Purity assessment of ginsenoside Rg1 using quantitative ¹H nuclear magnetic resonance

Ginseng herbs comprise a group of the most popular herbs, including Panax ginseng, P. notoginseng and P. quinquefolius (Family Araliaceae), which are used as traditional Chinese medicine (TCM) and are some of the best-selling natural products in the world. The accurate quantification of ginsenoside Rg1 is one of the major aspects of its quality control. However, the purity of the commercial Rg1 chemical reference substance (CRS) is often measured with high-performance chromatography coupled with an ultraviolet detector (HPLC-UV), which is a selective detector with unequal responses to different compounds; thus, this detector introduces probable error to purity assessments. In the present study, quantitative nuclear magnetic resonance (qNMR), due to its absolute quantification ability, was applied to accurately assess the purity of Rg1 CRS. Phenylmethyl phthalate was used as the internal standard (IS) to calibrate the purity of Rg1 CRS. The proton signal of Rg1 CRS in methanol-d4 at 4.37ppm was selected to avoid interfering signals, enabling accurate quantitative analysis. The relaxation delay, number of scans, and NMR windowing were optimized for data acquisition. For post-processing, the Lorentz/Gauss deconvolution method was employed to increase the signal accuracy by separating the impurities and noise in the integrated region of the quantitative proton. The method validation showed that the developed method has acceptable sensitivity, linearity, precision, and accuracy. The purity of the commercial Rg1 CRS examined with the method developed in this research was 90.34±0.21%, which was obviously lower than that reported by the manufacturer (>98.0%, HPLC-UV). The cross-method validation shows that the commonly used HPLC-UV, HPLC-ELSD (evaporative light scattering detector) and even LC-MS (mass spectrometry) methods provide significantly higher purity values of Rg1 CRS compared with the qNMR method, and the accuracy of these LC-based methods largely depend on the amount of the sample that was loaded and the properties of the impurities.