Toxicological evaluation of complex mixtures: fingerprinting and multivariate analysis

Area of Science:

• Environmental Chemistry
• Toxicology
• Chemometrics

Background:

• Complex mixtures, such as exhaust particle extracts, pose challenges for toxicological evaluation.
• Traditional methods struggle to identify individual contributors to toxicity within these mixtures.

Purpose of the Study:

• To develop and illustrate a strategy for toxicological evaluation of complex mixtures using chemical fingerprinting and pattern recognition.
• To correlate chemical fingerprints with toxicological endpoints, identify major toxicity contributors, and predict the toxicity of new mixtures.

Main Methods:

• Utilized chemical fingerprinting via full scan gas chromatography-mass spectrometry (GC-MS) for characterizing organic extracts of exhaust particles.
• Employed automated curve resolution to process complex GC-MS data into individual compound peaks and spectra.
• Applied Projections to Latent Structures (PLS) regression modeling to link GC-MS data with mutagenicity (Ames Salmonella assay) results.

Main Results:

• The PLS regression model successfully correlated GC-MS data with observed mutagenicity.
• Identified specific chemical peaks that co-vary with mutagenicity, enabling chemical identification from spectra.
• Demonstrated the model's capability to predict mutagenicity from GC-MS chromatograms of new samples.

Conclusions:

• Chemical fingerprinting combined with multivariate data analysis provides a robust strategy for toxicological evaluation of complex mixtures.
• This approach facilitates the identification of toxic components and the prediction of toxicity for environmental and industrial chemicals.