(PTFS10-03-24) Profiling and Clustering of Herbs and Spices Using the Agilent Q-TOF LC/MS System and Mass Profiler Professional Software

Objectives: The objective of this study was to conduct non-targeted analysis in order to classify and cluster different herbs and spices (H/S) based on phytochemicals using multivariate statistical analysis.

Methods: Seventeen commercially available H/S (cinnamon, coriander, ginger, cumin, parsley, black pepper, garlic, turmeric, onion powder, paprika, chili powder, rosemary, cilantro, oregano, basil, red pepper, and bay leaf, McCormick brand) from 7 botanical families (Lauraceae, Apiaceae, Zingiberaceae, Piperaceae, Amaryllidaceae, Solanaceae, and Lamiaceae), and from 6 parts of the plant (bark, seed, root, leaf, fruit, and bulb) were extracted with acidified 70% acetone in triplicates. Non-targeted analysis was performed on ultra-high- performance liquid chromatography (UHPLC) coupled with electrospray ionization quadrupole time of flight (Q-TOF) system. Data were extracted on Agilent MassHunter Profinder using recursive feature extraction and imported into Agilent Mass Profiler Professional (MPP, version 14.0) software for statistical analysis.

Results: A total of 437 compound groups were obtained from 17 H/S after data extraction on Profinder. Alignments of mass and retention time were performed on MPP. Three subsets of compounds were obtained by filtering out abundance level less than 10⁶ counts, and not present in all replicates of the given H/S, or part of the plant, or botanical family, resulting in 216, 40 and 20 unique entities, respectively. Principal component analysis (PCA) performed on the subsets of the given H/S, part of the plant, and botanical family showed the first three PC explained 42.49, 66.52 and 63.22% of the variance among the data. Hierarchical clustering analysis showed that samples were clustered by the botanical family based on phytochemical profile for Solanaceae, Lauraceae, and Amaryllidaceae.

Conclusions: Multivariate statistical analysis indicates similarity of the phytochemical profile in family of Solanaceae, Lauraceae or Amaryllidaceae, while other H/S from the same botanical family showed a difference. This study demonstrates the feasibility of developing a model that could be utilized to characterize various phytochemicals, classify place of origin, and detect adulteration in H/S.

Funding Sources: This project was supported by a gift from McCormick Science Institute and various donor funds.