Spectral Fingerprinting for Determination of Panax Species, Growing Site, and Treatment

The purpose of this project was to develop a simple chemical identification method for three Panax species. Spectral fingerprints of Panax ginseng, P. notoginseng, and P. quinquefolium grown in Canada, China, and the US were obtained using ultraviolet (UV), near-infrared (NIR), and mass spectrometry (MS) and analyzed by principal components analysis (PCA), soft independent modeling of class analogy (SIMCA), partial least squares-discriminant analysis (PLS-DA), and fuzzy rule-building expert system (FuRES). UV and MS data were obtained directly from a methanol-water extract (no separation) and the NIR data were obtained from finely powdered solids. MS data were collected on LCQ and Exactive instruments (Thermo Scientific) with injection across a guard column. UV data were collected on the LCQ from a diode array detector (DAD) downstream of the guard column and from 96-well plate reader (96WPR) (Molecular Devices). With PCA, all methods were capable of discriminating between the 3 species. P. notoginseng and P. quinquefolium were tightly clustered, compared to the P. ginseng. Preparation methods for P. ginseng (i.e. red, white, or raw) were easily distinguished. MS and NIR could differentiate between P. quinguefolium grown in Canada, China, and the US. SIMCA, based on one principal component, was accurate at better than 96% in discriminating between the 3 species. PLS-DA modeling provided 100% accuracy in predicting the species from MS data and 94–98% for UV data. FuRES was 100% accurate for MS and 82–97% accurate for UV. There were differences in the UV data sets since the transmission of the 96WPR ended at 230nm. Transmission for the DAD extended below 200nm and provided better discrimination. These data suggest that P. quinquefolium can be distinguished from the other species and the growing site identified.