Virtual control groups in non-clinical toxicology – A replicability challenge
Thomas Steger‐Hartmann, Guillemette Duchâteau-Nguyen, Frank Bringezu, Manuela Onidi, Martina Stirn · ALTEX · 2025
The concept of Virtual control groups (VCGs) describes the use of historical control data (HCD) to reduce or replace concurrent control groups (CCGs) in animal studies after careful matching of the HCD with the study and animal characteristics of a planned study (Steger-Hartmann et al., 2020).A prerequisite for the implementation of the VCG concept in animal safety studies is a careful analysis of the impact of CCG replacement or reduction on the study outcome.Any impairment of the study outcome with subsequent effects on risk assessment conclusions needs to be avoided (Golden et al., 2024).Before VCGs can be applied in animal safety studies, the concept will need to provide sufficient evidence that it is fitfor-purpose for a specific in vivo study type, which will in turn require qualification and validation procedures in the different industrial fields of animal safety testing involving various regulatory bodies. How to qualify the VCG concept?A frequently encountered and well-described problem during qualification and validation of a new replacement or reduction approach is the definition of the primary reference to which the new approach or method is compared to, as well as the definition of pre-defined acceptance criteria.The existing and often well-established in vivo method is generally considered the primary reference.During the validation process, the results of the new approach are compared to this standard of truth and are usually assessed using contingency tables to group true positives and true negatives together with the respective false outcomes hereby assessing the concordance of results.The concordance between the new method and the conventional assay represents the replicability, which is defined as "obtaining consistent results across studies aimed at answering the same scientific question, each of which has obtained its own data" (National Academies of Sciences, Engineering, and Medicine, 2019).