Submitted to: Journal of Food Composition and Analysis
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 6/15/2017
Publication Date: 12/1/2017
Citation: Harnly, J.M., Lu, Y., Sun, J., Chen, P. 2017. Botanical supplements: detecting the transition from ingredients to supplements. Journal of Food Composition and Analysis. 64:85-92. https://doi.org/10.1016/j.jfca.2017.06.010.
DOI: https://doi.org/10.1016/j.jfca.2017.06.010 Interpretive Summary: Authentication of dietary supplements is difficult because their chemical profile is different from the original botanical ingredient from which they are made. The differences arise from the extraction process and possible addition of other ingredients. This study develops a flow injection mass spectrometric method for identifying those components found in both the original ingredient and the finished commercial supplement. Cross correlation of the mass spectra of the ingredient and the supplement produces a correlation spectrum for a subset of components common to both materials. This correlation spectrum is used as a template that is applied to the spectra of the reference botanical ingredients and the supplements. The resulting supplement spectra are compared to a model constructed for the reference ingredients using soft independent modeling of class analogy (a one class classifier). This makes it possible to verify that ingredients listed on the label are in the supplement.
Technical Abstract: Methods were developed using flow injection mass spectrometry (FIMS) and chemometrics for the comparison of spectral similarities and differences of 3 botanical ingredients and their supplements: Echinacea purpurea aerial samples and solid and liquid supplements, E. purpurea root samples and solid supplements, and E. angustifolia root samples and solid and liquid supplements. The similarities of the Echinacea ingredients and supplements were first put in perspective by comparison with samples of Actaea and Panax species. Principal component analysis (PCA) showed that 3 genus, provided separate clusters and within the Echinacea cluster, the supplements appeared closely related to the species and plant part appearing on their label. However, soft independent modeling of class analogy (SIMCA) and nested pooled analysis of variance (pANOVA) showed that the species, plant parts and supplements were statistically different. PCA loadings and ANOVA identified compounds that were different between the ingredients and supplements and spectral correlation (i.e., point by point multiplication of spectra) provided simplified spectra that identified compounds found in both the ingredients and supplements. Correlation spectra were used as filters to produce modified spectra of the ingredients and supplements that could be used to verify the presence of specific Echinacea species and plant parts in the commercial supplements. Unfiltered spectra verified the presence of Echinacea in only 3 of the 25 supplements, whereas filtered spectra identified Echinacea in 14 of the 25 supplements.