Interoperability and multi-source data integration in metabolomics for the identification common Metabomic Syndrome phenotypes

Precision medicine is opening up a new paradigm in the management of chronic diseases, using multiple data (biological, environmental, lifestyle) to propose personalized treatments. In this context, metabolomics, and in particular the untargeted approach, represents a powerful tool for metabolic phenotyping. However, this approach, although promising, has major limitations for large cohort studies, due to its semi-quantitative nature and its sensitivity to analytical errors introducing biases, notably via batch effects, which alter the reproducibility and integration of results. These effects, specific to each metabolite and lacking a global model, restrain inter- study comparisons, increase false discoveries, and limit the clinical translation of identified biomarkers. Thus, the aim of this thesis was to contribute to

1- improving the reproducibility of non-targeted metabolomics data based on mass spectrometry;

2- developing methods for integrating multi-source metabolomics data to identify common phenotypes; and finally,

3- applying the strategy developed to the context of metabolic syndrome. Our results have enabled us to propose an ecosystem of solutions for all the steps involved in the vertical integration of metabolomic data: automated processes for data extraction and quality assessment, correction of batch effects and long-term analytical variability, and data integration methods for multi- cohort metabolomic studies.

Finally, all these methodologies were applied to data from two human cohort studies, to discover common and age-specific metabolic features associated with metabolic syndrome. Our results thus offer promising avenues for the study of longitudinal and integrative data in the field of metabolic research.

Key words: Untargeted metabolomics, reproducibility, multi-source data integration, phenotypes, metabolic diseases

Publications

• Salanon, E., Comte, B., Centeno, D., Durand, S., Pujos-Guillot, E., & Boccard, J. (2024). An alternative for the robust assessment of the repeatability and reproducibility of analytical measurements using bivariate dispersion. Chemometrics and Intelligent Laboratory Systems, 250, 105148. https://doi.org/10.1016/j.chemolab.2024.105148
• Salanon, E., Comte, B., Centeno, D., Durand, S., Pujos-Guillot, E., & Boccard, J. A workflow for the multivariate and univariate assessment of the reliability of batch correction methods centered on the Batch conformity index. Submitted to Metabolomics.
• Salanon, E., Comte, B., Centeno, D., Durand, S., Boccard, J., & Pujos-Guillot, E. Improving metabolomics data comparability without long-term quality controls using a post-acquisition correction strategy. Accepted in Analytica Chimica Acta.
• Salanon, E., Jules, E., Comte, B., & Boccard, J., Pujos-Guillot, E. A cooperative learning framework for the integration of metabolomic data from cohorts and common phenotype identification. In revision in Computational and Structural Biotechnology Journal
• Salanon, E., Boccard, J., Durand, S., Comte, B., & Pujos-Guillot, E. (In preparation). Towards automated decision support in the analytical reliability assessment of biomarker candidates: A multicriteria optimization framework.
• Salanon, E., Fu, A., Apte, A. P., Mahmoud, U., Belkhatir, Z., Shukla-Dave, A., & Deasy,
• J. O. (2024). Cancer radiomic feature variations due to reconstruction kernel choice and integral tube current. bioRxiv. https://doi.org/10.1101/2024.06.04.596806

Oral communications

• Elfried   Salanon,   Blandine   Comte,   Estelle   Pujos-Guillot,  Julien        Boccard.  Assessment  of  repeatability  and  reproducibility  in  untargeted  LC/MS metabolomics:  beyond  the  limits  of  the  relative  standard  deviation. Metabolomics 2023, Niagara Falls, Canada, 18-22 June 2023.
• Elfried Salanon, Blandine Comte, Estelle Pujos-Guillot, Julien Boccard. Towards automated decision support in the analytical reliability assessment of biomarker  candidates:  a  multi-criteria  optimization  framework. 34th International Symposium on Pharmaceutical and Biomedical Analysis (PBA 2024), Geneva, Switzerland, 9-12 September 2024.
• Elfried Salanon, Etienne Jules, Blandine Comte, Julien Boccard, Estelle Pujos- Guillot. Enhancing Biomarker Discovery in Metabolomics through Cooperative Learning-Based Vertical Data Integration. Metabolomics 2025, Prague, Czech Republic, 22-26 June 2023.
• Elfried Salanon, Julien Boccard, Blandine Comte, Estelle Pujos-Guillot. Cross- study metabolomics data integration for the identification of common metabolic syndrome phenotypes. Submitted. NuGOweek 2025, Dublin Ireland, 22-25 Sept 2025

Patent EU 25305946.3

• Salanon E, Comte B, Boccard J, Pujos-Guillot E. Method for the harmonization of metabolomic measurements from liquid chromatography/high-resolution mass spectrometry.