Reference Data for Evaluation of In Vitro Test Methods and Computational Models

NICEATM is compiling high-quality data from in vivo testing of chemicals for potential endocrine disruptor activity. These data have potential use for:

  • Developing adverse outcome pathways or models of endocrine activity.
  • Prioritizing chemicals for further testing.
  • Evaluating species-specific responses to chemicals.

Curated Database of Rodent Uterotrophic Bioactivity

NICEATM searched the peer-reviewed literature for uterotrophic assay data on 1812 chemicals tested in ToxCast Phase II. The uterotrophic assay (EPA OPPTS 890.1600) measures the estrogenic activity of a substance by assessing its effect on the weight of the uterus in a female rodent.

The initial search yielded over 1000 papers, of which 670 were potentially relevant based on the inclusion of uterine weight as a measured endpoint. From these 670 manuscripts, 2615 individual chemical/study/protocol combinations were extracted, providing data on 235 chemicals with unique CASRNs.

Only study design protocols that met six minimum criteria were included in the Guideline-Like Uterotrophy Database (GL-UTDB). The GL-UTDB contains information from 458 guideline-like (GL) studies extracted from 93 publications, providing high-quality in vivo estrogenic bioactivity data on 118 chemicals with unique CASRNs (103 of which are in the ToxCast/Tox21 inventory). We included all chemicals in the studies returned by our search, some of which were not in the ToxCast library but were included in publications that also examined ToxCast chemicals. We performed an additional round of manual quality assurance on all study information in the GL-UTDB to confirm data entry accuracy. All data from the GL-UTDB are available via the NICEATM Integrated Chemical Environment.

Curation of Data from the Rodent Hershberger Assay

In a similar effort, OECD, NICEATM, and EPA scientists collaborated to develop a reference database and evaluate an in vitro model to identify chemicals with the potential to interact with the androgen receptor (AR). The AR model is potentially a rapid, cost-effective replacement for the in vivo Hershberger assay (EPA OPPTS 890.1400).

  • Browne et al. 2018 describes assembly and curation of a data set of results for the Hershberger and other in vivo assays for AR activity. Ultimately, 49 chemicals were identified with reproducible AR pathway responses confirmed in at least two in vivo rodent studies. These 49 chemicals could be considered reference chemicals useful for validating alternative methods. The list of chemicals and the in vivo AR activity data are available in the NICEATM Integrated Chemical Environment.
  • Kleinstreuer et al. 2018 describes use of the reference chemicals identified through the data curation effort to interrogate the performance of a ToxCast/Tox21 AR model based on 11 high-throughput assays. The AR model had 100% positive predictive value for the in vivo response, where chemicals with conclusive AR model results were consistently positive in vivo.