Debiopharm

Ulf Andersson, the exceptional leader of Debiopharm’s preclinical safety and translational pharmacology teams, is at the forefront of integrating safety and efficacy in drug development. Formerly steering AstraZeneca’s safety science team in CVRM diseases, his wealth of experience spans oncology and antimicrobials. Ulf’s unique perspective, driven by a molecular approach, elevates risk assessment with quantitative precision. His commitment to advancing medicine through cutting-edge AI-driven models and prescient insights promises groundbreaking results in predicting toxicity and optimising drug combinations.

In an interview with Pharma Tech Outlook Europe, Ulf Andersson discusses the integration of AI in toxicology testing and how it should be developed in parallel with advancements in multi-organ and multi-microphysiological systems.

What are some significant challenges that have recently impacted the pharmaceutical industry, particularly in the field of toxicology testing?

One of the recent major challenges in the field is the Food and Drug Administration (FDA) Modernisation Act, which implies a move away from animal experimentation. This shift poses significant difficulties as the currently available models for assessing the safety of molecules primarily concentrate on a limited number of cell types, such as liver, kidney, and heart cardiomyocytes. Since the human body consists of approximately 200 different cell types, relying solely on these limited cell-based models may not fully capture the complexities and intricacies of human biology, potentially leading to incomplete or inaccurate safety assessments of new drugs and medical products. The challenge, therefore, lies in finding alternative testing methods that can adequately represent the diverse array of cell types in the human body, ensuring a more comprehensive understanding of the potential toxicities and effects of the molecules being evaluated.

Despite the push for alternative methods, there is still a prevailing belief among scientists that in vivo toxicology studies are necessary to provide a more holistic and reliable characterisation of a molecule’s toxicity profile. As a result, finding a balance between opting for new in vitro testing techniques while retaining the value of in vivo studies remains an ongoing and critical objective for the industry. As we move towards in vitro testing to replace early animal experimentation, it’s scientifically justified to continue in vivo toxicology studies to comprehensively assess a molecule’s toxicity. This ensures that all possible angles are explored in describing the potential risks associated with the compound.

Today, there is a growing emphasis on incorporating pharmacology strategies, particularly in oncology, where there is a push to replace in vivo testing in xenograft models with patient-derived organoids (PDOs) and patient-derived xenograft (PDX) cells. While this trend is ongoing, scientists involved in the projects still feel the need to validate the findings with non-clinical studies.

“AI models hold great promise in predicting toxicity and drug combinations, but many existing algorithms lack interpretability.”

The other challenge in the industry is the increasing expectation to integrate AI into strategies to enhance translatability, clinical efficacy, and safety. AI models hold great promise in predicting toxicity and drug combinations, but many existing algorithms lack interpretability. This lack of transparency poses regulatory challenges when attempting to justify the safety predictions made by AI software. Therefore, collaborations with AI companies are underway to improve the ability to predict the clinical efficacy of drugs in translational pharmacology. Due to the current limitations and lack of trust in AI models, scientists still rely on non-clinical in vivo models to verify AI predictions. It will take further advancements and evidence to convince the scientific community that AI can truly revolutionise clinical efficacy and safety assessments.

What are some technological trends in the pharmaceutical industry that excite you for the future?

In parallel with the advancements in AI, there is a recognition of the need to enhance multi-organ and multi-microphysiological systems (MPs) to further improve drug safety and efficacy assessments. While single or dual-organ microphysiological systems have been developed and utilised, the potential of multi-organ models in predicting drug metabolism and fate is highly appealing. These models offer a more comprehensive representation of the human body’s complexity, allowing for a more robust evaluation of drug safety and pharmacokinetics (DMPK).

The field is still in its early stages, and prospective validation is relatively limited, despite some companies having advanced models and retrospective analyses showing promising predictability. The challenge lies in conducting controlled studies to optimise molecules using MPs models for safety assessment and subsequently demonstrating their clinical benefits. Nevertheless, this area presents an exciting prospect for the future of drug development.

In toxicology, there is a notable shift from solely relying on establishing the no-observed adverse effect level in a traditional toxicology study to setting clinical doses. The industry is now actively exploring quantitative risk assessment in conjunction with pharmacokinetic-pharmacodynamic (PKPD) modelling and population modelling. By integrating these approaches, it becomes possible to predict the probability of observing adverse events in clinical studies based on non-clinical safety signals and given doses. This transformative approach in the pharmaceutical industry underscores the importance of combining PKPD modelling with toxicology to enhance predictions of clinical outcomes.

The successful implementation of these strategies depends on a deep understanding of the molecular basis behind the observed toxicological findings. Such insights are crucial in constructing relevant models and improving the accuracy and reliability of predictions in drug development and safety assessments. As the pharmaceutical industry advances in these areas, it moves closer to achieving more efficient and effective drug development processes with improved patient safety profiles.

What advice would you offer to senior leaders and CXOs working in the industry?

It is crucial for the toxicology field to continue evolving and exploring the potential of in vitro models. We should not solely rely on others to pave the way but actively contribute to the advancement of science. While there may be some resistance, particularly in smaller biotech and pharma companies, it is important to integrate these models into our safety profiling of molecules.

Moving away from reliance on animal models and towards quantitative risk assessment in toxicology, combined with a deep molecular understanding of toxicity mechanisms, is a valuable direction to pursue. This shift will not only contribute to the evolution of science but also provide us with increased confidence in the use of microphysiological systems and AI approaches. To enhance our understanding and interpretation of these models, it is crucial to be able to explain the insights they provide. This requires combining advanced computational modelling, such as adverse outcome pathways, with toxicology data and incorporating molecular mechanisms into the analyses.

By merging our understanding of adverse outcome pathways with advanced computational modelling and toxicology data, we can make significant strides in predicting and assessing safety outcomes. This integration of different approaches holds promise for enhancing our confidence in microphysiological systems and AI methodologies.