Development of Simple Analytical Method for B-Group Vitamins in Nutritional Products: Enzymatic Digestion and UPLC-MS/MS Quantification

A method for the simultaneous determination of seven B-group vitamers including thiamine, riboflavin, nicotinamide, niacin, pyridoxine, pyridoxal, and pyridoxamine in nutritional products by using enzymatic digestion followed by LC-MS/MS quantification was studied. The LC-MS/MS conditions such as MS transitions, mobile phase programs, and ammonium formate buffer concentrations, and sample treatment procedures (e.g., concentrations of buffer solution, digestion temperature, and digestion time) were investigated. The analytical method performance was evaluated by multiple criteria such as selectivity, linearity, detection and quantification limits, repeatability, reproducibility, and recovery by using real sample matrices. The validated method was successfully applied to analyze vitamin B concentrations in different nutritional products like ultra-heat-treated milk, powdered milk, and nutritional powder. Vitamin B concentrations varied over a wide range from lower than detection limits to about 9000 µg/100 g, depending on vitamin groups, compound forms, and sample types. The measured concentrations of B-group vitamins in our samples were generally in good agreement with values of label claims.


Introduction
e B-group vitamins comprise eight types of water-soluble vitamins (i.e., vitamin B1, B2, B3, B5, B6, B7, B9, and B12) and one type can exist in different forms, also called vitamers [1]. Vitamins in general and B-group vitamins in particular are essential for maintaining various body functions in humans, and vitamin deficiencies may cause adverse health effects [1,2]. Because foods are the most important vitamin sources, a nutritious and balanced diet can eliminate deficiencies [1][2][3]. However, negative impacts of vitamin overdose have also been noted [4].
e World Health Organization has tabulated recommended vitamin amounts in human nutrition [2]. e development of a reliable and effective method for the simultaneous determination of multiple vitamin types and forms in food and nutritional products is needed to characterize the occurrence of such nutrients in food and to provide useful information for regulatory agencies, manufacturers, and consumers [5,6]. In order to extract B-group vitamins from food matrices, several extraction and purification methods were applied, e.g., solid-phase extraction (SPE) with C18 adsorbent [7,8], dispersive SPE with molecularly imprinted biopolymers [9], treatment with precipitation reagents [5,10], and extraction with acetonitrile [6,11]. Each method has its own advantages and limitations. From a green chemistry point of view, the use of organic solvents [6,11], heavy metal salts [5], and potentially toxic chemicals like trichloroacetic acid [10] should be avoided. In addition, strong acid hydrolysis may be responsible for the appearance of some impurities, which interfere with chromatographic signals of some B-group vitamins [5]. e application of enzymatic digestion to extract B-group vitamins from complex food matrices rich in proteins and carbohydrates has been reported by various studies [12,13]. As for instrumental analysis of B-group vitamins, liquid chromatography with mass spectrometry detection (especially ultra-performance liquid chromatography-tandem mass spectrometry UPLC-MS/MS) has become the standard method due to its outstanding specificity, sensitivity, and efficiency [12][13][14].
In the present study, four types with seven forms of B-group vitamins were examined, including vitamin B1 (thiamine), B2 (riboflavin), B3 (nicotinamide and niacin), and B6 (pyridoxine, pyridoxal, and pyridoxamine). We focused on three nutritional products such as ultra-heattreated milk, powdered milk, and nutritional powder, which are usually fortified with micronutrients including B-group vitamins. e samples were treated by a simple and "green" extraction method, which utilized enzymatic digestion without the use of organic solvents and heavy metal salts. Analytical method performance (i.e., UPLC-MS/MS with isotope dilution/internal standard quantification) was strictly validated, showing adequate specificity and accuracy. Concentrations of individual vitamers over a wide range (from µg to mg per 100 g) in nutritional products were reported with comprehensive insights into their total levels, profiles, and validity of nutrition food labels. To our knowledge, this is among the first studies to investigate the occurrence of multiple B-group vitamers in nutritional products in Vietnam as well as Southeast Asia.

Instrumentation and Optimization of LC-MS/MS
Conditions. In this study, a liquid chromatograph (ACQ-UITY UPLC H-Class; Waters) equipped with a tandem mass spectrometer (Xevo TQD; Waters) and an ACQUITY UPLC BEH C18 column (100 mm × 2.1 mm × 1.7 μm, 130Å; Waters) was used. Basic parameters of the LC-MS/MS system in this study are summarized in Table 1.
Based on the basic parameters shown in Table 1 with mobile phase A as 20 mM ammonium formate solution referred from the AOAC Official Method 2015.14 [14], we performed three experiments to subsequently determine the following: (1) MS transitions of target compounds and internal standards; (2) effect of mobile phase programs on retention times and separation efficiency; and (3) effect of ammonium formate concentration in mobile phase A on signal intensities of the analytes. e MS transition parameters (e.g., quantitative and qualitative ions, cone voltage, and collision energy) were automatically optimized. Different programs of the mobile phase were investigated, including one isocratic program and three gradient programs ( Table 2). Concentrations of ammonium formate for optimization ranged from 5 to 50 mM.

Optimization of Sample Preparation Procedure.
Optimization experiments for the sample preparation procedure were performed by using a nutritional powder sample. Extraction conditions showing highest recovered amounts of vitamins in this sample were selected. We investigated the effects of (1) concentrations of ammonium formate in a range of 5 to 100 mM, (2) digestion temperature in a range of 35 to 50°C, and (3) digestion time in a range of 3 to 16 h, on extraction efficiency of 7 vitamins. e nutritional powder sample (10 ± 0.3 g) was reconstituted in distilled water (total weight 100 ± 2 g) by using a magnetic stirrer. An aliquot of 1 g reconstituted sample was transferred to a 50 mL tube, spiked with internal standards (100 µL of ISSM standard) and 5 mL of the enzyme cocktail solution, and mixed by using a vortex mixer. e sample tube was then incubated in a thermostatic shaker overnight. e extract was transferred to a 25 mL volumetric flask, made up to 25 mL with 50 mM ammonium formate solution, and filtered through a 0.2 µm PTFE membrane before LC-MS/MS analysis.

Method Validation.
e specificity/selectivity and linearity of quantification for seven vitamers were confirmed according to the criteria proposed by the AOAC International [14] and the Commission Decision 2002/657/EC of the European Communities [15]. Based on the optimized procedure, additional nutritional product samples were analyzed to evaluate method performance. ree types of sample matrices were examined including ultra-heat-treated (UHT) milk, powdered milk, and nutritional powder. For each sample type, repeatability (within-day replicate analysis of a real sample, n � 6), reproducibility (between-days replicate analysis of a real sample, n � 4), method detection and quantification limits (replicate analysis of a low-concentration sample, n � 10), and recovery (replicate analysis of matrix-spiked samples at three spiking levels, n � 3 for each level) were evaluated. e accuracy of our method was also confirmed by method development and analytical results of infant formula samples from the AOAC's interlaboratory study (in 2018) and the Nestlé's internal tests (in 2019 and 2020).

Application of the Validated Method to Nutritional
Products. e validated analytical method was applied to determine concentrations of 7 B-group vitamers in different nutritional products including UHT milk (n � 5), powdered milk (n � 5), and nutritional powder (n � 5) samples. All the samples were obtained randomly from several dealerships and groceries in Hanoi City, Vietnam, between 2019 and 2020.

MS/MS Conditions.
e MS/MS conditions including mass transitions, cone voltage, and collision energy of seven vitamers and respective internal standards were automatically optimized and the results are shown in Table 3. e mass transitions of the target compounds in our study are in good agreement with those reported by the AOAC Official Method 2015.14 [14]. For each compound, one quantitative ion (product ion) and two qualitative ions (precursor ion and one product ion) were assigned. is scheme is in accordance with performance criteria of LC-MS/MS proposed by the Commission Decision 2002/657/ EC, showing 4 identification points [15]. e MS/MS conditions tabulated in Tables 1 and 3 were used for next experiments.

Mobile Phase Programs.
With mobile phase A as 20 mM ammonium formate solution and mobile phase B as methanol, we analyzed the working standard WS5 (individual concentrations ranging from 6 to 600 ng/mL) under four programs listed in Table 2. Retention times of the target compounds are shown in Table 4 and chromatograms are presented in Figure S1 of Supplementary data. In the isocratic program, all the vitamers were eluted within 2.5 min and some compounds were almost coeluted (e.g., NIC, PYL, and PYN). In addition, peaks of native and internal standards of NIA and PYM were not sharp and balanced (i.e., tailing peaks). A suitable gradient program is therefore necessary. e program Gradients 1 and 2 gave tailing peaks for NIA and PYM. e program Gradient 3 was selected because it provided well-resolved, sharp, and balanced peaks of all the compounds.

Effect of Ammonium Formate Concentrations in Mobile Phase.
e mobile phase A with concentrations of ammonium formate of 5, 10, 20, and 50 mM was investigated. Signal intensities of seven vitamers in the working standard WS5 are presented in Figure 1. In general, signals   decreased with the increasing salt concentrations. e intensities were not significantly different with salt concentrations from 5 to 20 mM, which were much higher than those from 50 mM. With salt concentration of 5 mM, signals of RIB, PYM, and PYL were not balanced with shoulder or tailing peaks. erefore, we selected mobile phase A as 10 mM ammonium formate solution. Some other studies used 20 mM ammonium formate [14,16], but concentration of 10 mM was also applied elsewhere [17]. Finally, the mobile phases A (10 mM ammonium formate) and B (methanol) were used with program Gradient 3 ( Table 2). Total runtime was 10 min.

Effect of Ammonium Formate Concentrations.
e test samples (about 1 g of reconstituted nutrition powder) were spiked with internal standards and digested with 5 mL of enzyme cocktail at 37°C for 12 h. en, the digested mixtures were transferred to 25 mL volumetric flasks and diluted with     Figure 2. Accordingly, the highest extraction efficiency of all compounds was obtained at salt concentration of 50 mM. is solution (i.e., 50 mM ammonium formate) was also used by the AOAC Official Method 2015.14 [14].

Effect of Digestion Temperature.
e test samples (about 1 g of reconstituted nutrition powder) were spiked with internal standards and digested with 5 mL of enzyme cocktail solution at for 12 h at different temperatures as 35, 37, 40, 45, and 50°C. en, the digested mixtures were transferred to 25 mL volumetric flasks and diluted with 50 mM ammonium format solution. Concentrations of seven vitamers in the test samples with different digestion temperatures are shown in Figure 3. It is obvious that the sample treated at 37°C has the highest concentrations of all compounds. From 40°C onwards, amounts of vitamins recovered from the matrix decreased with the increasing temperature. Among B-group vitamins, vitamin B1 (THI) is relatively sensitive to high temperature as compared to others (e.g., vitamin B2 and B6) [18]. However, the temperature range investigated in this study did not exceed 50°C and is still lower than normal water temperature used to prepare formula (e.g., about 70°C) [19]. In addition, concentrations of THI did not change significantly over the temperature range from 37 to 50°C (RSD � 7%), suggesting that vitamin stability is not likely to be an important factor. erefore, the extraction efficiency was largely attributed to enzyme activities, which were strongly affected by incubation temperature. e optimum temperatures of papain, α-amylase, and acid phosphatase were 65, 37, and 37°C, respectively [20][21][22]. As a result, digestion temperature higher than 37°C may inhibit enzymatic activity of α-amylase and acid phosphatase, which prevents complete liberation of free vitamins. Based on the optimization results, digestion temperature of 37°C was selected.

Effect of Digestion Time.
e test samples (about 1 g of reconstituted nutrition powder) were spiked with internal standards and digested with 5 mL of enzyme cocktail solution at 37°C over different periods as 3, 6, 9, 12, 14, and 16 h. As shown in Figure 4, extracted amounts of vitamins from the matrix gradually increased from 3 h to 14 h and then decreased after 16 h digestion. Concentrations of most compounds obtained after 12 h and 14 h digestion were not significantly different. erefore, the optimal digestion time varied between 12 h and 14 h. As compared to other sample preparation methods (e.g., protein precipitation and acid hydrolysis), operation time of the enzymatic digestion method is much longer [12]. In our laboratory, the samples were incubated overnight, which helps to save time during working hours. Furthermore, accuracy is minimally dependent on small changes in digestion time by working near the peak of the signal/digestion time relationship.

Optimal Sample Preparation
Procedure. Before analysis, the samples (i.e., UTH milk, powdered milk, and nutritional powder) were homogenized thoroughly. e powdered milk or nutritional powder sample (10 ± 0.3 g) was reconstituted in distilled water (total weight 100 ± 2 g) by using a magnetic stirrer. An aliquot of 1 g UHT milk or reconstituted powder sample was transferred to a 50 mL tube, spiked with internal standards and 5 mL of the enzyme cocktail solution, and mixed by using a vortex mixer. e sample tube was then incubated at 37°C in a thermostatic shaker overnight (about 12 to 14 h). e extract was transferred to a 25 mL volumetric flask, filled up to 25 mL with 50 mM ammonium formate solution, and filtered through a 0.2 µm PTFE membrane before LC-MS/MS analysis.

Method Validation Results
3.4.1. Method Specificity/Selectivity. As described above, the specificity and selectivity of our analytical method meet the requirements of the AOAC International (i.e., each compound was confirmed by one precursor ion and two product ions) [14] and the Commission Decision 2002/657/EC (i.e., 4 identification points) [15]. Procedural blank samples (n � 10) were analyzed together with test and real samples, showing no signal of any target compound. Possible contamination from chemicals, reagents, and laboratory ware and environment was excluded. Results of the standards, samples, and matrix-spiked samples also indicated good agreement between signals (i.e., retention time and mass transitions) of all compounds from standards and from samples. In addition, the isotope dilution/internal standard method for quantification utilizing stable isotope-labelled compounds has been demonstrated as the most appropriate technique for quality control in LC-MS/MS. In general, our quantification method is specific and selective for the determination of multiple B-group vitamin forms in food matrix. Representative chromatograms of vitamers and internal standards in each sample type are shown in Figure S2, indicating distinct and clear appearance of these compounds in sample matrices.

Calibration Curves.
We prepared seven working standard solutions (WS1 to WS7) to construct calibration curves for all the target compounds. Concentration ranges of seven vitamers in the calibration standards were as follows: THI (2.7-136), RIB (11.5- working ranges. e biases (relative errors from known concentrations) were less than ±8% for all compounds, which meet the requirements of the AOAC International [14].

Method Detection Limits, Quantification Limits, and
Quantification Ranges. Based on screening results, we selected real samples with low vitamin concentrations (i.e., individual concentrations ranging from 1.70 ± 0.10 for PYL to 11.9 ± 1.0 µg/100 g for NIC) and performed replicate analysis (n � 10) to determine method detection limits (MDLs) and quantification limits (MQLs). e MDLs and MQLs were estimated as three times and ten times of standard deviations, respectively. As shown in Table 6, MDLs of seven vitamers ranged from 0.28 to 2.8 µg/100 g for liquid samples and from 2.8 to 28 µg/100 g for powder samples, showing high sensitivity of our method. e quantification ranges span over three orders of magnitude, indicating that the method can be applied for samples with wide vitamin concentration ranges (i.e., µg to mg per 100 g).

Method Repeatability, Reproducibility, and Recovery.
e repeatability (within-day replicate analysis of real samples, RSD r ) and reproducibility (between-days replicate analysis of real samples, RSD R ) of our method were determined for three sample types including UHT milk, powdered milk, and nutritional powder.
ere is no significant difference in the RSD r (2.9-9.1%) and RSD R (2.1-9.8%) values of individual vitamers between the three sample types (Table 7). e RSD values of our method were lower than 10%, indicating adequate precision. Our RSD values also satisfy the criteria proposed by the AOAC International for RSD r < 15% and RSD R < 22% at a concentration level of 100 ppb [23]. Representative samples of the three matrices were spiked with known amounts of vitamin standards at three concentration levels and analyzed to determine method recovery. e spiking levels varied between compounds and sample types, spanning over a wide range from 50 to 5000 µg/100 g. e recoveries of all target compounds ranged from 80 to 107%, showing good accuracy. ese recovery values fall within the acceptable range (80 to 110%) expected by the AOAC International for concentration levels of 100 ppb to 10 ppm [23]. e reliability of our method was also confirmed by method development results of infant formula samples from the AOAC's interlaboratory study in 2018 [14]. In addition, we also joined the Nestlé's internal tests in 2019 and 2020 for infant formula samples. Ratios of vitamin B1, B2, B3, and B6 concentrations measured by our method and the assigned values ranged from 79 to 113% with z-score values ≤2, suggesting acceptable levels of accuracy.

Concentrations of B-Group Vitamins in Nutritional
Products.
e validated method was applied to analyze concentrations of seven B-group vitamers in nutritional products obtained from dealerships and groceries in Hanoi City, Vietnam. e analytical results of real samples are tabulated in Table 8. THI, RIB, NIC, and PYN were quantified in all the samples. NIA was detected in milk      Journal of Analytical Methods in Chemistry usually reported as "vitamin B3" or "niacin," although NIA was found at low levels or even not detected. Our findings were in good agreement with those reported for infant formula products [24,25]. ese observations suggest the need for manufacturer responsibility to report exact nutrition information in their product labels. NIC was also more abundant than NIA in human milk [26] and fresh cow, goat, and buffalo milk [27]. In our samples, PYN accounted for 90 to 100% (98 ± 4%) of total vitamin B6. In the powdered milk samples, proportions of PYN, PYL, and PYM in total vitamin B6 were 93 ± 3%, 4 ± 2%, and 2 ± 1%. Significant levels of PYL were found in infant formula samples (22-26% of PYN + PYL) [14]. PYL was major vitamer found in human milk (87-97% of PYN + PYL + PYM) [26]. Besides, significant percentages of PYL and some other vitamin B6 forms (e.g., pyridoxal 5′-phosphate and 4-pyridoxic acid) were detected in fresh milk samples [27]. erefore, detailed and comprehensive analysis of different vitamin forms, rather than total vitamins and major vitamers, is needed in future studies.

Conclusions
In this study, a simple and solvent-free analytical method for the extraction of B-group vitamins in nutritional products was studied. Liquid samples (including UHT milk and reconstituted milk and nutritional powder) were digested with enzyme mixtures, diluted by ammonium formate solution, and filtered before quantification by using LC-MS/ MS method. Our optimized method shows adequate specificity, accuracy, repeatability, reproducibility, and sensitivity for the simultaneous determination of seven B-group vitamers at a wide concentration range up to three orders of magnitude (e.g., µg to mg per 100 g). e method was successfully applied to measure vitamin B concentrations in several nutritional products purchased from Hanoi City, Vietnam, showing agreement between measured and labelled values of total vitamin levels. However, the discrepancies resulting from vitamer recognition and contribution (i.e., for vitamins B3 and B6) should be taken into account. Our results suggest that detailed characterization of vitamin forms in food and nutritional products is necessary.
Data Availability e main part of the research data is included within the article. Other data can be made available from the corresponding author upon request.

Conflicts of Interest
e authors declare that there are no conflicts of interest regarding the publication of this paper.