Deriving bio-equivalents from in vitro bioassays: assessment of existing uncertainties and strategies to improve accuracy and reporting.

Bio-equivalents (e.g., 17β-estradiol or dioxin equivalents) are commonly employed to quantify the in vitro effects of complex human or environmental samples. However, there is no generally accepted data analysis strategy for estimating and reporting bio-equivalents. Therefore, the aims of the present study are to 1) identify common mathematical models for the derivation of bio-equivalents from the literature, 2) assess the ability of those models to correctly predict bio-equivalents, and 3) propose measures to reduce uncertainty in their calculation and reporting. We compiled a database of 234 publications that report bio-equivalents. From the database, we extracted 3 data analysis strategies commonly used to estimate bio-equivalents. These models are based on linear or nonlinear interpolation, and the comparison of effect concentrations (ECX ). To assess their accuracy, we employed simulated data sets in different scenarios. The results indicate that all models lead to a considerable misestimation of bio-equivalents if certain mathematical assumptions (e.g., goodness of fit, parallelism of dose-response curves) are violated. However, nonlinear interpolation is most suitable to predict bio-equivalents from single-point estimates. Regardless of the model, subsequent linear extrapolation of bio-equivalents generates additional inaccuracy if the prerequisite of parallel dose-response curves is not met. When all these factors are taken into consideration, it becomes clear that data analysis introduces considerable uncertainty in the derived bio-equivalents. To improve accuracy and transparency of bio-equivalents, we propose a novel data analysis strategy and a checklist for reporting Minimum Information about Bio-equivalent ESTimates (MIBEST).