For decades, the path from a promising antibody discovery in the lab to a life-saving therapy for patients has been fraught with uncertainty. Many drugs that show incredible potential in early tests ultimately fail in human trials, costing billions and delaying medical progress. Today we’re speaking with Ivan Kairatov, a biopharma expert with deep experience in therapeutic development, about a new platform poised to change this paradigm. We’ll explore the critical, often-overlooked mismatches between preclinical models and human biology, see how a new genetic approach creates a more faithful replica of our immune system, and discuss how this innovation can de-risk drug development, enhance patient safety, and accelerate the delivery of next-generation medicines.
Many promising antibody drugs fail during human trials despite strong early results. What specific biological mismatches in conventional lab models, particularly involving an antibody’s Fc domain and immune cell receptors, create these misleading outcomes? Could you provide a concrete example of this process?
This is the central challenge we’ve been grappling with for years. The issue lies in a fundamental, yet often underappreciated, part of antibody biology. We tend to focus on the “head” of the antibody that binds to a target, but the “tail,” or Fc domain, is just as crucial. It acts like a set of instructions for the immune system, which are “read” by Fc gamma receptors on our immune cells. The problem is that these receptors differ substantially between humans and the animal models we rely on. So, an antibody tested in a standard mouse might send a signal to dampen inflammation, but in a human, that same antibody could trigger a powerful, and potentially dangerous, killing response. A tragic and well-known example is the development of anti-CD40L antibodies. They looked reassuring in preclinical studies but went on to cause severe, fatal blood clots in patients. The reason? The mechanism causing the clots—direct activation of human platelets—simply doesn’t exist in standard lab mice, so the risk was completely invisible until it was too late.
Your new platform is built on a mouse model using a precise “knock-in” genetic strategy. How does this approach create a more faithful replica of human immune responses compared to other humanized models, especially when it comes to capturing changes during inflammation?
It’s all about precision and biological context. Instead of just adding human genes randomly, this platform was built on a very detailed blueprint. The research consortium first created an exhaustive map showing exactly which Fc gamma receptors are present on which immune cells in humans, and how that compares to our existing models. Using that map, they employed a “knock-in” strategy to replace the mouse genes with their human counterparts in the correct locations. This ensures that when we test a therapeutic antibody, it’s interacting with immune cells in a way that truly mirrors what happens in a patient’s body. What makes this model particularly powerful is its ability to reflect the dynamic nature of the immune system. During inflammation, receptor expression on immune cells changes dramatically. This model captures those shifts, providing a much more accurate picture of how a drug will perform in a diseased state, not just in a healthy baseline system.
The development of anti-CD40L antibodies, which led to unforeseen and dangerous blood clots in patients, revealed a major blind spot in preclinical safety testing. How does your platform specifically detect risks like human platelet activation, a mechanism that is invisible in standard animal models?
That case was a wake-up call for the entire industry. The core issue was that standard models lack the specific immune architecture that leads to that particular adverse event. Human platelets have Fc gamma receptors that can be activated by certain antibody complexes, triggering the clotting cascade. Since this machinery is absent in mice, the risk was completely missed. This new platform addresses that blind spot directly. By humanizing the Fc gamma receptor system with such high fidelity, it creates a biological context where these human-specific interactions can actually occur. If an antibody candidate has the potential to activate human platelets, we will now see that signal in this model. It allows us to screen for these hidden dangers early on, ensuring that a tragedy like the one seen with anti-CD40L antibodies isn’t repeated. We can now ask the right safety questions and get meaningful answers before a drug ever enters a human trial.
Beyond safety, this model reportedly helps rank antibody candidates by effectiveness. Could you walk us through the process of how a research team would use this platform to compare different antibody designs and confidently select the most promising one to advance toward clinical trials?
This is where the platform really accelerates development. Modern antibody engineering is all about subtle fine-tuning—making tiny molecular changes to the Fc domain to get a desired immune response. In the past, predicting the real-world impact of these changes was largely guesswork. Now, a research team can take, say, three different versions of their antibody, each with a slightly different Fc design, and test them head-to-head in this model within a specific disease context, like cancer. They can then directly measure and compare the outcomes. For example, they can reliably see which version is best at instructing immune cells to eliminate tumor cells, or which is most effective at removing a population of harmful immune cells in an autoimmune setting. The data isn’t just theoretical; it’s a direct readout of biological effectiveness in a system that behaves like a human’s. This allows the team to confidently select the single best candidate to move forward, saving immense time and resources by avoiding the pursuit of less effective designs.
Regulatory agencies are increasingly emphasizing the need for better predictive models. How does the commercial availability of this platform, developed through an academic-industry partnership, help both drug developers and researchers meet these evolving standards and accelerate the adoption of more accurate testing methods?
The timing is perfect. Regulatory bodies like the FDA are actively pushing the industry to move beyond legacy models and adopt technologies that can provide stronger, more predictive evidence before starting human trials. This platform is a direct answer to that call. The fact that it was developed through a collaboration between top-tier academic institutions and industry leaders like argenx gives it immense credibility. Making it commercially available through partners like genOway is the crucial final step. It democratizes the technology, so it’s not limited to a single institution. Now, any drug developer, from a small biotech to a large pharmaceutical company, can access this superior model to generate the high-quality, biologically relevant data that regulators want to see. This helps them build a much stronger case for their drug candidate and, ultimately, accelerates the entire approval pathway.
What is your forecast for antibody drug development over the next decade, now that more accurate predictive tools are becoming available?
I believe we are on the cusp of a major acceleration in innovation. For years, our ability to engineer incredibly sophisticated antibodies has outpaced our ability to accurately test them. This created a bottleneck. With tools like this new platform, we can finally close that gap. Over the next decade, I forecast that we will see a significant drop in the failure rate of antibody drugs in late-stage clinical trials. We’ll be able to tackle more complex diseases with finely-tuned therapeutics because we can confidently predict their effects. This will not only bring more effective and safer medicines to patients faster, but it will also encourage more ambitious research, as scientists will no longer be held back by the limitations of outdated preclinical models. We are moving from an era of educated guessing to one of predictive, data-driven design.
