Abstract
A targeted metabolomics LC-ESI-QqQ-MS/MS application for the determination of cocoa shell based on 15 non-polar key metabolites was developed, validated according to recognized guidelines, and used to predict the cocoa shell content in various cocoa products. For the cocoa shell prediction, different PLSR models based on different cocoa shell calibration series were developed and their suitability and prediction quality were compared. By analysing samples from different origins and harvest years with known shell content, the prediction model could be confirmed. The predicted shell content could be verified with a deviation of about 1% cocoa shell. The presented method demonstrates the suitability of the targeted application of metabolomic profiling for the determination of cocoa shell and its applicability in routine analysis is discussed.