Mining the European chemical safety registrations could help identify chemical toxicity of untested chemicals
The chemical universe can now be mined. This is thanks to a new database complemented with the guidance developed to help make sense of it. This new database will be presented at a stakeholders’ event in Brussels to be held on 26 February 2016 and in Washington on 1 March 2016. The first four articles analysing the database and two guidance documents are due to be published online on 12 February 2016, to coincide with their presentation at the American Association for the Advancement of Science (AAAS) conference in Washington, DC, USA.
The motivation for adopting such an approach was simple. We first just wanted to understand, how the chemical universe looks like with all the animal test data currently submitted to the public database of ECHA, the European Chemical Agency in Helsinki. Due to the requirements of the REACH Directive, about 800,000 studies on 10,000 chemicals have already been registered in the ECHA database by December 2014. This is the largest toxicological database in the world.
It took Tom Luechtefeld, one of my PhD students at John Hopkins Bloomberg School of Public Health, one year and about 500 pages of programming to train a computer to make sense of all the free text information stored in the public database of ECHA, the European Chemical Agency in Helsinki. Until then, this information was not computer readable. Previously, to state that a given chemical is an eye irritant, the information on file might read: “category 1”, “corrosive”, “cat. I”, “highly irritating.”
Cutting animal testing
What we found is astonishing: much less chemicals are labelled with the different hazards than expected. Only one in four chemicals produced any effect in rabbit eyes, for example. But it is shocking, how often some chemicals have been tested in the same test, for example on average three times in rabbit eyes. Two chemicals present in the database were tested 90 times, sixty-nine chemicals have been tested 45 times. This shows the importance of the mutual acceptance of data between countries brokered by the OECD. It also demonstrates the role of REACH in requesting that all companies interested in one substance to work together and share their data.
However, this waste of animals now shows, for the first time with so much evidence, how bad this test is. Until now, it has been pretty much a lottery! If a substance was a severe irritant in the first test, there is a 20% probability that it is mild in the repetition and 10% that it shows as non-irritant. The other way around, many hazardous substances will go undetected. These obvious limitations should open up for much faster replacement of the rabbit eye test. Many alternative methods have proven better than just the reproducibility of the animal test.
The full potential of the new database lies in computational toxicology. Similar chemicals have similar toxicity. This is called read-across. In other words, it is possible to conclude on the properties of non-tested chemicals based on the tested properties of a group of chemicals with structural similarities.
Thus, our Hopkins team created a similarity map of chemicals. It shows strong proof that toxic properties cluster. This means that we can now tell their properties analysing their tested neighbours. This approach even overcomes the problems of the lottery. That is, if some neighbours were wrongly classified, others, similar ones, will show the actual result. If the vicinity of the chemical is contradictory or ambiguous, animal testing might still be required, but with the growth of the database the number of such situations should decline.
I hope that some concerns by ECHA currently preventing the database to be made fully public can soon be overcome. It would be a scandal, if we carry on testing on animals while sufficient information is publicly available. This is not the spirit of the REACH and the animal welfare legislation in Europe. More than 20,000 additional chemicals will be registered by 2018.
Guidance for read-across
To make full use of read-across and the database, a team of thirty experts have developed, in parallel, guidance on how to do adequate read-across with high certainty. This guidance is published together with the data-mining and will be presented at stakeholder fora in Brussels and Washington. But our Hopkins team wants to go one step further: a web-based tool to automatically do the read-across. Nobody should test on animals just because it is too complicated to try the existing data first. A spin-off company, tentatively called ToxTrack, is currently being set up and looking for investors.
And there are many more potential applications outside of the REACH directive. There are similar programs emerging in America and Asia, such as the US Toxic Substance Control Act reauthorisation as well as Korea and Taiwan REACH or similar plans in China. Imagine, for example, that a company wants to go for greener chemistry, to get rid of the toxicants to protect workers and consumers. How would they know where to start? Such a web-based tool could help identify potential problems. Similarly, a chemist could assess the toxicity of their structure into the similarity map, even before synthesizing it. If there are flags for toxicity, they might choose a different molecule, instead of discovering toxic liabilities at the end of a costly product development cycle. In essence, this database allows us to learn from the past to avoid animal testing in the future.
Thomas is head of the Centers for Alternatives to Animal Testing (CAAT) at Johns Hopkins Bloomberg School of Public Health, Baltimore, Maryland, USA and for CAAT Europe in Constance, Germany
Photo credit: SUWIT NGAOKAEW via Shutterstock