A practical approach for regulatory acceptance of virtual control groups: vehicle pooling and biological relevance assessment.

The implementation of virtual control groups (VCG) to replace concurrent control groups (CCG) in toxicity studies is challenged by the need for strict covariate matching. This is particularly true in diverse administration vehicles, which severely fragment historical control data (HCD). To overcome this limitation, this study consolidated region-specific HCD from physicochemically similar aqueous vehicles (saline, distilled water, and 0.5% methylcellulose) into a site-specific pool for 13-week rat studies in the Republic of Korea, thus securing a sufficiently powered VCG pool. To rigorously evaluate VCG reproducibility beyond strict statistical thresholds (e.g., p < 0.05), a category-based assessment framework was introduced as a validation methodology focusing on the direction and magnitude (e.g., percentage change) of alterations. Although body and organ weights showed high statistical agreement between VCGs and CCGs, the exact reproduction of statistical significance for clinical pathology parameters was relatively low (25%-40%) owing to limited sample sizes and inherent biological noise. Nevertheless, the category-based assessment demonstrated 100% reproducibility based on the biological relevance of all evaluated clinical pathology parameters. Crucially, the VCG-based analysis confirmed that the study director's fundamental toxicological interpretations and target organ signals remained robustly preserved. Overall, pooling HCD across similar vehicles resolves sample size constraints, and assessing VCG reproducibility via biological relevance provides a practical framework to enhance regulatory acceptance.