viscosity-prediction-models-for-bispecific-antibodies
High viscosity is a major roadblock in bispecific antibody development, complicating injectability and manufacturing. Discover how advanced viscosity prediction models are transforming formulation, helping to de-risk development and accelerate timelines. Read on to learn more.
Menu
Tackling the Viscosity Challenge in Bispecific Antibody Formulation
Frequently Asked Questions (FAQ)
Current Situation
Typical Market Trends
Current Challenges and How They Are Solved
Tackling the Viscosity Challenge in Bispecific Antibody Formulation
How predictive models are helping de-risk development and accelerate timelines for complex biologics.
Bispecific antibodies are a big step forward in medicine, offering new ways to treat complex diseases like cancer and autoimmune disorders.[1] The global market for these therapies is expanding fast, expected to reach over USD 110 billion by 2030.[1] Unlike traditional monoclonal antibodies, bispecifics can bind to two different targets at the same time.[3] This dual-targeting opens the door to new treatments, but it also brings unique challenges for making and formulating them.[4, 5]
One tough problem, especially for drug product development leaders, is dealing with high viscosity. To give a bispecific antibody conveniently, often as a shot under the skin, it needs to be highly concentrated.[6, 8] But at concentrations over 100 mg/mL, these complex proteins can get thick and hard to inject, a problem that can stop a promising candidate in its tracks.[6, 8]
Current Situation
As a Director in CMC or Drug Product Development, you know the pressure of getting a candidate from discovery to the clinic. High viscosity is a common roadblock. It doesn't just affect how a drug can be delivered, but it can also make them hard to manufacture, maybe causing problems like membrane clogging during filtration.[6, 8] The complex structures of bispecific antibodies often lead to strong protein-protein interactions, which is the main reason for high viscosity.[9] These interactions are tough to predict just from the molecule's structure, making formulation a tough, often trial-and-error process.
Typical Market Trends
The industry is looking for ways to get ahead of viscosity problems. The usual way involves lots of experimental screening. This means making the molecule, which is often in short supply early on, and testing it with many different formulation buffers and excipients. This takes a lot of time and resources.
A newer trend is using computer tools and predictive models.[10, 18] By using computational methods, teams can check candidates and formulations virtually before doing lab work. Machine learning, in particular, has become a powerful tool.[11, 16, 17] These models are trained on large datasets of antibody sequences and their measured properties to predict which molecules are likely to become too viscous at high concentrations.[11, 13, 16, 17] This move towards data-driven formulation is helping teams make better decisions earlier in development.
Current Challenges and How They Are Solved
The main challenge is how complex bispecific antibodies are. Their unique shapes and surface properties can lead to unexpected interactions that drive viscosity.[14] Predicting these interactions accurately is the main goal.
Here’s how the industry is handling this:
Early Stage Assessment: Teams are including "developability" checks more and more early in discovery. This means checking not just how well a molecule binds to its target, but also how easy it is to make and how stable it is.[15, 24]
Machine Learning Models: Scientists are developing smart machine learning models that analyze a molecule's sequence and structure to predict how it will act. These models can spot specific features, like charge distribution or hydrophobic patches on the antibody's surface, that are linked to high viscosity.[11, 16, 17] Some models were very accurate, correctly predicting whether a molecule's viscosity would be above a certain threshold with up to 87% accuracy.[10, 18]
Rational Engineering: When a promising candidate shows a high viscosity risk, it’s not always the end of the road. By understanding what causes it at the molecular level, scientists can sometimes re-engineer the antibody, making specific amino acid changes to reduce viscosity while keeping its therapeutic function.
These approaches look good, but they have limits.[13, 14, 16] The accuracy of any predictive model depends on the quality and diversity of the data it was trained on. Many models are built on data from standard monoclonal antibodies, and using them for the many different kinds of bispecifics is still being researched.
How Leukocare Can Support These Challenges
This is where a good formulation partner comes in handy. At Leukocare, we've developed a special way that mixes data science with deep knowledge of formulation to handle the specific challenges of bispecific antibodies.
Our platform uses AI and machine learning to help with formulation. Instead of just doing broad experiments, we use predictive algorithms to check your molecule and find the best formulation strategies from the start.[19, 20] This helps us create a custom, data-driven plan that uses less of your valuable drug and speeds up development.
Here's how we do it:[21]
In-Silico Analysis: We start by analyzing your bispecific antibody with computers to understand its unique structure and physical and chemical properties. This helps us guess potential challenges, like viscosity and aggregation.
AI-Guided Formulation Design: Our AI tools suggest specific excipient combinations that should stabilize your molecule and keep viscosity low at high concentrations.
Targeted Wet-Lab Experiments: Our models help us design smaller, more focused lab studies. We confirm the performance of the top-ranked formulations, making sure we find a great, strong solution quickly.[20]
Predictive Stability Modeling: After initial formulation, we use predictive modeling to guess the long-term stability of your drug, giving you confidence as you head toward clinical trials and commercialization.
Value Provided to Customers
If you lead drug product development, working with Leukocare has clear benefits. We aim to make your project less risky and add strategic value.
Speed: By using predictive modeling to focus our efforts, we speed up formulation development, helping you meet tight deadlines.
Save Material: Our targeted approach means we use a lot less of your valuable, often scarce, drug substance compared to traditional, high-throughput screening methods.[21]
Less Risk: By spotting and reducing formulation risks early, we increase the chance of developing a stable, manufacturable, and commercially viable product.[21]
Partnership: We work like an extension of your team, giving you not just data, but a collaborative partner who really gets the science. We provide the structure, speed, and substance needed to build a strong CMC story for investors and regulators.
In a competitive field like bispecific antibody development, having a clear and smart formulation strategy is a big plus. By using predictive modeling, we can go beyond old methods and get these important new therapies to patients faster.
Frequently Asked Questions (FAQ)
Q1: What makes viscosity a bigger challenge for bispecific antibodies compared to standard mAbs?
Bispecific antibodies are structurally more complex. Their asymmetric design and dual-target binding can create unique surface properties and intermolecular interactions that aren't usually seen in standard monoclonal antibodies. These factors can lead to a higher tendency for protein-protein interactions, which is a key reason for increased viscosity at high concentrations.
Q2: How accurate are computer viscosity prediction models?[14]
The accuracy varies depending on the model and the data it was trained on. Some recent machine learning models report high predictive accuracy for classifying antibodies as having either low or high viscosity. Still, these models are tools to guide development, not replace experimental verification.[13, 16] Their real value is in focusing lab work on the most promising candidates and formulation strategies.
Q3: Can a bispecific antibody with high viscosity be fixed?
In many cases, yes. High viscosity can often be reduced through formulation changes, like adjusting the pH or adding specific excipients that disrupt protein-protein interactions. Sometimes, if the issue is spotted early enough, small changes to the antibody's amino acid sequence can also reduce viscosity without affecting how it works.
Q4: How much material is needed for a formulation study using predictive modeling?[14]
A key benefit of a data-driven approach is that it greatly reduces the amount of material needed compared to traditional high-throughput screening. By using computer analysis to design smaller, more focused experiments, the material requirements can be much lower, which is a big advantage during early-stage development when material is often limited.
Q5: At what stage should we start thinking about viscosity and formulation?[21]
The earlier, the better. Including developability and manufacturability checks, like viscosity risk, early in candidate selection is becoming a standard practice.[15, 24] Dealing with these potential issues early on can stop costly delays and failures later in development.
