Diet-Related Metabolomic Signature of Long-Term Breast Cancer Risk Using Penalized Regression: An Exploratory Study in the SU.VI.MAX Cohort


Diet has been recognized as a modifiable risk factor for breast cancer. Highlighting predictive diet-related biomarkers would be of great public health relevance to identify at-risk subjects. The aim of this exploratory study was to select diet-relatedmetabolites discriminating women at higher risk of breast cancer using untargeted metabolomics.


Baseline plasma samples of 200 incident breast cancer cases and matched controls, from a nested case-control study within the Supplémentation en Vitamines et Minéraux Antioxydants (SU.VI.MAX) cohort, were analyzed by untargeted LC-MS. Diet-relatedmetabolites were identified by partial correlation with dietary exposures, and best predictors of breast cancer risk were then selected by Elastic Net penalized regression. The selection stability was assessed using bootstrap resampling.


595 ions were selected as candidate diet-related metabolites. Fourteen of them were selected by Elastic Net regression as breast cancer risk discriminant ions. A lower level of piperine (a compound from pepper) and higher levels of acetyltributylcitrate (an alternative plasticizer to phthalates), pregnene-triol sulfate (a steroid sulfate), and 2-amino-4-cyano butanoic acid (a metabolite linked to microbiota metabolism) were observed in plasma from women who subsequently developed breast cancer. This metabolomic signaturewas related to several dietary exposures such as a “Western” dietary pattern and higher alcohol and coffee intakes.


Our study suggested a diet-related plasma metabolic signature involving exogenous, steroid metabolites, and microbiota-related compounds associated with long-term breast cancer risk that should be confirmed in large-scale independent studies.


These results could help to identify healthy women at higher risk of breast cancer and improve the understanding of nutrition and health relationship.