How Paratope Refinement Can Mitigate Antibody Polyspecificity

July 3, 2019

July 3, 2019 | There is often a lack of understanding in drug discovery about how polyspecificity drives antibody toxicity, says Jonny Finlay. This shows polyspecificity is an underappreciated phenomenon in therapeutic antibody development, but that these unwanted properties can be fully improved by paratope refinement.

Finlay is the CEO of UltraHuman, an antibody drug discovery biotech in the UK that is developing a series of therapeutics for inflammation and oncology. Prior to co-founding UltraHuman, Jonny led research teams in Biologics Discovery at Pfizer and Wyeth, where he carried out postdoctoral research in recombinant protein engineering at several institutes, including the Centre for Biologics Evaluation and Research, FDA.

On behalf of Bio-IT World, Kent Simmons, Senior Conference Director at Cambridge Healthtech Institute (CHI), spoke with Finlay about the role of polyspecificity in the toxicity and side effects of checkpoint inhibitors, the solvability of the problem, and where bottlenecks occur during the development stage of biotherapeutic R&D.

Editor's note: Simmons is producing the Antibody Forum as part of the upcoming Discovery on Target (DOT) conference in Boston, September 16-19. Finlay will be speaking on the program. Their conversation has been edited for length and clarity.

Bio-IT World: Could you talk for a moment about the role of polyspecificity in the toxicity and side effects of checkpoint inhibitors?

Jonny Finlay: I think polyspecificity is one of the most misunderstood and ignored factors in therapeutic antibody development. While antibody specificity has been a recognized problem in diagnostics for years, it is still under-appreciated as a risk factor in therapeutic antibody development. I believe the studies I will present at DoT really prove definitively that antibody polyspecificity is not just a rare, academic curiosity. It is a sincere clinical risk, causing unique side-effects that can be tracked from the clinic all the way back to the lab, to truly understand the causes. In therapeutics such as checkpoints inhibitors, ADCs, CD3 redirecting agents and CAR-T, this is of immense importance for both efficacy and safety.

Do you see this as a solvable problem, or is this a general problem with all molecules in this class?

I do believe that this problem is a risk for any antibody, as polyspecificity is a naturally occurring phenomenon. Having said that, I very much believe this is a solvable problem if we are willing to address the challenge. R&D groups from multiple large and mid-size drug discovery companies and academic groups have already identified accurate methods for weeding out polyreactive clones, which have broadly reactive "sticky" chemistry problems that can severely affect the pharmacokinetics and/or bioavailability of biologics. Polyspecificity is different, however, and is defined by reactivity to a limited number of apparently unrelated proteins in a given proteome. Frustratingly, it can be species-specific, as antibodies generated in mice for example are only tolerated across the mouse proteome, not the human one. By applying cutting-edge methods for screening antibodies for polyspecificity and other development risks across our antibody portfolio, we have a fantastic opportunity to raise the bar in success rates for therapeutic antibodies.

You and colleagues published a paper last fall outlining an interesting approach to this.  Could you please tell us more about this work?

In that study (DOI:https://doi.org/10.1080/19420862.2018.1550321), we were fascinated to see that a clinical-stage antibody had very high incidence of a highly specific and unusual side effect that had not been observed in other antibodies addressing the same target (PD1). For the first time, we took this type of specific clinical insight and traced it back to the fundamental issue with the antibody in question. Through proteomic and cell-based assay analyses we identified the probable root cause of the side-effect, which was remarkably specific, off-target receptor agonism. Moreover, we used our unique molecular engineering approaches to completely remove this off-target binding problem and create a molecule with ideal clinical development characteristics. We published some of that data, but not all. Further insights will be presented at DoT.

It looks like this work is being done through a new entity called "Ultrahuman."  Can you tell us more about that venture and what we’ll be seeing from them in the future?

We founded UltraHuman to directly address the issues outlined above, taking a holistic approach to antibody therapeutics. We believe that the next generation of biologic therapeutics must learn from the past to have a maximized chance of success in the clinic, so we need to fully address the factors that cause failures. That means maximizing molecular quality attributes like specificity and stability, addressing key aspects of immune function via Fc engineering, and even generating new molecular formats to reduce toxicity and significantly improve therapeutic index. There is a lot more to come from UltraHuman in the near future, with exciting new molecules coming thick and fast.

Could you share a few thoughts on where bottlenecks occur during the development stage of biotherapeutic R&D, and what needs to happen to overcome these?

The entire process of biologic drug discovery is better understood than ever before, but that actually creates more bottlenecks than it removes as we learn more and more about what causes failure. As a result, one of the biggest issues in successful progress in biologic drug discovery, in my opinion, is the struggle to convince the community to work harder on their molecules "up front" and to not ignore known risks that can cause painful and costly late-stage failure. Not all antibodies are the same. Not all are low risk. Ignoring these facts is false economy, as nothing is more expensive than failure.