the most significant challenges to overcome in early drug development. Poor solubility generally corresponds with lower bioavailability, which can have a detrimental effect on the efficacy of the drug product. Artificial intelligence (AI) and machine learning (ML) methods continue to revolutionize the pharmaceutical industry in many ways, including the development of solutions to accurately predict the most effective combinations of solubility enhancement technologies and excipients for drug development.

The challenge of poor solubility in drug development

Some estimates suggest that the efficacy of more than 70-90% of molecules is challenged by solubility issues¹. Traditionally, resolving these issues was conducted via empirical trial-and-error methodologies with the goal of finding the best technology to ensure the drug's bioavailability. The solubility and eventual bioavailability of a molecule depend on various physico-chemical factors as well as the properties of the excipients and the formulations. These factors must be understood both individually and in combination. Despite the expertise of the scientists involved, identifying the most effective combinations is a complex and intricate process, requiring detailed scientific experimentation and analysis.

Employing traditional empirical methods poses substantial challenges to drug development timelines and budgets due to the extensive time and resources required for systematic experimentation, data collection, and analysis, which often involve lengthy clinical trials, complex laboratory procedures, and meticulous documentation processes.

The rise of AI/ML in drug development

The surge in interest and advancements in AI/ML technologies, coupled with an increase in regulatory submissions involving these methods (exceeding 100 in 2021), has led the US Food and Drug Administration (FDA) to establish multiple specialized groups. These groups are tasked with accumulating knowledge, providing support, assessing, and even driving innovation in AI/ML applications within the pharmaceutical sector². This trend is further underscored by the anticipated growth of the global AI market in drug discovery, which is projected to reach a value of $4.9 billion by 2028³.

The substantial increase in published literature over the past two decades highlights the expanding role of computational chemistry in the pharmaceutical industry. This growth underscores how predictive modeling and similar methods are revolutionizing formulation development. These techniques address long-standing industry challenges, particularly those related to solubility and bioavailability. Techniques like Quantum Mechanics and Molecular Dynamics (QM/MD) are used in this transformation. They offer precise modeling capabilities by providing deep insights into molecular-level interactions, such as ligand binding, structural mechanisms, free energy evaluation, and spectroscopic characterization. This approach is proving to be more reliable than the traditional methods of selecting optimal technology options and ideal excipients via experiments set up. Notably, it leads to a reduction in false positives, streamlining the development process⁴.

Innovation is driving computational methods to even more aspects of drug development with growing interest in its applications from lead generation to the prediction of clinical trial results to advanced pharmaceutical manufacturing.

AI/ML predictive modeling for drug development: Quadrant 2™

For formulation scientists contending with solubility challenges in their compounds, collaboration with a contract development and manufacturing organization (CDMO) specializing in computational drug development can be beneficial. An example of this is the use of AI/ML-based solutions, such as Thermo Fisher Scientific's Quadrant 2™ platform. This technology assists in early formulation development through in-silico modelling. Developers can input the molecular structure and known physico-chemical properties of their compound into the platform to generate predictive models.

Quadrant 2™ utilizes proprietary algorithms alongside quantum mechanics/molecular dynamics (QM/MD), quantitative structure–activity relationship (QSAR) models, and aspects of ADMET (absorption, distribution, metabolism, excretion, and toxicity) analysis to create customized predictions. By analyzing specific data of the compound, these algorithms can identify promising solubility enhancement techniques and suitable formulation design for enabling technologies (e.g. amorphous dispersions).

Following this, experts at Thermo Fisher review the generated insights and recommendations, compiling them into a detailed report that aligns with the sponsor's business and clinical objectives. This in silico approach to formulation development aims to reduce the time and resources typically expended on empirical, trial-and-error methods. It addresses the risk associated with potentially having to revise solubility enhancement approaches post-proof of concept, which could otherwise lead to significant delays and additional costs in the development process.

To date, the Quadrant 2™ platform has been applied to more than 400 molecules across multiple therapeutic areas and druggable space for small molecule oral delivery modalities. Validation studies indicate the accuracy of the platform's technology selection is greater than 90%, and its excipient selection is verified to exceed 80% accuracy. The growing preference for in-silico modeling over traditional trial-and-error methods is primarily driven by the potential for significant time and resource savings. Many companies initially adopt conventional approaches for resolving solubility challenges but as the time and budget costs accumulate, they look for guidance in transitioning to an approach that integrates AI/ML technology. Additionally, the reduction of risk and minimization of barriers in each phase of development are also critical benefits. Predictive modeling can even help developers reduce the amount of API required and support sustainability by reducing energy use and material waste.

The future of AI/ML in drug development

The future of AI/ML and its continued success rests on the generation of even more high-quality data. Large data sets that have been carefully cleaned, curated, and standardized and that represent the druggable formulation space will ensure even better performance of the predictive technologies and algorithms moving forward. The capabilities of quantum computing will only get larger and more complex, permitting modeling of even larger and more complex molecular systems. The incorporation of real-world data including clinical-trial results, patient outcomes, and safety surveillance data promises to further drive precision and accuracy of predictions.

In the future, quantum computers will likely play a role in even more aspects of drug development, further improving the speed of new API discovery and the accuracy of optimal predictions for formulations. These innovations represent great promise with respect to addressing the unmet needs of patients for whom effective treatments are not yet available. The sizable reduction in development costs with AI/ML may even represent lower drug costs for consumers, allowing or improving access for many⁵.

As with all new research spaces where there is a lot of excitement and a lot of innovators moving concurrently, AI/ML may benefit from widespread collaboration across stakeholders. Development of accuracy standards could support the advancement and adoption of these tools, protecting the reputation of and increasing confidence in AI/ML use across multiple roles.

Regulators will also need to establish ways to consistently evaluate submissions involving AI/ML. The FDA has announced their plans to “develop and adopt a flexible risk-based regulatory framework that promotes innovation [with respect to the use of AI/ML] and protects patient safety⁵. They acknowledge the opportunities and challenges associated with AI/ML and are currently seeking feedback from stakeholders on several issues related to its use in development and manufacturing of drugs.

Conclusion

The integration of AI/ML technologies in drug development, particularly in addressing solubility and bioavailability challenges, marks a significant paradigm shift in the pharmaceutical industry. These advanced computational methods, as exemplified by platforms like Quadrant 2™, are transforming the traditional resource-intensive trial-and-error processes into more efficient, accurate, and cost-effective strategies. By harnessing the power of predictive modeling, AI/ML tools are not only optimizing the selection of solubility enhancement technologies and excipients, they are also paving the way for a more streamlined and sustainable development process.

This revolution in drug formulation is poised to accelerate the journey of new drugs from the laboratory to the patient, potentially improving outcomes and accessibility. As the industry continues to evolve, embracing these technological advancements will be key to overcoming long-standing challenges and unlocking new possibilities in the quest for more effective and accessible treatments. The future of drug development, powered by AI and ML, holds immense promise for both the pharmaceutical industry and patients worldwide.

To speak with one of our experts about AI/ML-based approaches to solubility and bioavailability enhancement in drug development, contact us.

References

1. Loftsson T, Brewster ME. Pharmaceutical applications of cyclodextrins: basic science and product development. J Pharm Pharmacol. 2010 Nov;62(11):1607-21. doi: 10.1111/j.2042-7158.2010.01030.x. PMID: 21039545.
2. US Food and Drug Administration. Artificial Intelligence and Machine Learning (AI/ML) for Drug Development. 2023. Available at: https://www.fda.gov/science-research/science-and-research-special-topics/artificial-intelligence-and-machine-learning-aiml-drug-development
3. Artificial intelligence/AI in drug discovery market: Global forecast to 2028. Markets and Markets website. 2024. Available at https://www.marketsandmarkets.com/Market-Reports/ai-in-drug-discovery-market-151193446.html.
4. Kar RK. Benefits of hybrid QM/MM over traditional classical mechanics in pharmaceutical systems. Drug Discov Today. 2023;28(1):103374.
5. AI Drug Discoveries to Cut Costs and Save Lives: Medicine’s Next Big Thing? [Blog post]. University of Central Florida. Available at https://mse.ucf.edu/ai-drug-discoveries-to-cut-costs-and-save-lives-medicines-next-big-thing/