Signal-Corrected LC-MS/MS Approaches in Amino Acid and Biogenic Amine Profiling for Chemometric Characterization of Commercial Dark Chocolates According to Cocoa Content and Manufacturer.

The chemical composition of chocolate is influenced by multiple factors, including cocoa genotype, fermentation, roasting, and formulation, which complicate the identification of reliable compositional descriptors. In this study, amino acid and biogenic amine profiles were determined by LC-MS/MS in 91 commercial dark chocolates (56-100% cocoa content) to assess their potential for product characterization. Because the analytical sequence extended over several days, different signal-correction strategies (including internal standard normalization and quality-control-based drift correction) were evaluated to compensate for instrumental variability prior to chemometric analysis. Signal reductions of up to 91% were observed, and the selected correction procedures improved measurement precision by up to 18-fold. A total of 19 amino acids and 8 amines were detected, with tyrosine, hydroxyproline, proline, leucine, phenylalanine, and valine being the most abundant compounds. Principal Component Analysis revealed a compositional gradient primarily associated with cocoa content, whereas classifications based on cocoa variety, geographical origin, and certification status showed limited discrimination. In contrast, Partial Least Squares-Discriminant Analysis achieved a balanced classification accuracy of 75.2% under repeated M-fold cross-validation when classifying samples according to manufacturer. Furthermore, permutation tests were carried out, confirming that the observed classification performance was highly unlikely to arise by chance. These findings indicate that amino acid and biogenic amine profiles in commercial dark chocolates are influenced mainly by cocoa content and manufacturer-related factors, whereas the effects of cocoa origin and variety appear less evident within the products evaluated.