Using a central composite design to optimize the analytical method for simultaneous determination of choline, pantothenic acid, ascorbic acid, and niacin in food supplements by capillary zone electrophoresis

26 HNUE JOURNAL OF SCIENCE DOI: 10.18173/2354-1059.2017-0051 Chemical and Biological Science 2017, Vol. 62, Issue 10, pp. 26-35 This paper is available online at USING A CENTRAL COMPOSITE DESIGN TO OPTIMIZE THE ANALYTICAL METHOD FOR SIMULTANEOUS DETERMINATION OF CHOLINE, PANTOTHENIC ACID, ASCORBIC ACID, AND NIACIN IN FOOD SUPPLEMENTS BY CAPILLARY ZONE ELECTROPHORESIS Nguyen Thanh Dam, Nguyen Manh Huy, Vu Minh Tuan, Duong Hong Anh and Pham Hung Viet VNU Key Laboratory of Analytical Technology for Environmental Quality and Food Safety Control (KLATEFOS), University of Science, Vietnam National University, Hanoi Abstract. This study focused on using a central composite design (CCD) to optimize the analytical method for simultaneous determination of choline, pantothenic acid (vitamin B5), ascorbic acid (vitamin C), and niacin (vitamin B3) by capillary zone electrophoresis (CZE). The optimization process was designed with 30 experiments, including four factors: Tris concentration, pH, separation voltage and sample injection time. The optimal condition was found: Tris concentration was 150 mmol/L, adjusted to pH = 9.5 by acetic acid with 10kV of applied voltage, and 60 s of sample injection time. Analytical methods were evaluated through limit of detection (LOD), limit of quantification (LOQ), linear range, repeatability, and recovery. The results showed that the method had good LOD (0.22 - 0.45 mg/L), extensive linear range (1 - 200 mg/L), high repetition with relative standard deviation (RSD) for peak area and migration time were less than 4% and 0.5% respectively, and relative recoveries ranged between 89 and 111%. The developed CZE-method could be used in the quality control of food supplements. Keywords: Choline, vitamins, capillary electrophoresis, response surface methodology, central composite design. 1. Introduction Choline is a water-soluble vitamin-like essential nutrient. This compound is the precursor molecule for the neurotransmitter acetylcholine, which is involved in many functions including memory and muscle control. It has been found that intake of choline during pregnancy can have long-term beneficial effects on memory for the child [6]. Nicotinic acid (niacin, vitamin B3), pantothenic acid (vitamin B5), and ascorbic acid (vitamin C), which are in the water-soluble vitamins group, play an important role in human health [5]. The deficiency of these compounds in the diet can cause serious health issues. Received November 15, 2017. Revised December 8, 2017. Accepted December 16, 2017. Contact Pham Hung Viet, e-mail address: phamhungviet@hus.edu.vn Using a central composite design to optimize the analytical method for simultaneous determination 27 Thus, there is a need to have reasonable analytical methods for their quality control in different types of food products. Several methods are used to measure choline [4, 7] and vitamins [2] in supplements and food products. However, only a few methods dealt with the simultaneous quantitative analysis of these compounds. In recent years, capillary zone electrophoresis (CZE) has been expanded to the simultaneous quantitative analysis of water-soluble vitamin in food products [5]. The chemical properties of these compounds, their ionic nature and water solubility, make them suitable for electrophoretic analysis. However, there are two main problems encountered during the development of CZE-method for this aim: (i) different structures and chemical properties of vitamins make the development of a single method for their simultaneous determination difficult; and (ii) foods are complex matrices. These difficulties can be overcome by performing an experimental design to choose the optimized condition to analysis choline, pantothenic acid, ascorbic acid, and niacin. Traditionally, optimization in analytical chemistry has been carried out by monitoring the influence of one factor at a time on an experimental response (one-variable-at-a-time). While only one parameter is changed, others are kept at a constant level. Its major disadvantage is that it does not include the interactive effects of the variables studied. Another disadvantage of this technique is the increase in the number of experiments necessary to conduct the research, which leads to an increase of time and expenses as well as an increase in the consumption of reagents and materials [3]. In order to overcome this problem, the optimization of analytical procedures has been carried out by using multivariate statistic techniques. Among the most relevant multivariate techniques used in analytical optimisation is central composite design (CCD), a type of symmetrical second- order experimental designs in response surface methodology (RSM). This paper describes the development of a simple CZE-method for the simultaneous determination of choline, pantothenic acid, ascorbic acid, and niacin, using an experimental design based on CCD technique. The method was validated and can be used to quantitative these vitamins in supplement foods. 2. Content 2.1. Experiment 2.1.1. Reagents and apparatus All chemicals were of analytical or reagent grade and purchased from Tokyo Chemical Industry (Japan) or Sigma-Aldrich (Germany), except lactic acid from Guangdong (China). Stock solutions (1000 mg/L) of choline, pantothenic acid, ascorbic acid, and niacin were used for the daily preparation of the standard solutions. Chemicals used for the preparation of BGEs included: Acetic acid (Ace), histidine (His), 2-(N- morpholino)ethanesulfonic acid (MES), lactic acid (Lac), tris(hydroxymethyl) aminomethane (Tris), 3-(N-morpholino)propanesulfonic acid (MOPS), N-Cyclohexyl-2- aminoethanesulfonic acid (CHES), N-cyclohexyl-3-aminopropanesulfonic acid (CAPS). Deionized water, purified using a water purification system from Millipore - model Simplicity UV (USA), was used for the preparation of all standard solutions and sample dilution if required. Nguyen Thanh Dam, Nguyen Manh Huy, Vu Minh Tuan, Duong Hong Anh and Pham Hung Viet 28 All experiments were performed on a hand-operated CE instrument in the VNU Key Laboratory of Analytical Technology for Environmental Quality and Food Safety Control (KLATEFOS), VNU University of Science. The electrophoresis section was based on a high voltage power supply (Spellman CZE2000, UK) with ±30 kV maximum output. Polyimide coated fused silica capillaries of 50 µm I.D. and 365 µm O.D. with 60 cm of total length and 49 cm of effective length (from Polymicro, USA) were used for the separations. Before their use, the capillaries were pre-conditioned with 1 M NaOH for 10 min and deionized water for 10 min, followed by flushing with the BGE. Detection was carried out with a miniaturized high-voltage C 4 D cells built in-house that is used a power supply of 12 VDC. The resulting signals were recorded with an e-corder 401 data acquisition system (eDAQ, Australia) connected to the USB-port of a personal computer. 2.1.2. Methodology * Design of experiment The composition of BGE was investigated before the development of the experimental design. Six different BGEs at pH 9.5 were evaluated: 100 mM Tris/Ace; 100 mM Tris/Lac; 60 mM Tris/40 mM CAPS; 100 mM Tris/5 mM MES; 100 mM Tris/5 mM CHES; 100 mM Tris/5 mM MOPS. After selecting the initial BGE component, further optimization of analysis conditions by using a central composite design with four independent variables (factors) was performed. The variables were: concentration of Tris (30 ÷ 150 mmol / L), pH (8.5 ÷ 9.5), applied voltage (10 ÷ 20 kV), and injection time (10 ÷ 60 s, Table 1). All statistical calculations were developed by the software Design Expert, v.10.0.4 (USA). Table 1. The values corresponding to -1, 0, +1, -2 (-α) and +2 (+α) in the experimental design Variables Levels -2 (-α) -1 0 +1 +2 (+α) X1 (Tris concentration, mmol/L) 30 60 90 120 150 X2 (pH) 8.5 8.75 9.0 9.25 9.5 X3 (applied voltage, kV) 10 12.5 15 17.5 20 X4 (injection time, s) 10 22.5 35 47.5 60 * Method validation The method was validated using the following performance criteria: limit of detection (LOD), limit of quantitation (LOQ), linearity, linear range, recovery and repeatability. LOD includes instrument detection limit (IDL), the method detection limit (MDL). The IDL was determined by measuring the standard solution (0.5 mg/L per each analyte), calculating the analyte concentrations that produce a signal/noise (S/N) ratio = 3. MDL was determined similarly but with a spiked sample of 2 mg/L per each analyte (on the real sample without analytes). The corresponding LOQ values are determined by the expression: LOQ = 3 LOD. The linearity, linear range were established through the Using a central composite design to optimize the analytical method for simultaneous determination 29 calibration curve obtained by triplicate analysis of choline, pantothenic acid, ascorbic acid and niacin at five concentration levels 2.0, 5.0, 10, 20, and 40 mg/L. Repeatability was determined by intra-assay and inter-assay precision. The intra- assay precision of the method expressed as the relative standard deviation (RSD) of peak area and migration time measurements (n = 11), was evaluated through the results obtained with the method operating over one day under the optimal condition, using solution at concentration of 20 mg/L of each analyte. The inter-assay precision was determined for the same solution but was performed for seven days (n = 7). Recovery was determined by standard addition method on deionized water and real samples at concentrations of 2.5 mg/L and 25 mg/L. Repeatability was determined by repeated measurements of intraday (11 times) and daily (7 days) for the 20 mg/L standard solution and calculated the RSD of the peak area and migration time. 2.2. Results and discussion 2.2.1. CZE method development * Investigation of BGE’s composition The BGE is an important factor that affects the detection and separation of the analytes. Among the analytes, pantothenic acid, ascorbic acid and niacin all contain the acid functional group (-COOH) in the molecule, so it would be preferable if they were detected as anions. In contrast, choline is a quaternary ammonium salt and should be detected in cationic form. For the simultaneous determination of four analytes in one run, the use of reverse polarity technique proves to be most appropriate. At this point, the positive voltage is applied to the electrode at the side that sample is injected while the electrode on the detector side is grounded to produce a negative polarization (towards the detector). It can be predicted that choline, in the form of cation, will move towards the cathode. Interestingly, in reverse polarization, under the effect of electroosmotic flow (EOF), the anion forms of vitamins will also move towards the cathode (in the same direction as the cation of choline) and will be detected by the C 4 D detector. To do this, the pH of the BGE must be relatively high to ensure that the EOF is powerful enough to be able to bring vitamins, which are quite large compounds. Figure 1. Investigation results of the background electrolytes solution Nguyen Thanh Dam, Nguyen Manh Huy, Vu Minh Tuan, Duong Hong Anh and Pham Hung Viet 30 In six BGEs used in this investigation, the Tris/CHES (pH 9.5) and Tris/Ace (pH 9.5) buffers gave the sharpest signal and highest peak area. However, in the case of the Tris/CHES buffer, after the peak of niacin signal, there was a significant rise in the background, which may affect the quantification of niacin (Figure 1). Thus, the Tris/Ace buffer was chosen as the background electrolyte for subsequent optimization. * Optimization design The central composite design consists of three parts: (i) a full factorial or fractional factorial design; (ii) an additional design, often a star design in which experimental points are at a distance α from its center; and (iii) a central point. This design presents the following characteristics: (1) Require an experiment number according to N = 2 k +2k+cp, where k is the factor number and (cp) is the replicate number of the central point; (2) α-values depend on the number of variables and can be calculated by α =2k/4. For example, if there are four variables, the α-value will be 2.00; (3) All factors are studied in five levels (−α, −1, 0, +1, + α). In this work, an experimental design using CCD with four factors was conducted: concentration of Tris (30 - 150 mmol/L), pH (8.5 - 9.5), voltage (10 - 20 kV), and injection time (10 - 60 s). In CE method, these are the factors that affect the separation, in which concentration of electrolyte and pH are the chemical factors while voltage and injection time are factors of the device. The experimental design was constructed by the use of a full 2 4 factorial design with six central and eight axial points. Therefore, the overall matrix of CCD design involved 30 experiments. This design was used to optimize and evaluate main effects, interaction effects, and quadratic effects. The nonlinear computer-generated quadratic model is given as: where Xi are the independent variables (factors); b0 is an intercept; bi, bii, bij are the regression coefficients; and Y is the response function. In this study, the peak areas of each target analyte are responses and denoted Y1, Y2, Y3, Y4 corresponding to the peak area of choline, pantothenic acid, ascorbic acid, and niacin, respectively. The experiments and the results of the model are presented in Table 2. The separation of choline, pantothenic acid, ascorbic acid, and niacin under study is achieved at a concentration of 150 mmol/L Tris, adjusted to pH 9.5 by acetic acid, voltage 10 kV, and 60 s of injection time. The responses of the model, R 2 values, were greater than 0.87, implying that the model was well fitted by the data for the response of area of peaks. The F-test for the models showed that the p-values of all responses were smaller than 0.0001 while the lack of fit F-test had p-values that were greater than 0.1 (0.1155 - 0.8172). This indicated a good correlation of the obtained regression equation, which shows that the obtained models were consistent and had high statistical reliability. In addition, a low coefficient of variation (CV%), less than 10% (1.21 - 9.48%), indicated that the experiments had high repeatability (Table 3). Using a central composite design to optimize the analytical method for simultaneous determination 31 Table 2. Experimental matrix and the results Run X1 X2 X3 X4 Y1 Y2 Y3 Y4 mmol/L kV s V.s V.s V.s V.s 1 30 9 15 35 0.005 0.027 0.015 0.024 2 60 8.75 12.5 22.5 0.006 0.022 0.010 0.020 3 60 8.75 12.5 47.5 0.008 0.044 0.022 0.048 4 60 8.75 17.5 22.5 0.004 0.014 0.007 0.013 5 60 8.75 17.5 47.5 0.006 0.030 0.015 0.030 6 60 9.25 12.5 22.5 0.005 0.022 0.010 0.019 7 60 9.25 12.5 47.5 0.008 0.046 0.019 0.041 8 60 9.25 17.5 22.5 0.004 0.017 0.007 0.015 9 60 9.25 17.5 47.5 0.006 0.032 0.013 0.028 10 90 8.5 15 35 0.005 0.022 0.012 0.024 11 90 9 10 35 0.007 0.035 0.021 0.035 12 90 9 15 10 0.003 0.012 0.005 0.013 13 90 9 15 35 0.006 0.028 0.009 0.022 14 90 9 15 35 0.006 0.027 0.014 0.028 15 90 9 15 35 0.007 0.028 0.009 0.025 16 90 9 15 35 0.006 0.027 0.009 0.028 17 90 9 15 35 0.006 0.027 0.008 0.030 18 90 9 15 35 0.006 0.026 0.008 0.026 19 90 9 15 60 0.009 0.052 0.024 0.057 20 90 9 20 35 0.004 0.018 0.010 0.018 21 90 9.5 15 35 0.005 0.031 0.017 0.029 22 120 8.75 12.5 22.5 0.005 0.017 0.006 0.018 23 120 8.75 12.5 47.5 0.009 0.038 0.012 0.041 24 120 8.75 17.5 22.5 0.003 0.012 0.004 0.013 25 120 8.75 17.5 47.5 0.005 0.024 0.007 0.026 26 120 9.25 12.5 22.5 0.005 0.023 0.008 0.022 27 120 9.25 12.5 47.5 0.010 0.051 0.015 0.046 28 120 9.25 17.5 22.5 0.004 0.015 0.005 0.015 29 120 9.25 17.5 47.5 0.006 0.037 0.011 0.032 30 150 9 15 35 0.006 0.026 0.010 0.029 Nguyen Thanh Dam, Nguyen Manh Huy, Vu Minh Tuan, Duong Hong Anh and Pham Hung Viet 32 Table 3. The results of analysis of variance and model testing Response Model Lack of fit R 2 CV(%) F p-value F p-value Y1 38 < 0.0001 2.56 0.1520 0.92 9.48 Y2 285 < 0.0001 2.99 0.1155 0.99 1.21 Y3 24 < 0.0001 0.78 0.6826 0.87 3.88 Y4 95 < 0.0001 0.58 0.8172 0.97 8.23 Before proceeding to validate the CZE method for the determination the vitamins in food supplements, the optimal conditions from the model was tested by comparing the analysis results with the predicted value of the model. As shown in Table 4, the test results were in the range predicted by the model and the deviation from the predicted values of the model was less than 10%. Thus, the predictive values of the model were highly accurate, in other words, the model was consistent. Figure 2. The 3D response surface shows the effect of the pair of variables on the peak area of choline Using a central composite design to optimize the analytical method for simultaneous determination 33 Table 4. The results of the experiment used to test the analytical conditions Peak area Predicted range (V.s) Predicted value (V.s) Analysis results (V.s) Choline 0.0110 - 0.0154 0.0130 0.0150 Pantothenic 0.0869 - 0.1234 0.1037 0.0777 Ascorbic 0.0303 - 0.1002 0.0560 0.0366 Niacin 0.0720 - 0.0890 0.0804 0.0740 2.2.2. Validation of the analytical method The CZE method was validated for the analyses of analytes by evaluation of the following parameters: LOD, LOQ, linear range, linearity, intra- and inter-assay precision and accuracy. The results are summarized in Table 5. It can be seen that the method had low LODs and LOQs. IDLs of choline, pantothenic acid, ascorbic acid and niacin were 0.12; 0.06; 0.13 and 0.09 mg/L, respectively. MDL and MQL were also low, MDL ranged from 0.22 to 0.50 mg/L, and MQL varied from 0.74 to 1.68 mg/L; pantothenic acid had the best, and ascorbic acid had the highest LOQ. The linear ranges of the analytes were extensive, with the range of niacin was widest, from 0.98 to 500 mg/L, while choline gave the narrowest linear range (1.17 to 200 mg/L). Repeatability for both peak area and migration time were very good, RSD of peak areas and migration times for intra-assay (n = 11) were lower than 4% and 0.5%, and for inter-assay (n = 7) were not greater than 10% and 4%, respectively. Table 5. Quantitative features for choline, pantothenic acid, ascorbic acid and niacin Parameter Choline Pantothenic acid Ascorbic acid Niacin IDL (mg/L) 0.12 0.06 0.13 0.09 IQL (mg/L) 0.36 0.18 0.39 0.27 MDL (mg/L) 0.45 0.22 0.50 0.35 MQL (mg/L) 1.52 0.74 1.68 1.15 Linear range (mg/L) 1.17 - 200 0.6 - 250 1.29 - 250 0.98 - 500 Linearity (R 2 ) 0.999 0.999 0.998 0.999 Intra-assay precision (20 mg/L, n = 11) RSD a % 1.7 1.5 1.7 3.3 RSD b % 0.4 0.2 0.2 0.2 Inter-assay precision (20 mg/L, n = 7) RSD a % 7.7 6.0 9.3 6.0 RSD b % 1.5 2.8 3.0 3.4 RSD a : relative standard deviation of peak area; RSD b : relative standard deviation of migration time Nguyen Thanh Dam, Nguyen Manh Huy, Vu Minh Tuan, Duong Hong Anh and Pham Hung Viet 34 The recovery of each analyte in standard solutions (deionized water background) at a concentration of 2.5 mg/L and 25 mg/L were 94 - 100% and 93 - 96%, respectively. In real sample background, recoveries were good at 89 to 111% and 91 to 109% for standard additions of 2 mg/L and 10 mg/L, respectively (Table 6). According to AOAC, these values are acceptable [1]. Figure 3. Electropherogram of choline, pantothenic acid, ascorbic acid, and niacin Table 6. Recoveries (%) of