2025 AOCS Annual Meeting & Expo.
Analytical
Berta Torres Cobos, PhD (she/her/hers)
Postdoctoral Researcher in Food Science
University of Barcelona
Beatriz Quintanilla Casas
Postdoctoral Researcher in Food Science
University of Copenhagen
Copenhagen, Denmark
Mercè Rovira
Researcher
Institute of Agrifood Research and Technology
Constantí, Catalonia, Spain
Agustí Romero
Researcher
Institute of Agrifood Research and Technology
Constantí, Catalonia, Spain
Francesc Guardiola
Full Professor of Nutrition and Food Science
University of Barcelona
Santa Coloma de Gramanet, Catalonia, Spain
Stefania Vichi
Associate Professor of Nutrition and Food Science
University of Barcelona
Santa Coloma de Gramanet, Catalonia, Spain
Alba Tres
Associate Professor
University of Barcelona
Santa Coloma de Gramanet, Catalonia, Spain
Food authentication is crucial to combating fraud and ensuring consumer trust, particularly for high-value products such as extra virgin olive oil (EVOO) and nuts, which are prone to mislabelling by origin or cultivar. This study compares analytical approaches—targeted profiling, fingerprinting, and untargeted profiling—applied to two case studies: (1) geographical authentication of EVOO using sesquiterpene hydrocarbons (SHs), and (2) varietal and geographical authentication of hazelnuts based on their unsaponifiable fraction.
In the EVOO study, 82 samples from seven countries were analysed by headspace-solid-phase microextraction coupled with gas chromatography-mass spectrometry (HS-SPME-GC-MS). Data were evaluated using both targeted profiling and fingerprinting approaches. Fingerprinting achieved 100% correct classification for all geographical origins in internal validation using partial least squares discriminant analysis (PLS-DA), outperforming targeted profiling (46–100%), which faced challenges in peak identification and missed critical information.
For hazelnuts, 176 samples from three geographical origins, two harvest years and different cultivars were analysed by GC-MS. Untargeted profiling was conducted using PARADISe, a PARAFAC2-based tool for efficient deconvolution of complex chromatographic data. Fingerprinting and untargeted profiling were compared using PLS-DA models, with fingerprinting showing higher classification accuracy for cultivar (90%) and geographical origin models (EU vs. non-EU 99%; Spain vs. Italy 95%) than untargeted profiling (cultivar 86%; EU vs. non-EU: 97%; Spain vs. Italy: 97%) in external validation. The exploration of the regression coefficients revealed that fingerprinting extracts more detailed information, while untargeted profiling provides better chemical interpretability and facilitates compound identification.
These findings highlight the effectiveness of fingerprinting in extracting key information for food authentication purposes, emphasizing the importance of selecting suitable analytical strategies tailored to specific authenticity goals, ensuring high accuracy and reliability.