Methodological Research On The Qualitative Analysis Of Ephedra Plants With HPLC And Chemometrics

Ephedra plants (Ephedraceae) are widely distributed in china. Theaerial parts of Ephedra sinica plants are used as a diaphoretic, anti-asthmaticand diuretic. The levels and types of plant secondary metabolites aredecided primarily by the species and habitats of Ephedra plants andmodulated by the picking times of day. Therefore, it is important toestablish an efficacious method for the quality control of species, habitatsand picking times of Ephedra plants.Fingerprint technology, which emphasizes on the systemiccharacterization of compositions of TCM, is widely used for the qualityassessment of herbs. High performance liquid chromatography (HPLC)fingerprint analysis has been highly recommended due to precision,sensitivity, reproducibility. Chemometric methods were used to processchromatograms for reducing retention time shift, baseline drift and noise, topre-process chromatographic data for removing within-class variance andredundant information, and to analyze chromatographic data.In this work, the feasibility of the qualitative analysis of species,habitats and picking times of Ephedra was evaluated with HPLC fingerprint and chemometric methods. Three species, two habitats and twopicking times of Ephedra plants were discriminated by the models ofback-propagation artificial neural network (BP-ANN) and discriminantanalysis (DA).OBJECTIVES:1. Establish a method to discriminate different species of Ephedraplants with HPLC fingerprint.2. Establish a method to discriminate different habitats of Ephedraplants with HPLC fingerprint.3. Establish a method to discriminate different picking times ofEphedra plants with HPLC fingerprint.METHODS:1. Measurement of HPLC fingerprints of Ephedra plants.The separation was performed using a C18column (300mm×3.9mmid,10μm) with a Phenomenex C18guard column (4.0mm×3.0mm id).The mobile phase consisted of water-phosphoric acid-triethylamine (A;100:0.1:0.1, v/v/v), water-phosphoric acid (B;100:0.1, v/v) and methanol(C). A:B:C was as follows:0.00min,98:0:2;17.00min,98:0:2;17.01min,0:98:2;22.00min,0:98:2;27.00min,0:75:25;50.00min,0:35:65. Theflow rate was1.0mL/min. The column temperature was30°C. Thedetection wavelengths were210nm from0to17min and350nm from17to50min. The loading volume was20μL. 2. Discrimination of different species of Ephedra plants.The three different species of Ephedra plants, Ephedra sinica,Ephedra intermedia and Ephedra equisetina, were discriminated byBP-ANN and DA.3. Discrimination of different habitats of Ephedra sinica plants.The Ephedra sinica plants from Shanxi and from Inner Mongolia,were discriminated by BP-ANN and DA.4. Discrimination of different picking times of Ephedra sinica plantsfrom Shanxi.The Ephedra sinica plants from Shanxi picked in the morning section(10:00–11:30am) and in the afternoon section (4:30–5:00pm) werediscriminated by BP-ANN and DA.RESULTS:1. An HPLC method for the fingerprints of the Ephedra plants.Ephedrine peak was successfully separated from pseudoephedrinepeak. The RSD (n=3) of peak areas of ephedrine and pseudoephedrine wererespectively2.6%and2.1%for the variation in flow rate (±0.02mL/min).The RSD (n=3) of peak areas of ephedrine and pseudoephedrine wereseparately0.4%and0.2%for the variation in column temperature (±1°C).2. Discrimination of different species of Ephedra plants.The HPLC fingerprint data of three different species of Ephedraplants, Ephedra sinica, Ephedra intermedia and Ephedra equisetina was reduced to8principle components (PCs), the cumulative contribution ratewas99.8%. The network structure parameters of BP-ANN model was8-6-1, the prediction accuracy was94.4%. The performance index of DAmodel was88.5%.3. Discrimination of two different habitats of Ephedra sinica plants.The HPLC fingerprint data of Ephedra sinica plants from Shanxi andfrom Inner Mongolia was reduced to11PCs, the cumulative contributionrate was99.9%. The network structure parameters of BP-ANN model was11-8-1, the prediction accuracy was88.9%. The performance index of DAmodel was84.0%.4. Discrimination of two different picking times of Ephedra sinicaplants from Shanxi.The HPLC fingerprint data of Ephedra sinica plants from Shanxipicked in the morning section (10:00–11:30am) and in the afternoonsection (4:30–5:00pm) was reduced to12PCs, the cumulative contributionrate was100.0%. The network structure parameters of BP-ANN model was12-7-1, the prediction accuracy was83.3%. The performance index of DAmodel was82.8%.CONCLUSIONS:The results showed that the species, habitats and picking times ofEphedra plants could be identified by the proposed approaches, HPLCcoupled with chemometric methods. The nonlinear CP-ANN models are significantly better than the linear DA models for the qualitative analysis ofherbs with multiple components although both CP-ANN and DA aregenerally satisfactory for their intended purpose.