HPLC fingerprint and UV–Vis spectroscopy coupled with chemometrics for Curcumae radix species discrimination and three bioactive compounds prediction

Abstract

Curcumae Radix with four botanical origins (Wenyujin, Huangyujin, Guiyujin, Lvyujin) has a long medical history in China. The four species have not been differentiated well in clinical application, though there were differences in pharmacological effects among them. For the aim of fast discrimination and quantitative analysis of Yujin, HPLC (high performance liquid chromatography) fingerprint and UV–Vis (ultraviolet–visible) spectroscopy technologies coupled with chemometrics were employed. The HPLC fingerprints of 26 Guiyujin, 11 Huangyujin, 11 Lvyujin, 29 Wenyujin and full 77 samples were separately established, with good similarities among same species except Wenyujin and poor consistence in full samples. UV–Vis spectra showed obvious three absorption peaks in the spectra of Huangyujin samples and only one in that of the rest samples. LDA (linear discriminant analysis) and Bagged Trees models established based on HPLC fingerprint data attained higher than 90% discrimination accuracy. Based on the spectral data, complete separation for the four species of Yujin was failed to achieve by PCA (principal component analysis), and the KNN (k-nearest neighbors) and SVM (support vector machine) models achieved no<90% discrimination accuracies. Furthermore, the SVMR (support vector machine regression) and Bagged Trees Regression models used to quickly predict the contents of germacrone, curcumenol and curdione performed well, with relatively lower RMSE (root mean square error) and higher R2 (coefficient of determination) values. The present study proposed relatively rapid methods for the determination of bioactive compounds and discrimination of Yujin.