Digital pathology aims to reduce animal testing by reusing archived tissue data

The pharmaceutical industry is increasingly turning to digital pathology and archived tissue samples to reduce reliance on new animal studies. By extracting additional molecular and histological data from existing formalin-fixed, paraffin-embedded blocks, researchers can answer questions that arise after initial toxicology trials without conducting further experiments. This approach aligns with a broader regulatory shift towards non-animal methods for preclinical safety assessments.

Aleksandra Zuraw, a veterinary pathologist at Charles River Laboratories, explained that the traditional use of glass slides has evolved from a one-way process into a reusable data source. Previously, a pathologist would examine a slide, record findings, and the specimen would be stored. Now, digitized slides provide pixel-level data that can be paired with pathology reports and molecular measurements. This allows for the training of computational models to recognise patterns associated with specific biological changes, effectively turning a single slide into a resource for questions beyond the original assessment.

Charles River Laboratories launched its Alternative Methods Advancement Project in April 2024 to facilitate this transition. One key initiative involves the use of virtual control groups. In standard toxicology studies, each dose group is paired with a concurrent control group that does not receive the drug. By replacing some of these live control animals with matched historical data, companies can reduce the total number of animals required. Zuraw noted that this requires establishing that earlier study conditions are comparable enough to provide a reliable baseline, ensuring that differences between studies are not mistaken for drug effects. The company’s team is currently working to generate sufficient matched data to provide these virtual controls wherever possible.

The reuse of archived tissue also addresses the limitations of fresh-frozen samples. While fresh tissue is ideal for molecular analysis, it requires planning before a study begins. If new questions arise after a study concludes and suitable fresh samples are unavailable, researchers have traditionally needed to conduct another animal experiment. However, formalin-fixed, paraffin-embedded blocks retain molecular information about how tissue responded to treatment. This allows scientists to extract additional insights from the same animals and blocks, preserving the context of the original study while avoiding new procedures.

This strategy supports guidance from the Organisation for Economic Co-operation and Development, which describes how preserved tissue can support molecular investigations, including studies of gene expression and proteins. The OECD emphasises that sample preparation and storage significantly affect result quality. In the United States, the FDA Modernization Act 2.0, enacted in 2022, allowed non-animal methods to support applications for human drug trials. The FDA has since published a roadmap to reduce animal testing in preclinical safety studies, aiming to make animal studies the exception rather than the norm over the next three to five years. New approach methodologies are intended to become the default, starting with monoclonal antibodies and expanding to other biologics and chemical entities.

Technological advancements are further accelerating this shift. Zuraw highlighted the growing use of molecular prediction from standard haematoxylin and eosin-stained slides. This approach applies new computational methods to digitized images, potentially eliminating the need for additional methodologies. Virtual staining is also expanding, with software capable of generating the appearance of a stain from an unstained tissue section captured by a scanner. While most current virtual staining still begins with a glass slide, the trend points towards a wider cycle where labs invest in systems to create and share digital images. Zuraw described this as a potential snowball effect, opening up data sources that were previously inaccessible.

Implementing these methods requires laboratories to demonstrate their efficacy in actual studies. Zuraw stressed the need for early adopters to generate precedent by embracing regulatory guidance and troubleshooting implementation issues. This work will allow others to build on existing frameworks without reinventing the process, ultimately facilitating a routine practice that spares future animal studies.