Historical Toxicology Data Can Inform Future Drug Safety Studies, New Analysis Suggests

Analysis of more than 800 drugs identifies recurring toxicity patterns that could inform preclinical study design and help reduce unnecessary animal testing.

DURHAM, NC, UNITED STATES, November 3, 2026 /EINPresswire.com/ -- The pharmaceutical industry has spent decades generating extensive toxicology data to evaluate the safety of new medicines. Yet much of this information remains embedded in individual study reports and regulatory submissions, limiting its utility beyond the original development programs.
A new analysis from AI startup, FlashPath, demonstrates how historical toxicology data can be systematically examined to identify recurring patterns of drug toxicity, providing a foundation for more informed preclinical study design.

The work, described in a new article by FlashPath CEO Sara Selitsky, draws on nonclinical reviews from five regulatory agencies, including the US Food and Drug Administration (FDA) and European Medicines Agency (EMA). Examining repeat-dose animal toxicity findings associated with 896 drugs, the researchers identified networks of related pathological and clinical findings that recur across compounds, revealing recognizable toxicity profiles without relying on prior knowledge of drug classes.

Among the findings was a profile that appeared in all 11 corticosteroids examined, which included changes in the adrenal gland, lymphoid tissues, bone marrow, and blood-cell counts. These effects, which have been previously established, were re-discovered by applying frontier AI methods to systematically organized and annotated data.

The implications of this work extend beyond identifying toxicities. For example, by identifying class effects, researchers can focus subsequent experiments on questions specific to the drug of interest, including exposure dependence, functional consequences, reversibility, and unexpected effects that distinguish a new compound from its predecessors.

“We have become extraordinarily good at generating safety data, but far less so about learning from it across development programs,” said Selitsky. “There is an opportunity to bring accumulated knowledge into the design of the next experiment. Understanding what has already been established allows us to ask better questions about what we still need to learn.”

The findings reflect a broader opportunity emerging in pharmaceutical research. Advances in computational methods and data infrastructure are making it increasingly feasible to integrate previously fragmented scientific evidence, potentially changing how drug safety is evaluated.

This represents a fundamental shift in the way safety science can operate. Rather than treating each toxicology study as a largely independent exercise, FlashPath is developing AI-forward infrastructure to connect and analyze historical safety data across compounds, therapeutic classes, and development programs. The objective is to make decades of accumulated evidence more accessible and useful to researchers, while establishing a foundation for increasingly sophisticated computational and predictive approaches.

This approach also addresses the growing regulatory interest in modernizing nonclinical safety assessment, including the use of historical evidence and new approach methodologies to support scientifically justified reductions in animal testing. The study indicates that making historical evidence more accessible, comparable, and reusable could improve the quality of future research while reducing unnecessary duplication of work.

About FlashPath
FlashPath is a computational safety science company developing infrastructure to aggregate, integrate, and analyze toxicology, pathology, and related drug safety data. By transforming fragmented scientific evidence into connected, analysis-ready datasets, FlashPath aims to enable more informed drug development decisions, advance predictive toxicology, and accelerate the development of safer medicines.
Read the article: What Can Toxicity Profiles Tell Us About the Next Study? — https://www.drugdiscoveryonline.com/doc/using-recurring-toxicity-profiles-to-inform-preclinical-study-design-0001

Media Contact:
contact@flashpathai.com

Ryan Taft
FlashPath
+ +1 858-736-1667