Takeaways from a panel discussion at the World Clinical Biomarkers & Companion Diagnostics Summit
A recent panel discussion, hosted by Nucleai at the World Clinical Biomarkers & CDx Summit, brought together experts from leading pharmaceutical and biotechnology companies to share insights and advances at the intersection of artificial intelligence (AI), digital pathology, and translational research and their potential impact on therapeutics development.
The panelists included Christine Ward, VP and Head of Translational Medicine at Moderna Therapeutics; Rolando Yanes, Global Diagnostics Lead at Amgen; Mark Gustavson, Head of Computational Pathology Translational Science at AstraZeneca; and Oscar Puig, Head of Translational Medicine and Diagnostics at Nucleai. The discussion spotlighted the potential of AI-powered digital pathology to transform drug development and diagnostics and fostered a vibrant exchange of plans for increasing industry adoption of this technology.
The Promise of AI and Digital Pathology : Unlocking a deeper context for next-gen therapeutics
The panelists agreed that an area where AI and digital pathology can bring the most value is in unlocking complex biomarker analysis. As the industry moves towards more sophisticated therapeutics like vaccines, bispecific antibodies, and antibody-drug conjugates (ADCs), there is a growing need for equally advanced biomarkers to enable hypothesis generation at scale.
Digital pathology, enhanced by AI algorithms, empowers teams to extract more information from tissue samples, including spatial context of biomarkers, enabling a deeper understanding of disease mechanisms and treatment effects. As Gustavson described, “Computational pathology holds the promise of being able to unlock things that we can’t just do by eye.”
This technology is particularly promising for enabling next-generation treatments and diagnostics. By providing a more nuanced and quantitative analysis of tissue samples, AI-powered digital pathology can help researchers identify novel biomarkers, optimize patient selection for clinical trials, and potentially expand the pool of patients who might benefit from a particular treatment.
“In oncology,” said Ward, “it’s really about understanding what’s happening in the tumor with sophistication. Then we can really position our drug in future studies to target populations where it’s showing benefit.”
Exciting Applications and Use Cases: ADCs, spatial scoring and more
One of the most exciting applications discussed was in the development of antibody-drug conjugates (ADCs). Gustavson emphasized that for ADCs, simply identifying the presence of a therapeutic target is not sufficient. The pattern of expression and spatial distribution of target cells can significantly impact drug efficacy.
“We had some great gastric cancer data that proved this,” said Gustavson. “The predictive power and what predicts outcome in gastric cancer within HER2 is the pattern of expression, not necessarily the level of expression.”
In his team’s study, AI-powered digital pathology allowed for precise quantification of target expression at the single-cell level and analysis of spatial relationships between cells, providing crucial insights for ADC development and patient selection.
Spatial analysis of the tumor microenvironment was identified as another critical application. In Ward’s team’s work on a cancer vaccine, digital pathology enabled detailed analysis of multiple cell populations within the same tissue sample, their interactions, and their spatial relationships within the tumor microenvironment. They were able to characterize activation of T cell populations and tumor immunosuppression in the microenvironment.
The recent development of Quantitative Continuous Scoring (QCS) was introduced as a powerful AI-driven digital pathology tool for biomarker analysis. This approach, discussed by Gustavson, allowed for truly quantitative assessment of single-cell protein expression with a lower limit of detection and broader dynamic range than traditional methods. The team then used QCS to analyze the spatial distribution of cells and understand the patterns of protein expression that were driving ADC efficacy.
Challenges in Implementation: from clinical utility to practical application
Despite the potential of AI and digital pathology, there are significant challenges facing widespread adoption. “At some point, digital pathology is going to be a tool that everybody uses and is really accurate,” said Puig. “The reality is quite different, though. Less than 10% of the labs are digital. There’s no reimbursement for the tools. Why is there this discrepancy?”
The panelists identified a major hurdle being the lack of payer reimbursement for these advanced tools and the pressure to prove clinical utility to drive adoption and secure reimbursement. Without demonstrating clear value in patient outcomes or clinical decision-making, implementation will remain challenging.
Strategies for Successful Integration of AI and Digital Pathology
The panel stressed the importance of early incorporation of digital pathology and AI in the drug development pipeline. Ward and Yanes advocated for using these technologies in early-stage research to validate biomarkers, generate hypotheses and inform go/no-go decisions, gaining insights in time to inform subsequent programs and convince clinical teams to take the next step. Gustavson agreed: “By the time you get to phase 3, you’re too late.”
When asked about the best ways to convince regulatory, clinical, medical and other internal stakeholders of the need for digital pathology, Yanes said, “You have to show value for the digital biology tool. It has to do with being able to either expand the patient pool or have more confidence that the patients you’re selecting are more likely to respond. You have to be able to communicate the value to the commercial team.”
Collaboration between pharmaceutical companies, diagnostic partners, and AI companies was identified as crucial for successful implementation. Yanes and Gustavson highlighted the complexity of integrating digital pathology into trial operations, emphasizing the need for increased alignment between his team, clinical operations, and testing labs.
Yanes also noted that “prior experience from the [technology] partner is very important.” Both Yanes and Ward said that they specifically seek out partners who have completed projects studying their biomarkers, tissues, and therapeutic areas of interest. “When we’re looking at partners, we need to see what work has been done before,” said Yanes. “That gives us the confidence to invest.”
Demonstrating Value: the Nucleai Approach
Puig outlined that Nucleai’s approach to implementing AI-powered digital pathology works because the Nucleai team is focussed on deeply understanding all of the customer requirements first.
“Communication and collaboration are super important, and as early as possible,” said Puig. “Early in the project, we can really understand what the problem is.”
Nucleai brings deep biological and AI solution expertise, and communicates closely with pharma partners to inspect the specific problems that can be addressed through AI and digital pathology, to deliver a trusted solution with tangible value.
The Future of AI and Digital Pathology
The panelists identified a number of key benefits on the horizon that had the greatest potential to widen adoption of these technologies. “We’re just beginning to scratch the surface,” said Gustavson.
- Recruiting more patients: Expanding patient pools has been a major obstacle to the development of certain therapeutics, and the panelists were optimistic about the potential of AI and digital pathology to help overcome this challenge. By providing more nuanced biomarker analysis, these technologies could help identify patients who might benefit from a therapy but would have been missed by traditional diagnostic methods.
- Studying the right patients: Improved patient selection and treatment outcomes were cited key benefits. The ability to analyze complex spatial relationships and expression patterns could lead to more precise patient stratification and personalized treatment strategies.
>The panel also touched on the rapidly evolving nature of AI technology and its ability to generate more and more sophisticated insights, thanks to advances in computational capabilities.
Next Steps: How R&D Teams Can Advance with AI and Digital Pathology
Ultimately, the panelists agreed that the integration of AI and digital pathology in translational research represents a significant opportunity to advance drug development and personalized medicine. While challenges remain, particularly in widespread adoption and proving clinical utility, the potential benefits are substantial. As these technologies continue to evolve and demonstrate value, they are poised to become integral tools in the development of next-generation therapeutics and diagnostics.
For translational research scientists, staying informed about these advancements and actively exploring their potential applications in ongoing research projects will be crucial. Engaging with AI and digital pathology experts, participating in collaborative initiatives, and advocating for the integration of these technologies in research pipelines can help drive the field forward and ultimately improve patient outcomes.
Connect with Nucleai’s AI-Spatial solution Experts
