The following blog pertains to the 2017 SOT Annual Meeting Scientific Session "Opportunities for Read-Across Development and Application Using QSAR Approaches."
As recently as 25 years ago, applying read-across techniques to predict the toxicity of a target chemical was rare. Trailblazers in 1980s, such as the US Environmental Protection Agency (US EPA) IRIS Program and World Health Organization (WHO) JECFA Committee, established reference doses and acceptable daily intakes on a select number of chemicals, such as 2-butanone and allyl heptanoate, using read-across. Read-across moved from imaginative to imaginable due to the data requirements of the 2006 European Union (EU) REACH Regulation, which left companies scrambling for studies to fulfill data requirements.
Presenters at the read-across workshop demonstrated that read-across is a valuable technique; however, each described challenges yet to be overcome, including the incorporation and interpretation of ’omics data, improving reproducibility, and quantifying uncertainty in read-across predictions.
Nicholas Ball, Dow Chemical, described Dow’s journey applying read-across techniques to P-series and E-series glycol ethers, including the challenges using experimental tools, such as in vivo transcriptomics, to cluster apparently similar chemicals (hint: when analyzing materials, start from the ground up and analyze all materials using the same study design and variables).
Emilio Benfenati, Mario Negri Institute, communicated that reproducibility in read-across is heavily influenced by the prediction tools used and described a survey conducted as part of the EU CALEIDOS Project, where participants using ToxRead software were in agreement for predicting the mutagenicity of seven substances, in contrast to disparate mutagenicity predictions using OECD QSAR Toolbox. Besides ToxRead, VEGAHUB features ToxWeight, a relatively new tool that combines read-across and QSAR predictions from VEGA and ToxRead, as well as ToxDelta (available in beta form on the website), a tool that evaluates dissimilarity between substances.
Grace Patlewicz, US EPA, identified the major challenge in read-across software: None include the evaluation of uncertainty in their read-across predictions. She highlighted the US EPA’s work developing GenRA (generalized read-across) software, and detailed GenRA’s use of in vitro bioactivity data from the US EPA ToxCast program as part of facilitating read-across predictions. Ideally, objective read-across functionality from GenRA will be incorporated into the US EPA CompTox dashboard.
As it is our Society’s annual meeting, it is only fitting to quote from one of toxicology’s founding fathers, Paracelsus, to remind ourselves of the importance of striving for the use of new and better methods and tools, with an eye towards tomorrow at all times: “Time is a brisk wind, for each hour it brings something new... but who can understand and measure its sharp breath, its mystery, and its design?”