Exploration of virtual control groups and Bayesian approaches for rat fertility studies

This study evaluates the potential of using historical control data (HCD) in rat fertility studies to replace/reduce concurrent control animal use and improve statistical analysis. This analysis was conducted using data sourced from a single test facility using consistent animal attributes and study procedures and consisted of 13 datasets from 12 rat fertility studies conducted between 2018 and 2023. For required fertility endpoints, we have evaluated full and hybrid Virtual Control Group (VCG) and Bayesian statistical approaches compared with standard statistics used for concurrent control groups (CCG). Our findings demonstrate that the sensitivity and specificity of most required fertility endpoints were generally similar between CCG and full or hybrid VCGs of sufficient sample size. Bayesian analyses leveraging all available HCD offered similar or superior sensitivity and specificity to CCGs and VCGs (full or hybrid) for all required fertility endpoints, with some exceptions. A retrospective case study implementing these approaches with HCD illustrated similar statistical performance across all approaches in addition to the benefit of reduced animal use. Overall, this effort illustrates the potential to improve sensitivity and reduce animal use with VCG (full or hybrid) or Bayesian approaches for required fertility endpoints in the rat.