Back

A collection of Predicted No-Effect Concentrations of human pharmaceuticals and their metabolites

Giunchi, V.; Poluzzi, E.

2023-12-13 pharmacology and toxicology
10.1101/2023.12.12.571257 bioRxiv
Show abstract

The environmental impact of pharmaceuticals is a growing concern, necessitating methodologies for risk assessment. Current evaluation methods rely on comparing pharmaceutical concentrations in exposed environments with relevant animal or plant tolerance thresholds, often represented by Predicted No-Effect Concentration (PNEC) values. However, challenges arise from the limited accessibility and standardization of PNEC data. This study addresses these challenges by consolidating PNEC values from diverse sources into a unified and accessible tool. Investigated data sources were the NORMAN Ecotoxicology Database, the Swedish National Formulary of Drugs (FASS) website, the EU Watch List working documents, the European Public Assessment Reports (EPAR), the US EPA ECOTOX database, the UBA ETOX database, the WikiPharma database, the AstraZeneca documents, the European Chemical Agency registration dossiers, and the AMR Industry Alliance database. We retrieved 93,287 PNEC values associated with 92,850 substances, primarily medicines or their metabolites. Notably, 352 substances had more than one PNEC value, with the highest discrepancies often attributed to in-silico predicted values. The resulting database, available in the related OSF repository as a spreadsheet file, includes source information and processing scripts and is freely available for risk assessment analyses. While acknowledging limitations, future efforts should prioritize integrating additional data sources, addressing misspellings, and enhancing information on PNEC derivation. Collaboration in PNEC data collection is crucial for advancing collective knowledge in pharmaceutical risk assessment.

Matching journals

The top 8 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.