Abstract
Drugs prescribed during pregnancy affect two populations simultaneously: fetuses and their mothers. Drug-induced fetal injury (teratogenicity) has a significant impact on current and future public health. Teratogenic risk designation of many drugs relies on associating rare fetal events with rare environmental exposures. Therefore we aim to develop preclinical predictive models of clinical teratogenicity. We collated public databases for drug-target-gene relationships for 619 drugs spanning the 5 pregnancy risk classes. Genes targeted by high risk but not low risk drugs demonstrated 79% accuracy (p < 0.0001 vs. random) for predicting high vs. low fetal risk on cross validation. Functional enrichment analysis revealed that target genes of drugs known to be safe in pregnancy contained no developmentally related terms, while target genes of known teratogens contained 85 developmentally related terms. Drug target gene signatures that are enriched for known developmental genes may provide valuable preclinical predictive information regarding drug pregnancy risk.