New strategies for the differentiation of fresh and frozen/thawed fish: Non-targeted metabolomics by LC-HRMS (part B)

Analytical methods able to differentiate fresh from frozen/thawed fish products are needed for suppliers and official authorities to prevent food frauds. In this study we applied a metabolomics approach to identify markers of freezing/thawing in skinless European sea bass (Dicentrarchus labrax) fish fillets. To this end, one hundred and twenty fish samples (sixty fresh and sixty frozen/thawed) were analysed in positive and negative ion modes by liquid chromatography coupled to high resolution mass spectrometry (LC-HRMS). A combination of a supervised classification method by partial least square-discriminant analysis (PLS-DA), and Recursive Feature Elimination (RFE) allowed us to discover the most informative metabolites useful to codify fresh and defrosted European sea bass. Eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA) metabolites were identified as the most relevant predictors. These were used to build a model capable of classifying European sea bass samples according to their fresh or frozen/thawed condition. The model showed accuracy, sensitivity and specificity of 100 %, when tested to an independent set of European sea bass undergoing the same experimental conditions. EPA and DHA were found to be altered even in a set of twenty fresh and frozen/thawed Atlantic salmon (Salmo salar), opening the way for the development of a targeted method aimed at providing robust species-independent tests to be applied in the seafood chain by producers and competent authorities.