Honoring the Control Group: Advanced Analytics Help Reduce Animal Use in Safety Studies
The control group is the foundation of nonclinical safety science. It is what lets us separate a compound's effects from the background noise of biology, and every conclusion about safety rests on it. The question we are asking is how to reduce animal usage while maintaining the scientific rigor and the confidence that controls provide the answer may lie in data and expertise the industry has spent decades generating. Toxicology studies run under tightly standardized conditions, and every one of them has used control animals whose data are meticulously collected and carefully documented.
Virtual Control Groups (VCGs) reuse that curated historical control data matched to the design of a new study, to partially replace the concurrent control groups of new studies. The animals that contributed data have already informed regulatory decisions once. VCGs ask whether their data can serve science again.
Current implementations suggest reductions of approximately 12% in control animals, with the potential for greater reductions as validation efforts progress and regulatory confidence grows. Multiplied across the volume of studies the industry conducts every year this equates to a substantial reduction in control animal use and every animal that is not needed is a meaningful outcome.
What the evidence shows
Our retrospective analysis in Regulatory Toxicology and Pharmacology examined 20 toxicology studies in which curated VCGs replaced concurrent controls, comparing outcomes at both the study-decision and endpoint level. The No Observed Adverse Effect Level (NOAEL), the value that anchors dose selection, matched in all 20 studies, a 100% concordance. We did observe some endpoint-level variability between concurrent and virtual controls, though the core study conclusions held. That variability is where our current work is focused, refining the models with clients to increase confidence.
The benefits extend beyond animal numbers. Larger, well-characterized historical datasets strengthen statistical power for rare spontaneous findings and give pathologists and toxicologists deeper context when something unexpected appears. The aim is to support scientific judgment and provide additional insights from historical precedent.
An investment in science, not a shortcut
Building a trustworthy VCG demands significant investment: curating and harmonizing decades of data, developing and validating the analytical models, establishing rigorous selection criteria, sustaining a robust, compliant infrastructure and working with regulators on qualifications. None of that is cheap or quick. We make that investment because the science is sound and because reducing the number of animals used in research is a goal worth pursuing.
Building confidence, step by step
The near-term path for VCG use is hybrid. Hybrid studies retain a reduced concurrent control group, supported by a curated virtual one. Prospective approaches, where VCGs and concurrent controls run in parallel, can show study-by-study whether the conclusions agree. Over time, as validation accumulates, the share of the control group that can be replaced may grow.
Regulatory engagement is central to that evidence. The European Medicines Agency has issued a draft qualification opinion on VCGs in place of concurrent controls in non-GLP dose-range-finding repeated-dose rat studies. The draft qualification opinion represents an important milestone in the regulatory consideration of VCGs.
What comes next
Three priorities stand out. First, rigorous validation of VCG models, including in GLP settings, are required. Second, clear, transparent criteria for when virtual control data can be considered comparable to a new study. Third, success in deploying VCGs and reducing control animal use relies on collaboration across sponsors, CROs, and scientific societies for implementation and harmonized frameworks.
Responsible use of animals in research has always meant asking whether we can learn the same thing with less. VCGs are an evidence-backed answer to that question, and one that respects both the control group and the animals that make it possible.
As validation efforts continue and industry collaboration expands, virtual control groups have the potential to reshape how nonclinical studies are designed. The opportunity is not simply to use fewer animals, but to apply decades of accumulated scientific knowledge in ways that strengthen decision-making while advancing the principles of responsible research.
Organizations exploring how virtual control groups may fit into their nonclinical development strategies can learn more about Charles River's research, expertise, and ongoing work in this evolving field by visiting the Charles River Virtual Control Groups resource center.
The editorial staff had no role in this post's creation.