Hill or mountain? Virtual control groups in DART testing - and the need for firm footing

The concept of virtual control groups (VCGs) has gained significant momentum in recent years as a potential approach to reduce the use of concurrent control animals (CCGs) in nonclinical toxicology, including Developmental and Reproductive Toxicology (DART). While early proposals emphasized the promise of substituting CCGs with well-curated historical control data (HCD), subsequent evaluations have revealed considerable methodological, statistical and practical challenges. Evidence from retrospective analyses, proof-of-concept studies and consortium-driven initiatives demonstrates that the performance of VCGs critically depends on data granularity, metadata completeness, laboratory consistency and the inherent variability of toxicological endpoints. Some recent studies indicate that VCGs or hybrid approaches can approximate CCG-based interpretations under certain conditions. Variable reproducibility of quantitative endpoints, uncertainties in matching criteria and unresolved questions surrounding pathology evaluation currently might limit use in regulatory toxicology. It is proposed that adapted study designs, improved collection and presentation of historical data, may offer alternative pathways to animal reduction without compromising human risk assessment.