ai-assisted-design-of-stable-liquid-formulations
Is the path to BLA for complex biologics too slow and unpredictable with traditional formulation? AI and machine learning are transforming how we design stable liquid formulations. Learn how to accelerate development with a predictive, data-centric model.
Menu
The Changing Blueprint: AI-Assisted Design of Stable Liquid Formulations
FAQ
Current Situation
Typical Market Trends
Current Challenges and How They Are Solved
How Leukocare Can Support These Challenges
Value Provided to Customers
The Changing Blueprint: AI-Assisted Design of Stable Liquid Formulations
If you're a Director of CMC, the path to a Biologics License Application (BLA) is full of challenges. You're aiming for a drug product that's stable, effective, and easy to make. With today's complex molecules, the journey feels more unpredictable than ever. Traditional formulation development, which relies on iterative screening, is slow and uses up precious material. It's a process that often feels more reactive than strategic.
A shift is happening. Early adoption of artificial intelligence and machine learning is beginning to change how we approach formulation, moving us toward a more predictive and data-centric model.
Current Situation
We need to shorten development timelines. During the COVID-19 pandemic, CMC timelines for monoclonal antibodies were cut from a typical 12–15 months to about 3 months in some cases [1]. While that pace isn't the norm, it reset expectations across the industry. We're now expected to move faster, but the molecules we're working with are increasingly complex. Advanced therapies, such as viral vectors, antibody-drug conjugates (ADCs), and RNA-based medicines, present unique stability issues that don't fit neatly into standard approaches. A formulation strategy that works for a standard mAb might be entirely unsuitable for a gene therapy vector, which is sensitive to everything from pH shifts to the choice of stabilizers [2, 3].
Typical Market Trends
The biopharmaceutical market is expanding, with the global advanced therapy medicinal products (ATMPs) market expected to grow significantly [2]. One forecast predicts the market will reach USD 26.5 billion by 2031, up from USD 10.8 billion in 2024. This growth is driven by investment in cell and gene therapies and a rising demand for personalized medicines [4, 5].
This trend has two major implications for CMC teams [4, 5]. First, the sheer diversity of new modalities means formulation can no longer be a one-size-fits-all process. Second, many of these innovative therapies are coming from small or virtual biotech firms. These companies often operate without in-house labs and rely heavily on outsourcing partners for their CMC strategy, making clear communication and robust data transfer essential.
Current Challenges and How They Are Solved
Formulation development is still a big bottleneck. The central challenge is stability. Biologics are sensitive to their environment; factors like temperature, pH, and agitation can cause proteins to aggregate or degrade, harming safety and efficacy [6]. For new modalities, these problems are magnified.
The conventional approach to solving this is experimental, using design of experiments (DoE) to screen a wide range of excipients and buffer conditions [3]. This method is resource-intensive, needing a lot of drug substance and time [7]. For early-stage companies where every milligram of material is valuable, this is a big problem. A failed formulation can lead to delays in clinical trials and add significant costs.
Computational tools are changing how this works. Artificial intelligence, and specifically machine learning, helps predict formulation outcomes with greater accuracy and speed [8]. By analyzing large datasets, AI models can identify patterns that link specific formulation components to stability outcomes. This allows development teams to focus on fewer, more promising conditions for testing, saving both time and material [10, 11, 9]. Regulatory bodies like the FDA and EMA are also adapting, recognizing AI's potential in manufacturing and creating guidelines for its use [12, 13].
How Leukocare Can Support These Challenges
This is where a focused partnership can make a difference. At Leukocare, we use a combination of our proprietary Smart Formulation Platform and AI-based stability prediction to address these specific challenges head-on [14, 15, 16]. Our approach is designed to provide more than just data; it's about delivering a clear, strategic path forward.
For a fast-track biotech leader under pressure from their board, our platform can accelerate the journey to a robust, commercial-ready formulation. By using predictive modeling, we tailor formulation design to aggressive timelines, minimizing the risk of missteps. We act as a strategic co-pilot, giving you the scientific expertise and regulatory know-how needed to move forward with confidence.
For a mid-size biotech or a large pharma company working on a new modality, we offer specialized support. We can take on the complex formulation challenges that challenge internal teams, such as ensuring lyostability or working with novel vectors. Our process is transparent, designed to support in-house drug product teams, not to take over. We start with a pilot project to demonstrate value, delivering reliable, data-driven results that build trust.
Value Provided to Customers
The value is in creating a more predictable and efficient path to the clinic and the market. By integrating AI-assisted design early in the process, we help our partners achieve several key outcomes:
Speed: A data-driven approach means a faster, cleaner path to BLA. We help you reach your goals sooner with a formulation designed for regulatory success.
Reduced Risk: Better predictions mean fewer failures. By de-risking the formulation process, we help protect your investment and ensure your clinical timelines stay on track.
Confidence: Our process provides the structure and substance needed to build a strong CMC case for investors and regulators. It's about making informed decisions backed by solid data.
The goal is to move formulation from a source of uncertainty to a strategic advantage, giving CMC teams the tools and the partnership they need to succeed.
FAQ
1. How does AI-assisted formulation actually work?
AI-assisted formulation uses machine learning algorithms to analyze data from previous formulation studies. The models identify relationships between formulation ingredients (like buffers and excipients), their concentrations, and the resulting stability of the drug product. This allows the system to predict which new formulations are most likely to be successful for a specific molecule, narrowing down the experimental work needed.
2. Is this a "black box" system, or can we understand the reasoning?
Transparency is important. While some AI models can be complex, the goal is always to provide a clear rationale for the recommended formulation designs. Our approach combines the predictive power of the models with the deep formulation experience of our scientists. We work collaboratively to ensure you understand the "why" behind the recommendations.
3. How much material is needed compared to traditional methods?
A significant advantage of this approach is the reduction in the amount of active pharmaceutical ingredient (API) required. By using predictive modeling to focus on the most promising formulation candidates, we can greatly reduce the scope of initial experimental screening, saving precious material for other development activities.
4. How does this integrate with our existing CMC workflow?
Our approach is designed to be flexible and collaborative. We can act as an extension of your team, taking on specific formulation challenges, or work as a strategic partner to guide the overall formulation strategy. We provide structured data and clear documentation that aligns with investor and regulatory needs, fitting smoothly into your existing quality systems and CMC milestones.
5. What is the typical timeline for a formulation project using this approach?
While every project is unique, AI-assisted design can significantly shorten the formulation development phase. By reducing the number of iterations and experiments, we can often arrive at an optimized, stable liquid formulation more quickly than with traditional screening methods alone, helping to keep demanding development timelines on track.
