Italian researchers have created a platform for prioritising possible persistent, bioaccumulative and toxic (PBT) chemicals for further assessment.
Working on behalf of Germany's Federal Environment Agency (UBA), a team from the Mario Negri Institute for Pharmacological Research in Milan has brought together a battery of in silico tests to screen for chemicals that may be "of concern for the environment and human health".
As part of project Prometheus, the researchers address P, B and T properties separately before integrating the results.
For persistence, the team has built models using data on ready biodegradability (RB) from different sources. Meanwhile, for bioaccumulation, it constructed a dataset with information on more than 700 molecules. The researchers then used software called SARpy to find structural fragments related to biodegradability. From this they compiled general rules for an in silico classification model.
Finally, for toxicity, the researchers chose to focus on fish data. This was mainly because the quality of existing models on aquatic invertebrates and algae remain "quite poor". In future, models for daphnia and algae can be added, they suggest.
Using decision trees, the researchers assign reliability values to data, giving a score between 0 and 1 for each endpoint, where 1 is optimal. They then combine P, B, or T values with the reliability scores.
For the final PBT assessment, the team uses a multiple criteria decision making (MCDM) method to bring together P, B and T predictions to give an overall evaluation, with a score close to 1 being PBT.
Led by Emilio Benfenati, the researchers have tested their platform using non-PBT chemicals and a set of substances already assigned as PBT by regulatory authorities.
As hoped, the platform assigned a ranking close to 1 for all chemicals "officially" labelled as PBT, and close to zero for non-PBT chemicals. The team concludes that the integrated model "separates successfully PBT vs non-PBT compounds".
The platform can be refined and improved, for example by taking into consideration additional endpoints such as endocrine disruption and human toxicity, they say. It should be possible to use the platform to create a PBT prioritisation tool that would be freely available to users, they add.