Tox21 and ToxCast (hereafter referred to as ToxCast) are two large US-based research programs where several thousand chemicals undergo testing in hundreds of in vitro bioassays. The data generated by this high-throughput screening (HTS) are being openly shared to enable external exploration and analysis aimed to advance predictive toxicology approaches. Two studies published this year by a group of academia- and industry-based European scientists have explored whether and how the ToxCast data could be used in the analysis of food-related chemicals’ safety.

The article by Ans Punt and colleagues, published on February 10, 2020, in a peer-reviewed journal Toxicological Sciences, investigated whether ToxCast data could deliver “mechanistic insights in the biological targets of food-relevant chemicals . . . grouped according to structural similarity.” The authors worked with a set of 556 direct food additives identified  by Agnes Karmaus within the ToxCast-tested chemicals (FPF reported). The study found that the identified chemical groups have been tested in a “limited number of [ToxCast] assays” only. Nonetheless, this testing delivered “sufficient results” for profiling these chemical groups for two types of biological targets, namely “DNA binding” and “nuclear receptor.” Many of the identified activities were related to the estrogen receptor-mediated actions. The authors concluded that HTS data “could add to the evidence considered for regulatory risk assessment of food chemicals and to the evaluation of desirable effects of nutrients and phytonutrients.”

In a follow-up article by James Firman and colleagues, published on November 30, 2020, in the peer-reviewed journal Chemical Research in Toxicology, the same group of 556 chemicals was used to explore how the ToxCast data could be utilized for toxicological read-across. In particular, the researchers addressed whether and to which extent “the HTS data could provide information enabling (1) the elucidation of underlying bioactivities associated with apical toxicological outcomes, (2) the closing of existing toxicological data gaps, and (3) the definition of the boundaries of chemical space across which bioactivity could reliably be extrapolated.”

With regard to the first two questions, the authors reported that many biological targets seen to be activated by the tested chemicals do not yet have any validated association to toxicity pathways. This deficiency currently limits the toxicological interpretation and practical application of in vitro tests. To address the third question, they focused on estrogenicity as a well-characterized endpoint and looked at it within the two exploratory case studies which focused “upon groupings of paraben-gallates and pyranone-type compounds” such as flavonoids. In both cases, HTS data was found to “reflect estrogenic potencies in a manner which broadly corresponded to established structure-activity group relationships.” This led the authors to conclude that HTS  data can indeed be utilized for “the identification of out-of-domain compounds,” thus helping to define the chemical space boundaries when performing read-across.


Punt, A., et al. (2020). “Potential of ToxCast data in the safety assessment of food chemicals.Toxicological Sciences 174: 326-340.

Firman, J. W., et al. (2020). “Exploring the potential of ToxCast data in supporting read-across for evaluation of food chemical safety.Chemical Research in Toxicology (published November 30, 2020).

Karmaus, A., et al. (2017). “Curation of food-relevant chemicals in ToxCast.” Food and Chemical Toxicology 103: 174-182.

Karmaus, A., et al. (2016). “Evaluation of food-relevant chemicals in the ToxCast high-throughput screening program.Food and Chemical Toxicology 92: 188-196.