Application of Multivariate Curve Resolution ... - ACS Publications


Application of Multivariate Curve Resolution...

0 downloads 106 Views 3MB Size

Article pubs.acs.org/IECR

Application of Multivariate Curve Resolution Method in the Quantitative Monitoring Transformation of Salvianolic Acid A Using Online UV Spectroscopy and Mass Spectroscopy Xintian Zheng,† Xingchu Gong,† Qin Li,‡,§ and Haibin Qu*,† †

Pharmaceutical Informatics Institute, Zhejiang University, Hangzhou 310058, P.R. China Department of Chemical Engineering, Curtin University of Technology, Perth, WA 6845, Australia § Environmental Engineering and Queensland Micro- and Nanotechnology Centre, Griffith University, Nathan Campus, Queensland 4111, Australia ‡

ABSTRACT: This work presents an exploratory study of monitoring the transformation process of salvianolic acid A (SAA) using an online UV spectroscopic analysis system. A chemometrics approach, based on multivariate curve resolution of spectrophotometric matrix, was applied to resolve the concentration and spectra profiles of reactant species and to evaluate the kinetic profile. Kinetic runs have been developed at temperatures ranging from 25 to 90 °C and pH values ranging from 2.5 to 10 in order to investigate the effects of these two variables during the transformation process of SAA. The degradation reaction of SAA was considered as first-order and the degradation kinetic constant of SAA increases when temperature increases. It was also found that SAA transforms rapidly when pH is higher than 7, but remains stable when pH is lower than 6.5. Two transformation products were auxiliarily identified by direct mass spectrometric analysis using Mass Work software. Our results suggest that the manufacturing process of Chinese medicine preparations with SAA as the main bioactive compound should not be operated at high temperature and high pH.

1. INTRODUCTION Salvianolic acid A (SAA) is one of the main effective components of Salvia miltiorrhiza Bunge (Danshen), and possesses a wide range of pharmacological activities, including antiplatelet, antifibrotic, anti-inflammatory, and anticarcinogenic properties.1 These functionalities of SAA are postulated to stem from its strong antioxidant activity.2−7 As a result, there is a great interest in SAA as a new drug candidate in the pharmaceutical field. As a trimer of caffeic acid and danshensu (3-(3,4-dihydroxyphenyl) lactic acid), SAA is not stable in aqueous solution and some Danshen preparations.8 Heating and acid−alkali treatment during the entire manufacturing process may result in the transformation of SAA. Undoubtedly, the stability of SAA will affect the efficacy and safety of potential drugs derived from SAA. To date SAA transformation in the manufacture, storage, and other production processes has not yet been reported. Up to now, few studies have focused on the transformation or degradation of the unstable bioactive components in Chinese herbal medicines.9−11 However, these studies were based on off-line determination methods. These off-line methods, typically time-consuming and laborious, are still applicable when the transformation is slow. However, when the transformation process takes place at a relatively fast reaction rate, side reactions such as hydrolysis, oxidation, and precipitation may occur simultaneously during the off-line sample measurement period. Thus a more practical fast-responding and nondestructive online monitoring method such as a continuous-flow system is required. This online monitoring offers advantages such as excellent reproducibility, automated process, and low consumption of samples and reagents. Online measurements also allow the study of kinetics of the degradation processes.12 © 2012 American Chemical Society

Spectroscopic methods including near-infrared (NIR) and Raman are widely used tools for qualitative and quantitative analyses in food, pharmaceutics, and environmental science.13−17 When coupled to online monitoring, spectroscopic methods provide large amounts of information which can be collected and interpreted rapidly from chemical processes. However, NIR spectra obtained from aqueous solutions are usually of poor quality due to the strong interference of water absorbance bands. Raman spectra mainly occur in the visible region and do not interfere with water, but the weak Raman signal at low concentrations makes it inferior in sensitivity.18 UV spectroscopy, on the other hand, is much more sensitive in aqueous solutions. Its light sources and detectors are broadly available and analysis with high signal-to-noise ratio is possible. The capital cost of a UV spectroscopy system is significantly lower than that of NIR and Raman spectroscopy. Thus, UV spectroscopy system is more suitable in the online analysis of Chinese herbal medicine manufacturing process. During the monitoring of transformation process by means of online UV spectroscopy, a series of spectra are collected as a function of time. As the transformation takes place, changes in absorption bands show the evolution of the process. However, raw UV absorption spectra are highly overlapping; they are not suitable as direct read-out signals. To improve the significance of online UV spectroscopic analysis system with respect to the characterization of transformation compounds in an evolving Received: Revised: Accepted: Published: 3238

July 16, 2011 January 4, 2012 January 4, 2012 January 4, 2012 dx.doi.org/10.1021/ie201536y | Ind. Eng.Chem. Res. 2012, 51, 3238−3245

Industrial & Engineering Chemistry Research

Article

consisted of 0.1% aqueous formic acid (A) and 0.1% formic acid in acetonitrile (B). The flow rate was 0.5 mL min −1. High purity nitrogen (N2) was used as the drying gas. For negative ESI analysis, the parameters were as follows: drying gas temperature 350 °C, capillary voltage 3500 V, drying gas flow 10 mL min −1, and nebulizer pressure 40 psig. For full-scan mass spectrometric (MS) analysis, the spectra were collected in the range of m/z 100−800. The scan data storage was in full mode for further Mass Works (Cerno Bioscience) analysis. In addition, the ion threshold was set to zero and the acquisition rate was set to 0.05. 2.3. Procedures. The transformation process was monitored by heating 70 mL 1.5 × 10 −5 M aqueous solution of SAA at different conditions as summarized in Table 1. The UV

mixture, many chemometrics methods can be employed. Unlike principal component analysis (PCA) and partial least-squares (PLS), usually containing little or no physical or chemical meaning in their principal components, multivariate curve resolution (MCR) method is proven to be a versatile chemometrics method for resolving the pure chemical species from a complex evolving system.19 This method combines the strengths of hard-modeling and self- or soft-modeling methods, allowing the extraction of pure spectra of components involved and the formation of concentration profiles of the chemical species in an evolutionary process. Furthermore, this concentration data can be used to develop kinetic models for the process with or without prior knowledge.20−23 In this paper, the transformation process of SAA was monitored with online UV spectrometry. Multivariate curve resolution was applied to resolve the absorption profiles and to extract the spectra of reaction species and the associated quantitative information. The results were fitted to a kinetics model for obtaining the rate constants. Complementary information about the possible identity of transformation compounds by direct mass spectrometric analysis using Mass Works software is also presented.

Table 1. Summary of the Experiment Conditions of SAA Transformation

2. MATERIALS AND METHODS 2.1. Reagents and Drugs. SAA and its transformed products with >95% purity validated by HPLC analysis were isolated and purified in our laboratory. HPLC-grade acetonitrile and formic acid were purchased from Merck (Darmstadt, Germany) and Tedia (Fairfield, OH, USA), respectively. Ultrahigh-purity water was produced using a Millipore Milli-Q System (Milford, MA, USA). All other chemicals were of analytical reagent grade. 2.2. Instruments and Conditions. A T6 UV spectrophotometer (Pukinje, Beijing, China), with UV win 5 software (Pukinje, Beijing, China), was used for spectra collection and was operated at the following conditions: scanning range 220− 400 nm; data interval 1 nm; scan speed medium; 5-mm pathway quartz flow-cell. Continuous-flow manifold was composed of a K501 transfusion pump (Lumiere Analytical, Germany), a reaction container with water-bath heating, a flow cell, and peristaltic pump tube (200 cm ×2 mm I.D.). During the procedure, the pump delivered sample at a constant flow rate of 20 mL min−1; sample solution reached the detection flow cell in 9 s. The schematic of online UV spectroscopic analysis system is shown in Figure 1.

kinetic studies

pH

influence of temperature

6.5

influence of pH

2.5 4.5 6.5 8.5 10

temperature °C

run time

time intervals

data matrix (time points × wavelengths)

25 60 70 80 90 90

7 days 9h 7.5 h 6.5 h 6h 10 h 10 h 6h 2h 2h

1 day 1 min 1 min 1 min 1 min 10 min 10 min 1 min 1 min 1 min

7 × 180 330× 180 265× 180 220× 180 200× 180 56 × 180 50× 180 200× 180 86× 180 50× 180

spectra of the continuous flow solution were recorded at a fixed time interval over the wavelength range of 220−400 at 1-nm intervals. After the transformation reaction was finished, all of the process spectra were exported to a data matrix for further chemometrics analysis. 2.4. Multivariate Curve Resolution Method. Multivariate curve resolution−alternating least-squares (MCR-ALS)19 is a flexible, iterative curve resolution method for decomposing a mixture data matrix into the pure contributions of all significant components in the system along the two measurement directions. In matrix form, this bilinear model can be expressed in the following equation:

D = CST + E In this work, D is the spectral data matrix containing rows of UV spectra recorded during a transformation process. The columns of the C matrix are the concentration profiles of the modeled components, and the rows in the ST matrix are their related pure UV spectra. E is the matrix of residuals not explained by the MCR-ALS model. A superior advantage of MCR-ALS is the possibility of working simultaneously with several data matrices. To do so, column-wise and/or row-wise augmented data matrices are used. In the context of the transformation experiments, the data analysis can be carried out simultaneously over several experiments at different temperatures, setting the corresponding data matrices one on top of another, forming a column-wise data matrix. The procedures of MCR-ALS method include rank analysis, initial estimates, and constrained alternating least-squares optimization. In this work, the rank (significant component) was

Figure 1. Scheme of the online UV spectrophotometric analysis system. P is a binary pump, flow rate: 20 mL min−1. T is a temperature controller system. PC is a workstation for spectra collection. Water bath was set at different temperatures as summarized in Table 1.

For direct mass spectrometric analysis, an Agilent quadrupole mass spectrometer (Agilent, IL, USA) with an ESI interface was used. All samples were infused into the mass spectrometer via a prefilter without column separation. The mobile phase 3239

dx.doi.org/10.1021/ie201536y | Ind. Eng.Chem. Res. 2012, 51, 3238−3245

Industrial & Engineering Chemistry Research

Article

Figure 2. Example of spectrophotometric data obtained in the transformation of SAA at 80 °C and pH 6.5 for about 7 h. The arrow indicates the changing direction of the absorbance spectra of SAA.

determined by singular value decomposition (SVD).24 The chemical rank can be obtained by the number of singular values larger than singular values associated with noise. MCR-ALS needs an initial estimation for the spectral or concentration profiles for each species. Several algorithms such as evolving factor analysis (EFA)25−27 or the determination of the purest variables28−30 can be used for this purpose. In this work, concentration profile type estimates were obtained from EFA. Once the initial estimates were constructed, an iterative alternating least-squares optimization of matrix ST was carried out until convergence was achieved. To minimize the ambiguity of the solution, non-negativity constraint to the concentration and spectra profiles is generally used, due to the fact that the concentrations of the chemical species and the intensity of the radiation absorbed or reflected by samples can never be negative. Besides non-negativity, several other constrains can be applied to the concentration and spectra profiles, such as unimodality (single maxima), closure (mass balance), hard-modeling, and selectivity.19,20,31 In this study, non-negativity in spectral and concentration profiles, unimodality, and closure in concentration profiles were used. After optimization, the resulting concentration and spectra profiles were used for further kinetic and qualitative analysis. All data processing was carried out using MATLAB with the MCR toolbox.32 2.5. Identification of Reaction Species. At different intervals of time, aliquots of 20 μL of the sample were collected and analyzed by direct mass spectrometry to identify transformation products. To obtain stable signals, each sample was scanned for 2 min. The acquired raw data were uploaded to Mass Works software for evaluation. SAA was used as the pure standard for internal calibration. The calculated accurate mass and possible formulas of the transformed products were proposed by Mass Works software.

3. RESULTS AND DISCUSSION 3.1. Effect of Temperature and pH on the Transformation of SAA. The transformation of SAA in aqueous solution was shown in the changes in the UV absorbance spectra (Figure 2). The initial spectra of SAA at 80 °C and pH 6.5 possess absorbance maximum at 286 nm and a peak valley at 250 nm. Under this temperature and pH, during about 6.5 h

Figure 3. MCR-ALS analysis of SAA transformation process at 60 and 80 °C, identical species spectra profiles were resolved.

of transformation, an additional absorbance maximum at 320 nm appeared, while the curve valley gradually shifted to 260 nm. Because the SAA solution was freshly prepared and 3240

dx.doi.org/10.1021/ie201536y | Ind. Eng.Chem. Res. 2012, 51, 3238−3245

Industrial & Engineering Chemistry Research

Article

Figure 4. MCR-ALS resolved concentration profiles of SAA transformation at different experiment conditions (A 60 °C, B 70 °C, C 80 °C, D 90 °C) and related spectra profile E ( SAA, --transformation products) using column-wise augmented data matrices.

Figure 6. (A) MS spectra of 12 samples extracted from transformation process at different time intervals; MCR resolved results of the MS spectra data matrix. (B) MS spectra of reactant, (C) MS spectra of transformation products, (D) related kinetic profiles ( reactant, --transformation products).

Figure 5. Concentration profiles resolved by soft-modeling MCR-ALS (―) vs. concentration profiles fitted to a first-order reaction model (---) and concentration profiles fitted to a second-order reaction model ( · ).

purified at the beginning of the process, these changes in the absorbance spectra can only be attributed to the conversion of SAA to its transformation products. The influence of temperature on the transformation of SAA in aqueous solutions was investigated in the range of 25−90 °C and at pH 6.5. The spectroscopic data did not change at 25 °C for several days. In contrast, the transformation was completed in a few hours at 90 °C. Significant differences in reaction rate were observed. In the slow process at 25 °C, the measurements were carried out by manual procedure at each time interval. For the fast process, the continuous flow system was used for sampling. The effect of pH on the transformation was studied while temperature was set to 90 °C. It was shown that stability was attained in the acid range from pH 2.5−6.5. The spectra remained unchanged for more than 10 h at pH = 2.5. However, when pH was greater than 7, the conversion was severely accelerated by the increase in pH. At pH 10, the transformation reaction was completed within 1.5 h.

The pH-dependent characteristic of the transformation can be analyzed from the chemical structure of SAA. The ester bond in SAA can be easily hydrolyzed under alkaline condition, which may result in SAA chemical structure change. On the contrary, the phenolic hydroxyl groups and the carboxyl group in SAA are all stable under acid conditions. However, reaction mechanism in various media requires further investigation. All these results suggest that for SAA to remain stable during the manufacturing process, a combination of high temperature and alkaline environment should be avoided as much as possible. 3.2. MCR-ALS Analysis Transformation Experiments. To further study the kinetics of the transformation process, an exploratory analysis of the process evolution by means of MCRALS was carried out. As the first step, a rank analysis was performed with singular value decomposition (SVD). This identified a relative separation between the second and third eigenvalues, and the first two components contributed 99.9% variance. This result indicated the presence of two chemical 3241

dx.doi.org/10.1021/ie201536y | Ind. Eng.Chem. Res. 2012, 51, 3238−3245

Industrial & Engineering Chemistry Research

Article

Table 2. Mass Works Analysis Results for the SAA Transformation Product Molecular Ion [M − H]− m/z = 491a row

formula

mono isotope

mass error (mDa)

mass error (PPM)

spectral accuracy

RMSEb

DBEc

1 2 3 4 5 6 7 8 9 10

C26H19O10 C27H15O6N4 C25H19O9N2 C28H15O7N2 C29H15O8 C24H15O10N2 C23H15O9N4 C30H11O4N4 C22H19O13 C21H19O12N2

491.0978 491.0992 491.1091 491.0879 491.0767 491.0727 491.0839 491.078 491.0826 491.0938

8.0218 9.3592 19.2552 −1.8742 −13.1075 −17.1303 −5.8969 −11.7701 −7.2343 3.9991

16.3348 19.0581 39.2092 −3.8163 −26.6907 −34.8822 −12.0078 −23.9674 −14.7311 8.1433

98.8484 98.6705 98.6675 98.4743 98.2584 98.063 97.8277 96.715 96.3391 96.0909

111 128 129 147 168 187 210 317 354 378

17.5 22.5 17.5 22.5 22.5 18.5 18.5 27.5 13.5 13.5

a The ordering rule of the first ten possible results is mainly based on spectral accuracy. bRMSE is root-mean-square error. cDBE is double bond equivalent.

Table 3. Mass Works Analysis Results for the SAA Transformation Product Molecular Ion [M − H]− m/z = 509a row

formula

mono isotope

mass error (mDa)

mass error (PPM)

spectral accuracy

RMSEb

DBEc

1 2 3 4 5 6 7 8 9 10

C26H21O11 C25H21O10N2 C24H17O11N2 C23H17O10N4 C27H17O7N4 C28H17O8N2 C29H17O9 C22H21O14 C21H21O13N2 C20H21O12N4

509.108 509.1196 509.0832 509.0945 509.1097 509.0985 509.0873 509.0931 509.1044 509.1156

5.5865 16.8199 −19.5656 −8.3322 6.9239 −4.3095 −15.5429 −9.6696 1.5638 12.7972

10.9733 33.0383 −38.4315 −16.3665 13.6002 −8.4648 −30.5299 −18.9934 3.0716 25.1367

99.0328 99.0208 98.7263 98.506 97.5403 97.3443 97.1286 96.8266 96.5844 96.3349

93 94 122 144 236 255 276 305 328 352

16.5 16.5 17.5 17.5 21.5 21.5 21.5 12.5 12.5 12.5

a

The ordering rule of the first ten possible results is mainly based on spectral accuracy. bRMSE is root-mean-square error. cDBE is double bond equivalent.

MCR-ALS was fitted to a hard kinetic model and rate constants were calculated, a new estimated concentration profile was generated based on these rate constants, and then used as the initial estimate of a new MCR-ALS optimization. This iteration was stopped when the least lack of fit was reached, and then the most reasonable results were selected as the kinetic model for the transformation process. In our study, the kinetic model is suggested to be a simple pseudo-first-order reaction A→B (r = k[A]) from the shape of the concentration profiles resolved by MCR-ALS. To confirm this hypothesis, the calculated concentration profiles were also fitted to a second-order reaction model r = k[A]2. Smaller fitting error was obtained when the reaction were considered as first-order, and there was a close agreement between the resolved concentration profile and hard-modeling fitted one, as shown in Figure 5. Thus, no more optimization was used. The kinetic constant was calculated using the following equation:

species in this reaction system and the other factors might be related to system noise. MCR-ALS optimization was then started with the initial estimation of concentration profiles by EFA. To reflect the physiochemical reality of the process, soft constraints such as non-negativity of concentration and spectra profiles, unimodality of concentration profiles, and mass balance of concentration profiles were employed. Kinetic runs at different temperatures were analyzed simultaneously from the construction of the column-wise augmented data matrix. This extension of matrix was arranged by putting one matrix on the top of another and keeping common wavelengths in the same column. This augmented matrix analysis was potentially applicable to these temperature-dependent data sets as the spectra profiles of chemical species in these data sets were identical. Formerly separated MCR analysis results of single matrix are shown in Figure 3, which confirmed the identical spectra profiles of species in data sets of different temperature. Figure 4 shows the MCR-ALS resolved concentration and spectra profiles using column-wise augmented matrix. Concentration profiles were calculated over the whole experimental time range, however, in Figure 4, the time window was fixed between 0 and 200 time intervals for a better display of the effect of temperature on the process. 3.3. Kinetics Calculations. Kinetic spectroscopic monitoring provided information about the evolution of the species involved in this reaction and derived kinetic parameters about the underlying reaction model. As recently reported,33 a combined hard- and soft-modeling step was incorporated in order to minimize the rotational ambiguity and to obtain more accurate results. Once the optimized concentration profile by

[A]t = [A]0 e−kt

(1)

In this equation, [A]0 is the analytical concentration of SAA before reaction, [A]t is time-dependent equilibrium concentration of its transformation product, and k is the rate constant. The calculated rate constants at 60, 70, 80, and 90 °C were 0.00348, 0.00404, 0.00586, and 0.00687 min −1, respectively. It was obvious that kinetic constant increases as temperature increases. 3.4. Direct Mass Spectrometric Analysis of the Transformation Process of SAA. Extracts from the transformation process at 12 different time points were analyzed by mass spectrometry to confirm the results of online UV experiments. Each sample analyzed by mass spectrometry 3242

dx.doi.org/10.1021/ie201536y | Ind. Eng.Chem. Res. 2012, 51, 3238−3245

Industrial & Engineering Chemistry Research

Article

Figure 7. Mass spectroscopy of two pairs of isomer transformation products.

analysis does not exclude the possibility of multispecies existence as the reaction final products which are however identified as a single component. In reactions such as the one proposed in this work, the formation of several species with the same kinetic profiles (for instance, two species having equal concentration profiles or spectra profiles) can produce the socalled rank deficiency of spectral matrices measured on them. Therefore, in this work, some species could be considered as a

gave a data matrix whose rows had the MS spectra at different scan times and whose columns had the profiles at different m/z values. The same scan time ranging from 0.8 to1 min was selected for the raw data of all 12 samples, then mean MS spectra were obtained for each sample; the results are shown in Figure 6A. The MS spectra data matrix was resolved by MCRALS into related species MS spectra and kinetic profiles, as shown in Figure 6B−D. Our MCR combined UV spectra 3243

dx.doi.org/10.1021/ie201536y | Ind. Eng.Chem. Res. 2012, 51, 3238−3245

Industrial & Engineering Chemistry Research

Article

6.5, SAA was relatively stable. However, it transformed rapidly when pH was higher than 7. Results obtained in this paper clearly indicate that preparations with SAA as the main bioactive compounds should avoid high temperature and high pH environment during manufacturing processes.

mixture of chemical compounds with different molecular structures instead of a pure chemical compound with a single molecular structure. The identification of chemical species in the reaction system was implemented with the MS detection and further confirmed with Mass Works software analysis. Accurate mass measurements of these transformation products were conducted using this software. A comprehensive calibration involving m/z and peak shape had been utilized to obtain the high spectral accuracy for achieving more reliable composition determination.34 The quasi-molecular ions [M − H]− of the transformation products were at m/z 509 and 491. Results shown in Tables 2 and 3 suggest that there are two kinds of transformation products in the reaction system: C26H21O11 and C26H19O10. These results are in agreement with our previous study which found that SAA (C26H22O10, [M − H]− m/z = 493) was oxidized in aqueous solution at 90 °C, and the reaction products included salvianolic acid C (C26H20O10, [M − H]− m/z = 491), isosalvianolic acid C (C26H20O10, [M − H]− m/z = 491), and two new polyphenolic acids (C26H22O11, [M − H]− m/z = 509).35 The mass spectra of the standards of these transformation products are shown in Figure 7, which confirmed the resolved MS spectra based on MCR. The high similarity among the UV spectra of these four oxidation products is shown in Figure 8. This phenomenon confirmed



AUTHOR INFORMATION

Corresponding Author

*Tel: + 86-571-88208428. Fax: + 86-571-88208428. E-mail: [email protected].



ACKNOWLEDGMENTS This work was financially supported by China International Science and Technology Cooperation Project (2010DFB33630) and the key project from Large Science & Technology Special Grant of Zhejiang Province (200813004-1). We thank Dr. Mo Huanbiao for his thoughtful comments and professional English editing for this paper. Our sincere gratitude also goes to Weijian Li from Lumiere Technology for his help in the Mass Works data processing. In addition, we thank Dr. Jinzhong Xu for supplying the pure standards of SAA and its transformation products.



REFERENCES

(1) Kang, H. S.; Chung, H. Y.; Byun, D. S.; Choi, J. S. Further isolation of antioxidative (+)-1-hydroxypinoresinol-1-O-beta-D-glucoside from the rhizome of Salvia miltiorrhiza that acts on peroxynitrite, total ROS and 1,1-diphenyl-2-picrylhydrazyl radical. Arch. Pharm. Res. 2003, 26, 24. (2) Fan, H. Y.; Fu, F. H.; Yang, M. Y.; Xu, H.; Zhang, A. H.; Liu, K. Antiplatelet and antithrombotic activities of salvianolic acid A. Thromb. Res. 2010, 126, 17. (3) Huang, Z. S.; Zeng, C. L.; Zhu, L. J.; Jiang, L.; Li, N.; Hu, H. Salvianolic acid A inhibits platelet activation and arterial thrombosis via inhibition of phosphoinositide 3-kinase. J. Thromb. Haemost. 2010, 8, 1383. (4) Wang, S. B.; Tian, S.; Yang, F.; Yang, H. G.; Yang, X. Y.; Du, G. H. Cardioprotective effect of salvianolic acid A on isoproterenolinduced myocardial infarction in rats. Eur. J. Pharmacol. 2009, 615, 125. (5) Lin, T. J.; Zhang, K. J.; Liu, G. T. Effects of salvianolic acid A on oxygen radicals released by rat neutrophils and on neutrophil function. Biochem. Pharmacol. 1996, 51, 1237. (6) Liu, C. H.; Hu, Y. Y.; Wang, X. L.; Liu, P.; Xu, L. M. Effects of salvianolic acid-A on NIH/3T3 fibroblast proliferation, collagen synthesis and gene expression. World J. Gastroenterol. 2000, 6, 361. (7) Wang, X. J.; Wang, Z. B.; Xu, J. X. Effect of salvianic acid A on lipid peroxidation and membrane permeability in mitochondria. J. Ethnopharmacol. 2005, 97, 441. (8) Xu, J. Z.; Shen, J.; Cheng, Y. Y.; Qu, H. B. Simultaneous detection of seven phenolic acids in Danshen injection using HPLC with ultraviolet detector. J. Zhejiang Univ. Sci. B 2008, 9, 728. (9) Jia, M. G.; Sun, W. J. Studies on transform of matrine and oxymatrine in Radix Sophorae Flavecentis and its prescription. Chin. J. Pharm. Anal. 2003, 23, 91. (10) Guo, Y. X.; Xiu, Z. L.; Zhang, D. J.; Wang, H.; Wang, L. X.; Xiao, H. B. Kinetics and mechanism of degradation of lithospermic acid B in aqueous solution. J. Pharm. Biomed. Anal. 2007, 43, 1249. (11) Guo, Y. X.; Zhang, D. J.; Wang, H.; Xiu, Z. L.; Wang, L. X.; Xiao, H. B. Hydrolytic kinetics of lithospermic acid B extracted from roots of Salvia miltiorrhiza. J. Pharm. Biomed. Anal. 2007, 43, 435. (12) Daneshvar, N.; Aber, S.; Khani, A.; Khataee, A. R. Study of imidaclopride removal from aqueous solution by adsorption onto granular activated carbon using an on-line spectrophotometric analysis system. J. Hazard. Mater. 2007, 144, 47.

Figure 8. UV spectra of four transformation products.

the intrinsic difficulties encountered in the isolation and characterization of each species by soft-modeling MCR-ALS. On the other hand, MCR combined with UV spectra is a direct and effective method that does not require separations; hence, it appears to be a useful method for online process monitoring in chemical processing.

4. CONCLUSIONS MCR is an excellent method that offers a calibration-free resolution of chemical compounds from a complex matrix and acquires chemically meaningful information without prior chromatographic separation. A methodology is proposed in this study to describe the transformation process of SAA by employing online UV spectroscopic monitoring combined with MCR. Two degradation products were found and characterized with mass spectrometry. The degradation of SAA was characterized as first-order reaction. The kinetic constant was found to increase as temperature increases. When pH was lower than 3244

dx.doi.org/10.1021/ie201536y | Ind. Eng.Chem. Res. 2012, 51, 3238−3245

Industrial & Engineering Chemistry Research



(13) Luykx, M. A. M. D; van Ruth, S. M. An overview of analytical methods for determining the geographical origin of food products. Food Chem. 2008, 107, 897. (14) Karoui, R.; De Baerdemaeker, J. A review of the analytical methods coupled with chemometric tools for the determination of the quality and identity of dairy products. Food Chem. 2007, 102, 621. (15) Roggo, Y.; Chalus, P.; Maurer, L.; Lema-Martinez, C.; Edmond, A; Jent, N. A review of near infrared spectroscopy and chemometrics in pharmaceutical technologies. J. Pharm. Biomed. Anal. 2007, 44, 683. (16) Park, S. C.; Kim, M. J.; Noh, J.; Chung, H.; Woo, Y.; Lee, J.; et al. Reliable and fast quantitative analysis of active ingredient in pharmaceutical suspension using Raman spectroscopy. Anal. Chim. Acta 2007, 593, 46. (17) Costa, J. C. S.; Sant’Ana, A. C.; Corio, P.; Temperini, M. L. A. Chemical analysis of polycyclic aromatic hydrocarbons by surfaceenhanced Raman spectroscopy. Talanta 2006, 70, 1011. (18) Chalmers, J. M. Spectroscopy in Process Analysis; CRC Press: Boca Raton, FL, 2000. (19) Tauler, R. Multivariate curve resolution applied to second order data. Chemom. Intell. Lab. Syst. 1995, 30, 133. (20) de Juan, A.; Tauler, R. Chemometrics applied to unravel multicomponent processes and mixtures: Revisiting latest trends in multivariate resolution. Anal. Chim. Acta 2003, 500, 195. (21) Navea, S.; de Juan, A.; Tauler, R. Modeling temperaturedependent protein structural transitions by combined near-IR and mid-IR spectroscopies and multivariate curve resolution. Anal. Chem. 2003, 75, 5592. (22) Ruckebusch, C.; Duponchel, L.; Sombret, B.; Huvenne, J. P.; Saurina, J. Time-resolved step-scan FT-IR spectroscopy: focus on multivariate curve resolution. J. Chem. Inf. Comp. Sci. 2003, 43, 1966. (23) Kessler, W.; Kessler, R. W. Multivariate curve resolution: a method of evaluating the kinetics of biotechnological reactions. Anal. Bioanal. Chem. 2006, 384, 1087. (24) Golub, G. H.; Reinsh, C. Singular value decomposition and least squares solutions. Numer. Math. 1970, 14, 403. (25) Gampp, H.; Maeder, M.; Meyer, C.; Zuberbuhler, A. D. Calculation of equilibrium constants from multiwavelength spectroscopic dataIV: Model-free least-squares refinement by use of evolving factor analysis. Talanta 1986, 33, 943. (26) Maeder, M.; Zilian, A. Evolving factor analysis, a new multivariate technique in chromatography. Chemom. Intell. Lab. Syst. 1988, 3, 205. (27) Maeder, M. Evolving factor analysis for the resolution of overlapping chromatographic peaks. Anal. Chem. 1987, 59, 527. (28) Windig, W.; Guilment, J. Interactive self-modeling mixture analysis. Anal. Chem. 1991, 63, 1425. (29) Windig, W.; Stephenson, D. A. Self-modeling mixture analysis of second-derivative near-infrared spectral data using the SIMPLISMA approach. Anal. Chem. 1992, 64, 2735. (30) Windig, W.; Heckler, C. E.; Agblevor, F. A.; Evans, R. J. Selfmodeling mixture analysis of categorized pyrolysis mass spectral data with the SIMPLISMA approach. Chemom. Intell. Lab. Syst. 1992, 14, 195. (31) Tauler, R.; Smilde, A. K.; Kowalski, B. J. Selectivity, local rank, three-way data analysis and ambiguity in multivariate curve resolution. J. Chemom. 1995, 9, 31. (32) Gemperline, P. J.; Cash, E. Advantages of soft versus hard constraints in self-modeling curve resolution problems. Alternating least squares with penalty functions. Anal. Chem. 2003, 75, 4236. (33) de Juan, A.; Maeder, M.; Martinez, M.; Tauler, R. Combining hard- and soft-modelling to solve kinetic problems. Chemom. Intell. Lab. Syst. 2000, 54, 123. (34) Gu, M.; Wang, Y.; Zhao, X.; Gu, Z. Accurate mass filtering of ion chromatograms for metabolite identification using a unit mass resolution liquid chromatography/mass spectrometry system. Rapid Commun. Mass Spectrom. 2006, 20, 764. (35) Xu, J. Z.; Zeng, S. S.; Chen, X. Y.; Qu, H. B. Isolation and Identification of degradation products of Salvianolic Acid A by NMR and LC-MS. Fitoterapia 2011, 82, 260.

Article

NOTE ADDED AFTER ASAP PUBLICATION After this paper was published online February 13, 2012, corrections were made to the order of the surnames and given names in the author list. The revised version was published February 16, 2012.

3245

dx.doi.org/10.1021/ie201536y | Ind. Eng.Chem. Res. 2012, 51, 3238−3245