Résumé
Chronic exposure to complex mixtures of chemical contaminants (xenobiotics) is suspected to contribute to the onset of chronic diseases. The technological advances high-resolution mass spectrometry (HRMS), as well as the concept of exposome, have set the stage for the development of new non-targeted methods to characterize human exposure to xenobiotics without a priori. These innovative approaches may therefore allow changing scale to identify chemical risk factors in epidemiological studies. However, non-targeted approaches are still subject to a number of barriers, partly linked to the presence of these xenobiotics at trace levels in biological matrices. An optimization of every analytical (i.e. sample preparation) and bioinformatical (i.e. data processing, annotation) step of the workflow is thus required. The main objective of this work is to implement an HRMS-based non-targeted workflow applicable to epidemiological studies, to provide an operational solution to characterize the internal chemical exposome at a large scale. The undertaken developments allowed proposing a simple sample preparation workflow based on two complementary methods to expand the visible chemical space (up to 80% of features specific to one method). The optimization of various data processing tools, performed for the first time in an exposomics context, allowed demonstrating the necessity to adjust key parameters to accurately detect xenobiotics. Moreover, the development of a software to automatize suspect screening approaches using MS1 predictors, and of algorithms to compute confidence indices, allowed efficiently prioritizing features for manual curation. A large-scale application of this optimized workflow on 125 serum samples from the Pélagie cohort allowed demonstrating the robustness and sensitivity of this new workflow, and enriching the documented chemical exposome with the uncovering of new biomarkers of exposure.
Source: http://www.theses.fr/2022HESP0002
.