35018

Rapid detection of active ingredients in cortex moutan by near infrared diffuse reflection spectral technique

Yang Jidong 1 ; Wu Limin 2
Favorite this paper

1.Introduction
The cortex moutan is a kind of traditional Chinese medicine used widely in clinical treament. The medicine of different sources has various content and components, paeonol and paeoniflorin are its two main active ingredients, which will directly affect its quality and curative effects. It is reported that detection of the medicine includes UVS, CE, GC, and HPLC-MS, etc. Therefore, it is necessary to establish a simple and rapid quantitative and qualitative evalutation method.
2. Results and discussion
2.1 Experimental procedure
① Simultaneous determination of chemical values of paeonol and paeoniflorin by HPLC; ② Near infrared spectra collection; ③Model of cortex moutan calibration and prediction.
2.2 HPLC for determination of paeonol and paeoniflorin
2.3 Collection of near infrared diffuse reflection spectra
Collect diffuse reflection spectra of 121 cortex moutan samples from 6 origins, respectively. The absorption peaks show that combined frequency absorption of C-H key and O-H key for stretching vibration and bending vibration in molecular structure and in hydrone structure were located in wavenumber 4259 cm-1, 4744 cm-1and 5165 cm-1, and frequency doubling absorption of O-H key for same two vibrations in molecular structure were located in 6835 cm-1 .
2.4 Buildup of quantitative analysis model of paeonol and paeoniflorin
2.4.1 Near infrared spectrum pretreatment
Compared with several pretreatment such as original spectrum, vector normalization (vn), multiple scatter correction (MSC), first derivative, etc., RMSECV, RMSEP, R2 were used as evaluation index. Compared with original spectrua again, results shown spectra peaks become sharper and spectrogram difference of different samples more apparent, which avails to extract useful information.
2.4.2 Buildup of quantitative analysis model
OPUS 5.0 analysis software was adopted to set up PLS1 quantitative model of paeonol and paeoniflorin, respectively. Through internal interaction validation, influence of number of principal components (PCs) on mean square (RMSECV) was discussed to obtain the best parameters for buildup of quantitative calibration model.
2.4.3 External validation and method evaluation of NIR quantitative analysis model
prediction set samples were forecasted, compare NIR prediction value with their HPLC value the correlation coefficient R were measured as 0.9877 and 0.9971, respectively, which shown NIR prediction value is well correlated with HPLC measured value. external prediction root mean square error (RMSEP) of paeonol and paeoniflorin were calculated as 0.0662 and 0.0416, respectively. The results shown this method has high accuracy.
Take samples from Sichuan and Shandong, RSD of paeonol and paeoniflorin obtained were less than 1.0%, the results shown this method has good precision.
2.5 Buildup of cluster analysis model and external validation
Use OPUS 5.0 for cluster analysis of calibrating samples of cortex moutan. In spectral range 8427.4~3999.6 cm-1, to calculate spectral distance and inter-class distance, a higher recognition accuracy rate was obtained. Then mix forecast sample for cluster analysis prediction, samples clusters were divided into types to validate origins respectively.
3.Conclusion
This experiment combined NIR spectral technique with PLS and cluster analysis method, a new method for simultaneous determination of two main components and quickly identification of the quality of the medicine was established.