A new international study has shown that metabolomics-based chemical grouping is not only reproducible across laboratories, but can also be explained through known biological and toxicological mechanisms. Building on the successful MATCHING ring trial, the research demonstrates how metabolomic signatures can support more robust chemical grouping and read-across approaches, laying the foundations for a shared knowledge base that could strengthen chemical safety assessment and regulatory decision-making.
Led by the University of Birmingham, the MATCHING consortium has strengthened evidence that metabolomics can reliably group chemicals by their biological effects, supporting more robust chemical safety assessment and read-across. The consortium includes Phenome Centre Birmingham, the National Phenome Centre at Imperial College London, Syngenta, BASF Metabolome Solutions and BASF SE, Vrije Universiteit Amsterdam, and the European Chemicals Agency (ECHA). The work was funded by the Cefic Long-Range Research Initiative (LRI).
Metabolomics measures changes in small molecules involved in biological processes, showing how an organism responds to chemical exposure. It can therefore help group chemicals by their effects, rather than by structure alone. Such evidence allows grouping and read-across from similar substances to fill data gaps in safety data of chemicals, since regulators need strong biological and toxicological evidence to support these approaches.
The MetAbolomics ring-Trial for CHemical groupING (MATCHING), published in 2024, was a blinded international study. Five partner datasets that passed quality control grouped eight substances into the same three effect groups in female and male rats, despite differences in analytical and computational methods. However, the study did not examine the mechanisms behind the grouping.
The newly published follow-on study examined the metabolites and lipids that distinguished the three effect groups from controls. Across all groups and both sexes, the underlying metabolic changes were statistically consistent between the five partners. The researchers linked these changes to known toxicological mechanisms, including androgen receptor activity, peroxisome proliferation and anaemia. Combining this knowledge with consistently detected metabolites and the direction of change improved the ability to distinguish between the three effects.
Ultimately, the work provides a basis for a shared library of reproducible, biologically interpretable metabolomic signatures linked to known toxicological effects. The consortium proposes that laboratories contribute to such a knowledge base. Over time, it could support more consistent safety assessment, stronger mechanistic evidence for grouping and read-across, and wider regulatory use of metabolomics.
Professor Mark Viant, from the University of Birmingham, highlighted the scientific significance of the findings:
“This study shows that metabolomics can do more than group chemicals: the metabolic changes measured by independent laboratories can be linked to the known molecular mechanisms of the chemicals, giving a plausible toxicological basis for read-across.”
Tomasz Sobanski, from the European Chemicals Agency, emphasised the study’s potential regulatory value:
“This work is an important step towards making metabolomics more useful in regulatory toxicology. By combining reproducible laboratory findings with toxicological knowledge, the study offers a more practical way to interpret metabolic changes and strengthen confidence in chemical grouping and safety assessment.”
Notes for editors
- The published paper “Demonstrating the reliability of in vivo metabolomics-based chemical grouping: Part 2 – Consistency of group-specific metabolic effects” appeared in Archives of Toxicology and was published online on 7 October 2026 (DOI 10.1007/s00204-026-04570-1).
- The first MATCHING publication (Part 1), “Demonstrating the reliability of in vivo metabolomics based chemical grouping: towards best practice”, published in Archives of Toxicology in 2024, volume 98, pages 1111–1123 (DOI 10.1007/s00204-024-03680-y).
