Development of a streamlined rat whole embryo culture assay for classifying teratogenic potential of pharmaceutical compounds.

This study describes a novel rat whole embryo culture (rWEC) teratogenicity assay that applies a simplified experimental design and statistical prediction model, resulting in reduced animal requirements and increased throughput with low prediction error rate for classifying teratogenic potential of compounds. A total of 70 compounds (38 teratogens and 32 nonteratogens) were evaluated, and the prediction model was generated from a dataset of 59 compounds. The rWEC assay requires only one test concentration (1 mu M) and three structural endpoints (group average morphological scores of spinal cord, heart, and number of somite pairs), which are used in a recursive partition model for classifying teratogenic liability. The model fitting concordance between the WEC assay and in vivo outcome was 83% with a standard deviation (SD) of 4.9%. The predictivity for future compounds was evaluated by using two statistical methods. Fivefold cross-validation estimated the predictivity of this model at 73% (SD 5.8%). A second estimation of predictivity was obtained from an independent test set of 11 compounds that were not used to build the prediction model and reached 82% (SD 11.6%). The overall estimate for prediction concordance is 74% (SD 5.2%). There is no statistically significant difference (p value > 0.50) in the predictivity between this model and the model supporting European Center for the Validation of Alternative Methods WEC assay with predictivity of 80% (SD 10.6%). Overall, the streamlined WEC assay is estimated to reduce animal use and operational costs by more than 50%. It substantially improves results turnaround with no loss of predictivity.