The Rising Complexity of APIs and the Challenges We Face
My passion for synthetic organic chemistry dates back to my early years as a PhD candidate and postdoc in academia. After a decade working in late- and early-stage process development, I’ve developed an eye and a feeling for reactions and their realistic outcome. While the paper can tolerate the wildest ideas about synthesizing a molecule, the reality is often less forgiving, especially if said molecule needs to be synthesized at a scale larger than a few grams at a time.
Over the past two decades, I have witnessed a dramatic evolution in the landscape of active pharmaceutical ingredient (API) and drug substance development. While the usual synthetic route for a target molecule used to comprise roughly eight steps 20 years ago, drug candidates today are much more complex and often require 20 or more synthetic steps. This surge in complexity has profound implications for our industry, affecting everything from timelines to supply chains and manufacturing costs.
The development of APIs has become increasingly intricate because of this heightened molecular complexity, alongside increasingly stringent regulatory demands. This trend has resulted in extended timelines for investigational new drug (IND) applications, as each synthetic step requires meticulous development and optimization. Creating robust process routes is crucial for ensuring a reliable supply chain for clinical trials and beyond and to avoid the time and cost burdens associated with second-generation process development.
To tackle these challenges, integrating retrosynthetic analysis with artificial intelligence (AI) and machine learning (ML) has become essential. Traditional methods often fall short, thanks to a lack of practical considerations, such as cost and supplier availability. AI-enhanced retrosynthetic tools offer a more holistic approach, providing feasible, scalable synthetic routes that consider real-world supply chain constraints from the outset.
Leveraging AI and Advanced Data Analytics
Route scouting involves identifying and evaluating the most efficient and scalable synthetic pathways for producing a target API. This process starts with a comprehensive literature review, followed by the design of multiple potential routes that are then tried out on a small scale in the lab. The goal is to optimize efficiency, yield, and scalability while ensuring safety and environmental sustainability, and providing a clear pathway from initial discovery to commercial production.
AI helps to identify optimal synthesis routes through advanced techniques such as molecular modelling, data mining, and ML. These technologies allow for in silico retrosynthetic analysis, exploration of vast reaction databases, and simulation of reaction mechanisms. AI-driven optimization considers multiple objectives like cost, yield, and environmental impact, continuously refining its predictions with new data.
However, utilizing AI and ML algorithms for route scouting does not represent an ultimate and final solution for drug developers. Proposing a viable synthetic route for commercially viable products needs to consider the cost of goods, supply chain limitations and challenges, and process scalability. Experts with years of experience in the field are best suited to provide these inputs.
Our AI-enabled Route Scouting Service merges AI-driven retrosynthetic analysis with the decades of combined experience of our subject matter experts. This integration ensures that proposed routes are not only feasible but also commercially viable and practical. Our offering also addresses key industry challenges by integrating real-world supply chain data and commercial constraints into the retrosynthetic analysis process. This enables us to identify synthesis routes that are theoretically viable and optimized for factors like cost, availability of raw materials, and scalability. By considering these practical aspects from the outset, our approach mitigates risks of supply chain disruptions, accelerates drug development timelines, increases economic efficiency, and promotes reliable commercial manufacturing of complex APIs.
Comprehensive supply chain intelligence allows the proactive establishment of primary, secondary, and tertiary supply options, mitigating risks of shortages and disruptions. Overall, this ‘right-first-time’ philosophy, which integrates practical commercial factors early on in route design, accelerates the path from discovery to commercial manufacturing cost-effectively by aligning with supply realities, leading to substantial cost savings and efficiency gains throughout the manufacturing process.
Lonza’s AI-enabled Route Scouting Service exemplifies how technological innovation can address the pharmaceutical industry’s pressing challenges. It avoids late-stage redesigns by identifying feasible and scalable routes early, accelerating development timelines and reducing costs. Combining AI tools and high-throughput experimentation (HTE) allows for rapid assessment and optimization of promising routes.
An Eye to the Future
Looking ahead, we are developing fully automated robotic systems capable of multistep synthesis under the oversight of our experts. This innovation aims to further support clients in early development stages, helping to expedite the journey from lead compound discovery to clinical trials.
As AI and cheminformatics capabilities continue to advance, I anticipate several future developments that could further revolutionize drug development and manufacturing processes. Predictive modelling could be expanded to optimize synthesis routes not just for efficiency and cost but for plant fit, utilizing existing assets and equipment to the fullest. AI-driven retrosynthetic analysis could integrate real-time production data to dynamically adjust routes based on current plant capacities, inventories, and operational constraints.
Additionally, machine learning models trained on vast datasets of reaction data could enable de novo design of novel synthetic routes and methodologies, opening up new realms of synthetic possibilities. Cheminformatics tools may provide deeper insights into impurity prediction and formation, enabling proactive mitigation strategies. Ultimately, the tight integration of AI, real-world data, and human expertise could usher in an era of self-optimizing pharmaceutical manufacturing – processes that can dynamically adapt and self-correct in response to disruptions while always identifying the ideal pathway from both a chemical and operational standpoint.
Overcoming the complexities of pharmaceutical manufacturing is a challenging endeavor. Through our combined expertise and the power of AI, we are pioneering new ways to tackle these challenges head-on. By integrating advanced technological solutions with deep industry knowledge, we are not only streamlining the drug development process but ensuring it is more efficient, cost-effective, and scalable. As we continue to innovate and evolve, we are excited to lead the way in transforming pharmaceutical manufacturing for the better.