Replicability in EFDT Studies: Implications of Context Dependence for Virtual Control Groups (VCGs).

Reliable interpretation of animal toxicology studies depends on the ability to reproduce and replicate experimental findings. Recent discussions of Virtual Control Groups (VCGs) in nonclinical toxicology have emphasized the importance of matching study variables when constructing control datasets. However, the biological implications of variability among studies have received limited explicit attention. In embryo-fetal developmental toxicity (EFDT) studies, numerous intrinsic and extrinsic variables can influence pregnancy outcomes, fetal development, and the interpretation of findings. Drawing on previous analyses of EFDT variability and recent discussions of VCG implementation, 69 potential sources of non-treatment variability were identified and organized into five general domains, including animals, pre-study practices, the microenvironment, the macroenvironment, and study design/procedures. These domains differ in the degree to which they can be controlled or standardized across laboratories. Some variables may have negligible effects under many conditions, whereas others, alone or in combination, may influence study outcomes. EFDT outcomes are therefore inherently context-dependent, reflecting the combined influence of treatment and study context. Under such conditions, strict replicability of control-group outcomes across studies becomes biologically improbable. Recognition of this inherent variability has implications for expectations of study concordance and for the construction of VCGs as well as often-used historical control data (HCD). The development of reliable VCGs is therefore most feasible within individual laboratories where major domains of variability can be more closely aligned.