Simultaneous determination of dehydroacetic acid, benzoic acid, sorbic acid, methylparaben and ethylparaben in foods by high-performance liquid chromatography

In this study, an analytical method was established and validated to determine the preservatives such as dehydroacetic acid, benzoic acid, sorbic acid, methylparaben and ethylparaben. The level of preservatives was measured by solvent extraction method adding purification process with carrez reagent and by high-performance liquid chromatography (HPLC). The developed analytical method was successfully applied to determine the concentration of preservatives in various food samples including jam, cheese and soy sauce, displaying high accuracy (recoveries between 87.8% and 110%) and precision (%RSD less than 5.92% and 7.72% for intra-day and inter-day, respectively). To verify the applicability of the improved test method, selected 13 food items and collected 521 samples were monitored. As a result, all the cases met the Korea standard guidelines. Consequently, this study is expected to contribute to the safety management of preservatives for domestic distribution and imported food.


Introduction
Preservatives refer to the food additives for preventing a food spoilage resulting from the growth of microorganisms. In Korea, the common preservatives used for food preservation include dehydroacetic acid, sorbic acid, benzoic acid, paraoxybenzoic acid esters, propionic acid and salts thereof (MFDS, 2020). Internationally, the regulations for preservatives are precisely set in terms of the using conditions, the types of foods that can be used and the maximum used amount (Guarino et al., 2011). Each country has different tendency in the commercial preservatives. Further, what can be legally utilized may not be imported, distributed, or sold, since the preservatives do not meet the standards of the domestic food code (Shin, 2016). Therefore, to manage through the accurate preservative inspection process is crucial. In order to comply with forthcoming regulations and to have effective control over food safety, a rapid and reliable method of determining food preservatives is required. Several studies have been undergone on the determination of sorbic acid and benzoic acid in food samples. Many analytical methods, including high performance liquid chromatography (HPLC) (Phechkrajang and Yooyong, 2017;Jang et al., 2018;Timofeeva et al. 2019) and gas chromatography (GC) (Park et al., 2016;Ding et al., 2018;Tungkijanansin et al., 2020) were applied for the determination of these preservatives in various foods. As new sorts of foods continue to emerge and existing foods are tended to be combined with lots of preservatives, a superior purification process which can effectively reduce the food matrix is required. Further, an accurate separation and quantification analyses are needed. Throughout analyzing food, the complex food matrix hinders a quantitative analysis of separation technology. In real samples, the elimination of interfering substances from the food matrix is to be considered significant for accurate measurement of analytes. A sample preparation is applied using several extraction methods, such as liquid-liquid extraction (LLE) (Tighrine et al., 2019), dispersive liquid-liquid microextraction (DLLME) (Sun et al., 2014;Javanmardi et al., 2015;Ding et al., 2018), solid-phase microextraction (SPME) (Tungkijanansin et al., 2020) and solid-phase extraction (SPE) (Win et al., 2019). In the International Dairy Federation (IDF), extraction using only solvent may not sufficiently remove interfering substances, emulsified sample was analyzed using carrez solution (ISO, 2008). In order to efficiently analyze processed foods, among the Spe methods, HLB and C18 cartridges, which are excellent for removing proteins, salts and organic acids, and carrez solution, which are excellent for removing proteins and lipids.
The main objective of this study is to develop and validate an analytical method applicable to all food types where preservatives added (dehydroacetic acid, benzoic acid, sorbic acid, methylparaben, ethylparaben) for safety reasons. Currently, Korea Food Code proposed a steam distillation method and a solvent extraction method. Although the steam distillation is widely applied, it is time-consuming and complicating. Additionally, the solvent extraction method is limited to certain food types, and it does not go through any purification process, which makes the analysis difficult due to the interfering peaks of the matrix. In this study, three purification procedures (SPE(C18, HLB cartridge), carrez reagent solution) were compared in order to establish an optimal simultaneous analysis with five preservatives using HPLC. Further, a monitoring was conducted through the commercially processed foods to verify the applicability with the established analysis method for the preservatives. As a result, using HPLC, this study developed the fastest and simplest simultaneous analysis method for preservatives. The developed analytic method would enable the simultaneous analysis of preservatives (dehydroacetic acid, benzoic acid, sorbic acid, methylparaben, ethylparaben), and could be utilized for all types of food. As the pretreatment process is simple, and the purification effect is excellent, it will save much time and cost.
Meanwhile, validation was conducted by selecting the representative processed foods from food groups with standards for use of the preservatives. Jam was selected as a representative with high in sugar, soy sauce as the one with high salt, and cheese as the one having protein and fat.

Reagents
The standard-grade dehydroacetic acid, sorbic acid, benzoic acid, methylparaben, ethylparaben, tetra-n-butylammonium hydroxide and phosphoric acid were purchased from Sigma-Aldrich (St. Louis, MO, USA). Ethanol and acetonitrile for an extraction solvent, were obtained from Merck (Frankfurt, Germany). Ultrapure water (resistivity ≥ 18 MΩ) was obtained from a Milli-Q ultrapure water purification system (Millipore, Billerica, MA, USA). Potassium ferricyanide (Sigma-Aldrich, St. Louis, MO, USA) and zinc sulfate heptahydrate (Merck, Frankfurt, Germany) were used as the carrez solution I, carrez solution II. All regents used were analytical grade or better.

Preparation of solutions
Individual standard stock solutions were prepared by dissolving 100 mg of dehydroacetic acid (DHA), sorbic acid (SA), benzoic acid (BA), methylparaben (MP), ethylparaben (EP) in 100 ml of methanol using 100 ml volumetric flasks. With different concentrations, working standard solutions for DHA, SA, BA at concentrations of 0.3, 0.6, 1.2, 3, 6, 15, 30 and 60 mg/kg and parabens (EP, MP) at concentrations of 0.5, 1, 5, 10, 25, 50 and 100 mg/kg were produced by diluting known aliquots of the stock solutions with methanol. The stock solutions were prepared monthly while the working standard solutions were prepared daily. Carrez solution I was used by dissolving potassium ferrocyanide in water to a concentration of 15%, and Carrez solution II was used by dissolving zinc sulfate in water to a concentration of 30%.

Comparison of recovery according to purification process
SPE method using C18 cartridge C18 cartridges are known to be excellent at removing salts and organic acids, and have been studied using preservatives (Yun et al., 2019). For the recovery rate test, 40 mg/kg was added to the representative samples (jam, soy sauce, cheese). The solvent extraction of preservatives (DHA, SA, BA, MP, EP) in the samples of jam, soy sauce, cheese followed by respective purification by SPE was performed according to a procedure reported by Yun et al. (2019) with some modifications. In the solvent extraction process, 4 g of the homogenized sample was placed in a conical tube containing 40 ml ethanol and then sonicated for 10 min followed by centrifugation for 10 min at 1500 × g. During the purification, 5 ml of the supernatant was passed through an activated C18 SPE column (Waters, MA, US; the column was preconditioned with 5 ml methanol, 5 ml water and 1 ml acetonitrile before activation) followed by 5 ml of methanol. The total solution was dried using nitrogen, and the analytes were eluted with 1 ml of acetonitrile and filtered through a polyvinylidene fluoride (PVDF) 0.22 μm membrane syringe filter (Teknokroma, Barcelona, Spain).

SPE method using HLB cartridge
HLB cartridges are specialized for protein, salt, and lipid removal and comparative experiments were conducted with some modifications referring to the experimental method of Wang et al. (2013). This procedure is identical as the one using C18, and was experimented with only different cartridge by HLB.

Purification using carrez reagent
The solvent extraction and purification of preservatives in foods by carrez solutions were performed according to the procedure reported by Abedi et al. (2014) with some modifications. During the solvent extraction process, 4 g of the sample was placed in a conical tube. After adding 10 ml of ethanol, 1 ml potassium hexaferrocyanide (Carrez solution I) and 1 ml zinc acetate (Carrez solution II) were added. Then, the closed tube was shaken gently for 1 min to promote formation of a cloudy solution. The resultant was combined with ethanol 40 ml and sonicated for 5 min followed by centrifugation for 10 min at 1500 × g. The supernatant was filtered through PVDF 0.22 μm membrane syringe filter.

Method validation
The analytical method for the determination of DHA, SA, BA, MP and EP in the products were validated in terms of the linearity, limit of detection (LOD), limit of quantification (LOQ), selectivity, accuracy (recovery), precision (as percent relative standard deviation (%RSD)) and measurement uncertainty. The method was assessed according to the AOAC (Association of Official Analytical Chemistry) guideline for selectivity, accuracy, precision, LOD, LOQ (AOAC, 2012b), the GUM (guide to the expression of uncertainty in measurement) and EURACHEM guide for the measurement of uncertainty (ISO, 1993;EURACHEM, 2000). Linearity was evaluated using three replicates of calibration curve and calculating the correlation coefficient (r 2 ). The LOD and LOQ values were calculated from the calibration curves at eight concentrations. These were determined as 3.3 and 10 σ/S, respectively, where σ refers to the standard deviation of the intercept and S refers to the slope of the regression line determined from the calibration curve. A selectivity was verified by analyzing the chromatograms of standard solution and the three representative samples (jam, cheese, soy sauce) spiked with the stock solution to examine matrix interferences. The accuracy and precision were evaluated by analyzing the three representative samples (jam, cheese, soy sauce) spiked with the stock solution to obtain the final concentrations of 20, 40 and 100 mg/kg. The intra-day (three repetitions on the same day) and inter-day (three repetitions over three different days) precision tests were performed. The measurement uncertainty of the concentration of DHA, SA, BA, MP, EP in processed foods was estimated according to the GUM and EURACHEM guide (ISO, 1993;EURACHEM, 2000). The uncertainty factors were measured based on the four components: The calibration curve, foodstuffs matrix (repeatability), preparation of standards and sample preparation. The concentration of preservatives, respectively at 40 mg/kg, was selected for calculating the uncertainty. The sample repeatability was determined by analyzing jam for 6 times. The measurement uncertainty (U), the expanded uncertainty, was calculated by multiplying the combined standard uncertainty by a coverage factor (k = 2) that yields a confidence level of approximately 95%.

Inter laboratory comparison: International proficiency test
To demonstrate the applicability of this study, the international comparative proficiency evaluations were conducted by participating in the Food Analysis Performance Assessment Scheme (FAPAS, Fera Science Ltd., York, UK), a proficiency program operated by the Department for Environment, Food and Rural Affairs in UK. FAPAS analyses the Z score (range of error between laboratories) as an international proficiency test for the 10 fields such as food nutrients, food additives, and pesticide residues, and presents the results. The outcome was evaluated as Z-score, and it was statistically satisfactory if z|≤ 2 z = x − x a σ p , where x is the participant's reported result, x a is the assigned value and σ P is the standard deviation for proficiency.

Optimization of HPLC conditions
UV scan (200-400 nm) was applied to the analyte in order to minimize the interfering compounds and to evaluate the selectivity of the method. Regarding the detection wavelength, there was a difference in the maximum absorption of the analyte. Korea's Food Code recommended λ = 217 nm for five kinds of preservatives. Nonethless, as a result of the wavelength test, a single wavelength of λ = 235 nm was found to be the best compromise for detecting five preservatives. Compared to 217 nm, the sensitivity of the parabens (Methylparaben, Ethylparaben) decreased slightly at 235 nm, but the sensitivity of sorbic acid was significantly improved (data not shown). A UV scan of the analyte displayed no noticeable differences between the standard solution and the UV full scan (200-400 nm) of the preservative peaks in the processed samples. Five preservatives were well separated with the clear peak by chromatographic conditions. Therefore, this method is proven to have a high selectivity when the actual matrix is applied (Fig. 1).

Established method
Above all, for the determination of the extraction solvent, the recovery rate test was conducted with four extraction solvents (acetonitrile, acetone, ethanol, methanol) and the resultants were compared. As a result of the solvent test, there was a no significant difference between the solvent recovery rate. However, though a small difference, ethanol showed superiority, and it was selected as the extraction solvent.
As a consequence of comparing the purification process in the three representative foods (jam, cheese, soy sauce), the carrez reagent method generally showed better results in terms of recovery rate and selectivity (Table 1). Although the solid-phase extraction method seemed to be good at removing the processed food matrix, but it seems that there was a loss in recovery during the process of pass through the cartridge and nitrogen concentration step. It is also considered that large variations occurred in these experimental steps. Also SPE cartridge (C18, HLB) approach took longer time due to the time-consuming loading using the pump and nitrogen concentration. Meanwhile the process with carrez reagent had high recovery rates, superior accuracy, variation and it took much less time. Thus, using carrez reagent method was determined as suitable for the purification process. In the case of the ethanol extraction method, since the centrifuged supernatant is analyzed after simple extraction, fat and protein are not sufficiently removed, and in particular, the emulsification properties of emulsified foods such as cheese and butter cannot be destroyed, so the recovery rate is low.
Carrez solutions are widely used in the purification of proteins and fats, breaking up emulsions that can interfere with the crystallization of small molecules such as carbohydrates (Chávez-Serví et al., 2004;Manzi and Pizzoferrato, 2013;Marconi et al., 2004;Pereda et al., 2009). Therefore, carrez clarification process is optimal for protein precipitation and turbidity removal (Oliveira-Neves et al., 2018). Consequently, the purification process using carrez reagent is considered to be good for various types of food analysis.

Method validation
The analytical methods showed satisfactory linearity, LOD, LOQ, accuracy and precision. The calibration curves were found to be linear for analytes. The correlation coefficients (r 2 ) of the preservatives displayed over 0.9998. The LODs and LOQs for preservatives are shown in Table 2. The accuracy (recovery) and precision of the method were evaluated using spiked blank samples of jam, cheese, soy sauce at three concentrations, 20, 40 and 100 mg/kg, respectively. The validation results are summarized in Table 2. The recovery of the preservatives is within the range of 87.7-110%  for each sample. The intra-day and inter-day precisions are expressed as RSDs and the precisions are in the range of 0.07-7.72%. These recoveries and precisions are within the criteria established by the AOAC guideline (AOAC, 2012b). The measurement uncertainty of the concentration of analytes in the processed foods, expressed as mg/kg, was calculated from the expanded standard uncertainty, with a coverage factor of k = 2, which indicates approximate 95% confidence. The total relative uncertainty was expressed based on the combined relative uncertainty of four factors: Calibration curve, sample matrix (repeatability), preparation of standards and sample preparation. The standard uncertainty was calculated by combining the relative uncertainty and the concentration of the analyzed jam (40 mg/kg). The expanded uncertainty was calculated based on the combined standard uncertainty. The final analyzed concentrations of DA, SA, BA, MP, EP in jam were 44.7 ± 2.2, 44.0 ± 2.1, 44.4 ± 2.0, 44.3 ± 2.1 and 43.4 ± 2.2 mg/kg, respectively. The results for each factor of measurement uncertainty are summarized in a Table 3.

International proficiency test results
The analyzed sample for the first international comparative proficiency evaluation was selected as soft drink (soft drink, 03160) and the z-score was 0.5 for sorbic acid and −0.6 for benzoic acid. The second analysis sample was jam (Jam, 20171), which was evaluated as −0.2 sorbic acid and 0.0 benzoic acid. The z-score (range of error between laboratories) ≤ 2 was considered statistically satisfactory. These results were to confirm the accuracy of the experimental method used in this study. Table 4 represents the result of monitoring to verify the applicability of the established method with carrez reagent. Figure 2 shows the distribution of each component for the monitoring results. Among the collected products, 2 out of 10 butter and 6 out of 10 margarine products were found to  be combined with DA. DA was detected in all the six cases of the labeling, and the range was 32-144 mg/kg, meeting the regulations of 500 mg/kg (Korean standard). As SA is the most widely used preservative, it is used in various ways, and it was found in diverse items such as cheese, fish cake, dried fish, pickles, sauces, syrup, jam, bread, and alcoholic beverages. A total of 114 items with SA on the labeling were collected: Processed cheese 10 cases, natural cheese 1 case, dried fish paste 8 cases, jam 1 case, sugar processed products 11 cases, peanut butter 2 cases, bread 3 cases, sauces 10 cases, fish meat products 15 cases, pickled food 19 cases, stewed food 1 case, salted fish 4 cases, alcoholic beverages 16 cases and carbonated drinks 4 cases. Among the labeled products, eight pickled foods, one dried fish, five sauces, one bread and nine alcoholic beverages were not detected as the components were less than the LOQ. SA showed the lowest detected amount of 7 mg/kg in stewed foods, and the highest detected amount of 2532 mg/kg in natural cheese. All of them were suitable for regulations for use, except bread as it was included in the product group without standard for use. Nonethless, due to margarine, the raw material, and it was judged to be suitable (the use standard did not exceed 2000 mg/kg).

Monitoring results of preservatives in domestic processed foods
As the product labeled with BA, 1 fruit and vegetable juice, 3 stewed foods, 4 pickled foods, and 12 carbonated beverages were collected. BA was found to be particularly widely used in the beverages. Among the labeled products, 3 stewed foods and 1 pickled food were classified as not detected, as they were below LOQ. The standard for use of BA was 600 mg/kg in fruit and vegetable juice and soft drink, and 1000 mg/kg in pickled foods.
Among the processed foods in the market, the products containing MP were not collected and detected in all foods.
As the product labeled with EP, a total of 40 items were collected: 7 soy sauce, 26 sauces, 5 pickled foods, and 2 stewed foods. It was found that they are relatively widely used in the sauces. The standard for use of EP is 250 mg/ kg in soy sauce and 200 mg/kg in sauces, pickled foods, and stewed foods. Except for the seven cases detecting the value below the LOQ (three cases from sauces, two cases from stewed foods, and two cases of pickled foods), all were suitable for the regulations.
In this study, sorbic acid was detected at 2532 mg/kg in natural cheese, which was within the standard, but at a very high level, followed by 1363.3 mg/kg in fish products, 1238 mg/kg in processed cheese, 801.1 mg/kg in syrup, and 791.3 mg/kg in pickles. According to Lee et al. (2013) previous study, processed cheese was detected the highest at 761.7 mg/kg, processed fish meat products 712 mg/kg, dried stored meat 585.3 mg/kg, salted seafood 492.4 mg/kg, and sausage 369.5 mg/kg. However, there were some differences in other items such as salted fish and sausages. In this study, benzoic acid was detected at 206.5 mg/kg in most beverages    Lee et al. (2013)reported benzoic acid was detected at 86.5 mg/kg in beverages and 13.6 mg/kg in pickles, showing a big difference, but like this study, it was not detected in jams. According to Shin et al. (2017), the intake risk is evaluated by the estimated daily intake (EDI), and the main causes of EDI for benzoic acid are beverages, and the main causes of EDI for sorbic acid are soybean paste, processed fish products, and pickles. In addition, Yoon et al. (2001) also reported that the exposure of benzoic acid is mostly due to beverages, so regular monitoring of beverages is necessary.
In this study, as a result of randomly collecting and examining processed foods sold in the Korean market, sorbic acid was not detected in all soybean paste, but was detected in 7 out of 25 cases in processed fish products, reaching ND-876 mg/kg and ND-791 mg/kg in 19 cases out of 26 in pickles. It was confirmed that sorbic acid is commonly used in pickles. The detection amount was all within the standard and was at a safe level. Benzoic acid was detected in beverages in 12 out of 22 cases, and the detected amount was ND-336 mg/kg, which was also far below the standard, confirming that it was a safe level. By conducting monitoring using such a quick and accurate method, it can be used as research data for the safe use of food additives through continuous research.

Conclusions
In this study, an analytical method for the simultaneous determination of preservatives (dehydroacetic acid, benzoic acid, sorbic acid, methylparaben, ethylparaben) in different kinds of foodstuffs by HPLC-DAD was utilized. The chromatography coupled to conductivity detection was developed, validated and applied for the analysis of the commercial samples. The usage of carrez reagent, an excellent range of 41.5-244 mg/kg. D As for the detection distribution of sorbic acid, it showed a wide range from 312 to 1,363 mg/kg for the processed fish products and 86.6 to 791 for the pickled food products material for removing interfering substances, was found to be extremely suitable for this study. The pre-treatment method using carrez reagent showed high recovery, timeconsuming, rapid and easy. Currently, the Korea Food Code recommends the steam distillation method and the solvent extraction method limited to livestock products. The stream distillation methods were used widely among the analytical laboratories (AOAC, 2012a), but impossible to process at the same time, so time-consuming and costly in solvents (Techakriengkrai and Surakarnkul, 2007). Analysis of the centrifuged supernatant after just solvent extraction for processed food analysis does not sufficiently remove fat and protein. In particular, there is a problem that the recovery rate is low because it cannot destroy the emulsification properties of emulsified foods such as cheese and butter. Through this study, the test method that expands the range of applicable foods will be revised by adding a purification process of carrez reagent to the solvent extraction method. In addition, this study result will help to improve the preservative test method and to establish a database in the near future. Therefore, building a database through continuous monitoring of raw materials and expanding the scope of application of experimental methods will be great help in providing a rapid and efficient preservative analysis method.

Conflict of interest
On behalf of all authors, the corresponding author states that there is no conflict of interest. None of the authors has any financial interest or conflict with industries or parties.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.