Molecules (Dec 2023)

Liquid Chromatography–High-Resolution Mass Spectrometry (LC-HRMS) Fingerprinting and Chemometrics for Coffee Classification and Authentication

  • Nerea Núñez,
  • Javier Saurina,
  • Oscar Núñez

DOI
https://doi.org/10.3390/molecules29010232
Journal volume & issue
Vol. 29, no. 1
p. 232

Abstract

Read online

Nowadays, the quality of natural products is an issue of great interest in our society due to the increase in adulteration cases in recent decades. Coffee, one of the most popular beverages worldwide, is a food product that is easily adulterated. To prevent fraudulent practices, it is necessary to develop feasible methodologies to authenticate and guarantee not only the coffee’s origin but also its variety, as well as its roasting degree. In the present study, a C18 reversed-phase liquid chromatography (LC) technique coupled to high-resolution mass spectrometry (HRMS) was applied to address the characterization and classification of Arabica and Robusta coffee samples from different production regions using chemometrics. The proposed non-targeted LC-HRMS method using electrospray ionization in negative mode was applied to the analysis of 306 coffee samples belonging to different groups depending on the variety (Arabica and Robusta), the growing region (e.g., Ethiopia, Colombia, Nicaragua, Indonesia, India, Uganda, Brazil, Cambodia and Vietnam), and the roasting degree. Analytes were recovered with hot water as the extracting solvent (coffee brewing). The data obtained were considered the source of potential descriptors to be exploited for the characterization and classification of the samples using principal component analysis (PCA) and partial least squares–discriminant analysis (PLS-DA). In addition, different adulteration cases, involving nearby production regions and different varieties, were evaluated by pairs (e.g., Vietnam Arabica—Vietnam Robusta, Vietnam Arabica—Cambodia and Vietnam Robusta—Cambodia). The coffee adulteration studies carried out with partial least squares (PLS) regression demonstrated the good capability of the proposed methodology to quantify adulterant levels down to 15%, accomplishing calibration and prediction errors below 2.7% and 11.6%, respectively.

Keywords