Application Of Near Infrared Spectroscopy In Quality Evaluation Of Radix Astragali Extract

Radix Astragali extract(RAE)is an extract of the dry root of the leguminous plant Astragalus membranaceus(Fisch.)Bge.var.mongholicus(Bge.)Hsiao or A.membranaceus(Fisch.)Bge..It mainly contains saponins,flavonoids,polysaccharides and other active components.Modern pharmacology and clinical experiments show that RAE has anti-tumor,anti-virus,anti-oxidation,and enhance cellular immunity.As an intermediate,RAE can be used directly without purification and separation.The quality of RAE is greatly affected by Radix Astragali,preparation process and quality standards,the quality of RAE of different manufacturers may vary greatly,which directly affects the quality and clinical efficacy of its products.At present,the Chinese Pharmacopoeia(2015 edition)contains the quality standards of Radix Astragali,but no quality standard for RAE.It has been reported that analytical methods such as high performance liquid chromatography(HPLC),ultraviolet spectrophotometry(UV)and thin layer chromatography are used for the quality control of RAE.However,current analytical techniques have shortcomings such as tedious sample operations,unfriendly environment and time-consuming.Therefore,it is very necessary to establish a rapid and valid analytical method to provide a reference for the rapid quality analysis and evaluation of RAE.Near infrared spectroscopy technology is a fast,non-destructive,green and environmentally friendly analysis method,which can realize the rapid analysis of the quality of RAE.The instruments commonly used in near infrared spectroscopy mainly include two major categories.One is analytical-grade near-infrared spectrometers,such as Antaris Ⅱ Fourier transform spectrometer,which has high resolution and strong predictive ability of the model,but has a large volume and suitable for laboratory measurement.The other is a portable near-infrared spectrometer,such as Micro NIR 1700 spectrometer,which has a low spectral resolution and the relatively weak predictive ability of the model,but is small in size,easy to carry,and suitable for on-site determination.If the prediction ability of the model is not significantly reduced,the purpose of reducing the size of the instrument and the cost of the instrument while improving the prediction ability of the model will have important significance for the promotion and application of the near-infrared technology.For this problem,this paper proposes a feasible solution based on the model transfer technology,that is,the model transfer method is used to realize the model transfer between two different categories(analysis-grade and portable-grade)of near-infrared spectrometers,which can not only improve prediction ability of the model of portable near-infrared spectrometer,but also perform on-site rapid screening and quality analysis of RAE,to a certain extent,reduce the application cost of near-infrared spectrometer.This paper applies near infrared spectroscopy to the quality analysis and evaluation of RAE,the main research contents are as follows1.Determination of active components in RAEAstragaloside Ⅳ(AST Ⅳ),calycosin-7-glucoside(CG)and astragalus polysaccharide(APS)were selected as the marker indicators for the quality control of RAE.The content of AST Ⅳ and CG in RAE was determined by HPLC method.The content of APS in RAE was determined by phenol sulfuric acid method.From the perspective of the content of active components,there are some differences in the quality of RAE samples from different manufacturers,while the quality difference between samples from the same concentration of the same manufacturer is smaller.The determination results of the three active components of RAE are used as reference data for the subsequent near infrared spectroscopy studies2.Identification of manufacturers and determination of active components for RAE with micro near infrared spectrometer.Using near-infrared spectroscopy,a method was established to rapidly determine the content of three active components in RAE from different manufacturers and distinguish samples of RAE from different manufacturers.First,the partial least squares regression(PLSR)model and the discriminant analysis(DA)model of two near-infrared spectroscopy instruments were established to simultaneously determine the content of active components and quickly distinguish the manufacturers of RAE samples.Then compare the prediction accuracy of the PLSR model and the DA model based on the two instruments in detail to evaluate the performance of the Micro-NIR 1700 spectrometer in the qualitative and quantitative analysis of the RAE.DA results showed that the discriminant models based on two instruments could rapidly discriminate the manufacturers of RAE samples,and the identification accuracy both achieved 100%.The values of Rp,RMSEP and RPD of the best PLSR models on two spectrometers are summarized belowMicro-NIR 1700 spectrometer:The values of Rp,RMSEP and RPD of the PLSR model of AST IV are 0.9958,0.0045%and 9.96,respectively.The values of Rp,RMSEP and RPD of the PLSR model of CG are 0.9969,0.0011%and 11.46,respectively.The values of Rp,RMSEP and RPD of the PLSR model APS are 0.9709,1.06%and 3.94,respectively.Antaris ⅡFT-NIR spectrometer:The values of Rp,RMSEP and RPD of the PLSR model of AST IV are 0.9974,0.0033%and 13.78,respectively.The values of Rp,RMSEP and RPD of the PLSR model of CG are 0.9989,0.0007%and 19.89,respectively.The values of Rp,RMSEP and RPD of the PLSR model of APS are 0.9742,0.91%and 4.36,respectively.As for the same samples,the values of Rp,RMSEP and RPD of the best PLSR models on two spectrometers are summarized below.Micro-NIR 1700 spectrometer:The values of Rp,RMSEP and RPD of the PLSR model of AST Ⅳ are 0.9946,0.0052%and 7.13,respectively.The values of Rp,RMSEP and RPD of the PLSR model of CG are 0.9914,0.0012%and 7.10,respectively.The values of Rp,RMSEP and RPD of the PLSR model of APS are 0.9742,0.91%and 3.99,respectively.Antaris Ⅱ FT-NIR spectrometer:The values of Rp,RMSEP and RPD of the PLSR model of AST IV are 0.9981,0.0026%and 15.64,respectively.The values of Rp,RMSEP and RPD of the PLSR model of CG are 0.9944,0.0010%and 9.09,respectively.The values of Rp,RMSEP and RPD of the PLSR model of APS are 0.9739,0.85%and 4.27,respectively.Quantitation results indicated that the PLSR models of two instruments had good prediction accuracy,but the predictive ability of the model of Micro-NIR 1700 spectrometer was weak.The above research results prove that the Micro-NIR 1700 spectrometer can be used for rapid identification of manufacturers and determination of three active components in RAE,which provides a convenient method for the rapid quality detection and on-site screening of RAE.At the same time,it provides a data reference for the subsequent model transfer research of near infrared spectroscopy.3.Study on antioxidant activity of RAE.The 1,1-diphenyl-2-trinitrophenylhydrazine radical(DPPH)scavenging method was used to determine the antioxidant activity of 82 samples of RAE.The PLSR was used to establish the correlation model between the near-infrared spectrum and the antioxidant activity index IC50 value.The model was optimized through the selection of pretreatment methods and spectral variables.Finally,the SPA method was used to select 10 variables from the raw spectrum to establish an optimized PLSR model.The corresponding results were as follows:The Rc,RMSEC,Rp,RMSEP,RMSECV and RPD values were 0.9851,0.0998 mg/mL,0.9957,0.0610 mg/mL,0.1338 mg/mL and 10.30,respectively.The above results indicated that the prediction accuracy of PLSR model is high and can accurately predict the antioxidant activity of RAE.The PLSR model also showed the potential application value in the quality control and evaluation of RAE.4.Study on model transfer of APS in RAE.For the model transfer between spectrometers of different manufacturers,took the content of APS in RAE as the research object,a new model transfer method based on spectral data transfer was proposed.The basic idea is to use the principal component analysis(PCA)method to achieve bidirectional transfer between the main instrument(Antaris Ⅱ FT-NIR spectrometer)and the target instrument(Micro-NIR 1700 spectrometer)models while ensuring that the spectral information is not lost as much as possible.The RMSEP values of the model evaluation parameters of the forward transfer and reverse transfer of the model by the PDS method are respectively 1.49%and 1.84%,which are higher than the prediction results of the target instrument,and the model transfer effect is poor.When the newly proposed model transfer method is used for model transfer,the RMSEP values of the model after forward and reverse transfer are respectively 1.18%and 1.16%,which are lower than the prediction results of the target instrument and close to the prediction results of the main instrument.To verify the transfer effect of the newly proposed model transfer method,the method of this paper was compared with the traditional PDS method.The results show that the newly proposed model transfer method had better model transfer effect and compared with the traditional PDS method,the method got the minimum RMSEP value and the prediction result was close to the main instrument.