Simultaneous Quantification of 16 Bisphenol Analogues in Food Matrices

1. Introduction

Bisphenol A (BPA) and its analogues are phenol-based chemicals broadly used in plastic industries [1,2], with a global production estimated at 5.5 billion tons in 2021 [3]. BPA is one of the most commonly produced compounds among bisphenols (BPs). It is widely used as raw material for the synthesis of polycarbonates and epoxy resins, or as an additive (such as an antioxidant or a stabilizing agent) to improve the properties of plastic materials [1,2]. Many products used in daily life contain BPA, including water pipes, food containers, medical equipment, toys or electronics, thus globally it has been demonstrated that populations have been chronically exposed to BPA via different pathways, including oral, dermal and hand-to-mouth transfer, as well as other mechanisms [4,5]. However, because of its adverse impact on human health and the environment, the contamination of commercial products with BPA has recently attracted tremendous attention. BPA is considered as an endocrine-disrupting chemical (EDC), responsible for impeding the function of estrogenic and androgenic hormones [6,7,8,9,10,11,12]. In addition, it is a highly polluting substance, mainly generated by effluents from the plastic industry and is commonly found among the contaminants identified in soils and waters [13,14]. The use of BPA has therefore been limited or prohibited in many countries [15,16] leading to an increasing demand by the plastic industries for bisphenol analogues to replace BPA.

Several bisphenol analogues have been produced to replace BPA in research and industry (Figure 1). Bisphenols S, F, and AF (BPS, BPF, BPAF), which display similar chemical structure in comparison to BPA, are the most used substituents in industry [17,18,19,20]. They exhibit similar stability and thermoplastic properties, with however different reactivities and processabilities, which sometimes hinder their efficient use in the manufacture of polymers. Although studies about the toxicological behaviour of the bisphenol analogues are limited, several reports demonstrate a wide variety of different toxicological mechanisms, including endocrine disruptive effects, cytotoxicity, genotoxicity or neurotoxic effects [17,18,19,20]. Despite the potential risks to human health, no restrictions have been implemented for most of the BPs on the market [21]. Therefore, the development of rapid and low-cost techniques for the determination of BPA analogues with high sensitivity are urgently needed. The detection and quantification of bisphenols in food matrices have already been described in the literature and is usually performed using liquid chromatography or gas chromatography, coupled to mass spectrometry (LC-MS and GC-MS) or tandem-mass spectrometry (LC-MS/MS and GC-MS/MS) [22]. However, the reported methodologies are usually limited to specific bisphenol analogues and cannot be applied to a broad range of BPs, furthermore, only a limited number of studies investigating the presence of bisphenols in food and beverages are reported. Other studies have employed aptasensors to detect BPA in food matrices and canned foods [23,24,25,26,27]. However, long turnaround times, expensive equipment and laborious processing prevent their widespread diagnostic use. Therefore, the development of new analytical methods allowing for the detection of a high number of different bisphenols in food is needed.

In this work, we developed and compared a new extraction and analytical method for selective and efficient detection of bisphenol analogues in canned food and beverages. A QuEChERS (Quick, Easy, Cheap, Effective, Rugged, and Safe) technique was used in combination with GC-MS for the simultaneous quantification of 16 bisphenols analogues (bisphenol A (BPA), bisphenol AF (BPAF), bisphenol AP (BPAP), bisphenol B (BPB), bisphenol BP (BPBP), bisphenol C (BPC), bisphenol E (BPE), bisphenol F (BPF), bisphenol G (BPG), bisphenol M (BPM), bisphenol PH (BPPH), bisphenol P (BPP), bisphenol S (BPS), bisphenol TMC (BPTMC), bisphenol Z (BPZ) and tetramethyl bisphenol F (TMBPF)), in various canned foods and beverages. Bisphenol analogues can also be found in non-canned foods and pose a clear risk to human health. However, due to the greater concentrations of BPs in canned foods [28], it is imperative to have a rapid detection method for several analogues simultaneously so that daily intake can be monitored and controlled. Therefore, detection of BPs in canned food remains an area of great interest. Several studies have highlighted the toxicity profiles of bisphenol analogues as well as their interactions with other compounds that can lead to an increased bioavailability and uptake of BPA to cells [29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44]. Yet, despite the established and growing evidence of their harmful effects to human health, BPA and its analogues are not prohibited with the exception of their use in cosmetic substances [45] and in plastic infant feeding bottles [46]. In recent years, European Union regulations have focused on detecting the limits for BPA products across a range of sectors [47,48,49,50,51,52,53]. However, even if the migration of BPA and its analogues are restricted within the 50 μg/kg limit in canned foods, smaller concentrations of these chemicals are still ingested and detectable in the human body. This demonstrates the need for novel analytical methods for the simultaneous analysis of a wide variety of bisphenols.

2. Materials and Methods

2.1. Chemicals

Bisphenol A (≥ 99%), bisphenol AF (≥ 99%), bisphenol AP (≥ 99%), bisphenol B (≥ 98%), bisphenol BP (≥ 98%), bisphenol C (≥ 98%), bisphenol E (≥ 98%), bisphenol F (≥ 98%), bisphenol G (≥ 98%), bisphenol M (≥ 99%), bisphenol P (≥ 99%), bisphenol PH (≥ 99%), bisphenol S (≥ 98%), bisphenol TMC (≥ 97%), and bisphenol Z (≥ 99%) were purchased from Neochema Gmbh pre-dissolved in acetonitrile and in a concentration of 100 ppm (µg/mL) (Stock solution 1: standards (STDs) mix 16 Bisphenols). This stock standard mixture was stored at -20°C. Tetramethyl bisphenol F (≥ 99%) standard was purchased from Sigma Aldrich Chemie GmbH. The stock solution was made by weighing 10 mg of standard and by dissolving it in 10 mL of methanol. The internal standard (ISTD), bisphenol A d-16 was purchased from Neochema Gmbh pre-dissolved in acetonitrile with a concentration of 100 ppm (µg/mL). Working solutions for calibrations were prepared by dilution of the stock standard mixture. The internal standard working solution was prepared separately and by dilution of the stock standard with acetonitrile (Sigma-Aldrich).

A solution of Bis(trimethylsilyl)trifluoroacetamide (BSTFA) was used for BPs derivatization and was purchased from Sigma Aldrich. Magnesium sulfate anhydrous (≥ 99.5%), sodium chloride, dichloromethane (≥ 99.8%), methanol (≥ 99.9%) and acetonitrile (≥ 99.9%), were also purchased from Sigma Aldrich. Nanopure water was provided by an ultrapure water system (ariumPro, Sartorius, Germany). The solid phase extraction method was carried out with a CHROMABOND^® HLB cartridge (3 mL, 200 mg which was purchased from Macherey-Nagel). The QuEChERS clean-up was performed with CHROMABOND^® Mix XX (1.20 g MgSO₄, 0.40 g CHROMABOND^® Diamino), which were purchased from Macherey-Nagel.

2.2. Samples and Sample Preparation

A total of 22 samples of canned food and beverages which were all readily available in Lausanne and the Swiss market were analyzed. All samples were stored at room temperature prior to preparation and returned to the refrigerator/freezer once preparation was completed. Food samples comprised of canned pineapple (n=2), canned peaches (n=1), canned ravioli (n=5), farce vol-au-vent (n=2), soup (n=2), fruit puree (n=5), canned tuna (n=1), cola light (n=1), lemon (n=1) and beer (n=2). Canned food samples were homogenized with an electric blender. In the case of the canned fruit sample the solid and the liquid part were analyzed discretely to study the migration effects of bisphenols. 10 g of each homogenized or liquid sample were added to a 50 mL tube followed by 100 µL of ISTD 1 ppm. The solid sample and any samples that were in the liquid phase once homogenized, had 5 mL of Evian water added to facilitate liquid-liquid extraction.

2.3. Extraction Method

10 mL of acetonitrile was added to each of the samples and followed by further homogenization with the vortex. Magnesium sulfate anhydrous (MgSO₄) and sodium chloride (NaCl) were added (4 g and 1 g respectively) to perform salting out of the liquid-liquid extraction phase of the QuEChERS method. The sample was then shaken either by hand or the vortex for 1 minute. The samples were then added to a centrifuge at 1000 rpm for 15 minutes. An amount of 5 mL of supernatant in the organic phase was collected and transferred into the CHROMABOND^® Mix XX and further shaken either by hand or vortex. The sample was then filtered using PTFE, 0.45 µm filter into a 40 mL glass tube. The filtered solution was then evaporated under N₂ flow.

2.4. Derivatization

Derivatization of BPs was performed with bis(trimethylsilyl)trifluoroacetamide (BSTFA) at a temperature of 60°C for 45 minutes. Four different volumes of BSTFA were tested: 20, 50, 70 and 100 μL. For the derivatization tests 500 μl of STD mix at 1 ppm were put in a vial and evaporated. The different volumes of BSTFA were added and for each volume acetonitrile was added to reach a total volume of 100 µL. Following derivatization and cooling at room temperature, the samples were subject to GC-MS analysis.

2.5. GC-MS Parameters

Following extraction using the QuEChERS method, the samples were subjected to gas chromatography-mass spectrometry model GCMS-QP 2010 Ultra (Shimadzu Corporation, Japan) in EI mode and utilizing LabSolutions software. The instrument was equipped with an OPTIMA-5 MS column with a diameter of 25 mm, length of 30 m, film thickness 0.25 µm. Helium was utilized as the carrier gas in constant pressure of 58.4 kPa with a flow rate at an initial temperature of 5.28 mL/min, a total flow of 24. 4mL/min and a column flow of 1.09 mL/min. The ion source temperature was adjusted to 250°C while the interface temperature was 280°C. A selected ion monitoring (SIM) mode was then used to quantify the analytes in the samples and the standard mixture of the calibration curve as shown in Table 1.

2.6. Calibration Curve and Controls for GC-MS

The calibration curve samples were prepared in water and spiked with 100 µL of ISTD, 1 ppm and with various volumes of a standard mixture of 16 BPs, 1 ppm and standard Mix 16 BPs, 100 ppb to produce concentrations: 50, 10, 5, 1, 0.5 and 0.1 µg/L (ppb). Blank samples were prepared in the same way as the calibration curve with 100 µL of ISTD, 1 ppm. The calibration curve and the blank followed the same preparation procedure as the samples. Each standard sample was prepared in duplicate and analyzed. Recovery tests (i.e., samples prepared by adding a known amount of standards and extracted with the same method described above) and triplicate measurements were conducted to study the efficiency and the reliability of the extraction method. The limit of detection (LOD) and limit of quantification (LOQ) were calculated for each bisphenol analyzed according to equations 1 and 2 respectively.

$$LOD=3\frac{S}{b}$$

(1)

$$LOQ=10\frac{S}{b}$$

(2)

where s is the residual standard deviation and b is the slope of the calibration curve.

2.7. LC-MS/MS Parameters

Similarly, to the QuEChERS GC-MS (SIM) method, the sample preparation and extraction, and the evaporation and recovery of the extract were carried out under the same conditions for QuEChERS LC-MS/MS. No sample derivatization was performed, but after evaporation the crude was resuspended in 1 mL of water and directly analysed in LC-MS/MS.

Chromatographic analysis was performed using a Shimadzu LCMS-8060. This was coupled to a triple quadrupole and equipped with three solvent modules; Nexera X2 LC-30AD, an autosampler Nexera X2 SIL-30AC, a column over unit CTO-20AC, two degassing units GDGU-205R and a valve unit FCV-20AH₂. All instruments were Shimadzu, Japan models. The LC separation was conducted on an Agilent Poroshell 120 EC-C18 treated column (Agilent technologies; 2.7 μm, 4.6 mm × 50 mm). The oven temperature was set and maintained at 40°C, with a temperature limit of 90°C. The injection volume was 20 μL. The flow rate was 0.25 mL/min and the total data acquisition was 30 min. The mobile phase consisted of a mixture of water Nanopure 5 mM ammonium acetate (solvent A) and methanol (solvent B). The analysis was conducted with a gradient mode where the organic mobile phase, solvent B, increased linearly. The time and percentages were as follows: initially 25% for 3 min, then increased to 100% over 3 to 20 min and maintained at 100% during 20 to 24 min; at 24 min the eluent was restored to the initial conditions for 6 min to equilibrate the column for the next injection. The pressure limit was 0 to 1000 bar. The mass spectrometry analysis was carried out on a triple quadrupole with an electrospray (ESI) source operating in negative mode. The interface temperature was set at 400°C and Argon was utilized as carrier gas. The nebulizing gas flow was 2 L/min, the heating gas flow 10 L/min while the drying gas flow was 10 L/min. The desolvation temperature was 650°C. MS data were acquired in 100 –1000 m/z range. Data acquisitions were performed using LabSolutions LCX3-TQ8060 while data processing was performed with LabSolutions Insight. All parameters related to the analytes analyzed are shown in Table 2.

2.8. Calibration Curve and Controls for LC-MS/MS

The calibration curve samples were prepared in the same manner to the QuEChERS GC-MS (SIM) Method of Extraction to produce concentrations: 50, 10, 5, 1, 0.5 and 0.1 µg/L (ppb). Blanks, recoveries and triplicate were also performed as described in section 2.6.

2.9. Statistical Analysis

The statistical treatment of the data set obtained from the analysis follows the “Guidelines for performance criteria and validation procedures of analytical methods used in controls of food contact materials - EUR 24105 EN (2009)” [54]. All the statistical values were calculated with the regression analysis output calculated with Excel software.

3. Results and Discussion

3.1. Optimization of the derivatization

The chromatographic response of bisphenol analogues was determined through derivatization due to their high polarity and non-volatile nature using silylation and performed with N, O-bis(trimethylsilyl)trifluoroacetamide (BSTFA; IUPAC: trimethylsilyl 2,2,2-trifluoro-N-trimethylsilylethanimidate) [55]. The results of the derivatization tests (Figure 2) showed that the best results are linked to the use of 50 mL of BSTFA/50 mL of acetonitrile and 70 mL of BSTFA/30 mL of acetonitrile. For this reason, 50 mL of BSTFA and 50 μL of acetonitrile were chosen for the derivatization of the samples.

3.2. Statystical validation of the extraction and analytical method

The developed method employs an internal standard (BPA-d16), which enables to account for variation in the response of the chromatographic system, the exact volume of sample injected, and all losses during the extraction method. Calibration curves were then constructed using the ratio of C_BPs/C_BPA-d16. Table 3 shows the details of the extraction method demonstrating good correlation coefficient of determination (R²) and instrumental trueness (C_found/C_nominal x 100), and low LOQ and LOD for each bisphenol. Moreover, the linearity was evaluated with two different methods: residues for different calibration levels and adequateness of the linearity model. By residues calculation, there were obtained results that confirm the linearity of the method for every analyte. The data produced by the adequateness of the linearity model results confirm the predictive ability of each calibration curve [54].

The selectivity of the method was evaluated by the analysis of different blanks, showing high selectivity for each analyte and no interferences. Moreover, at least one blank was analyzed every time the analysis was performed.

A short-term repeatability test was conducted and statistically treated following the UNI ISO 5725-1. The repeatability of the test was confirmed using the Horwitz equation [56].

Evaluation of the recovery of the QuEChERS extraction methods was performed on every sample for a standard concentration of 10 ppb. Table 4 shows the results of the recovery for each type of sample obtained as geometric mean of the recoveries of the single samples analyzed.

For the range of concentration analyzed, the UNI CEI EN ISO/IEC 17025: 2005 indicates a different acceptable mean recovery (%) range. For lower concentrations, this range increases: in the order of 10^-19 (10 µg/kg) the range of mean recovery is 60-115% [57]. As reported in Table 4, in the fruit, vegetable, beverage, and canned foods matrices, all recoveries were within acceptable limits. The complexities of the matrices within canned food however presented criticalities especially in the case of BPS, where some recoveries were outside of the acceptable range. An exception was found in the tuna matrix with BPAP, BPM, BPP, BPBP and BPPH where recoveries were in the range of 25% - 46%.

3.3. Quantitative results

In only 2 samples (canned pineapple and fruit puree) of the 22 analyzed no bisphenols were detected. Of the sixteen possible analytes analyzed, the samples were positive for eight of the bisphenols: BPAF, BPF, BPE, BPA, BPB, BPS, BPM, and TMBPF (Table 5). The most frequently detected analytes and indeed those found at higher concentrations were BPA and BPS. BPA was detected in 78% of all cases in concentrations between 3.21-40.65 µg/kg while BPS was found in 48% of cases in the concentration range of 5.58-11.11 µg/kg.

In the vegetable soup and canned fruit (Table 5, A) low concentrations of BPAF, BPF, BPE, BPA and BPB were found with 3.21 µg/kg of BPA being detected. In three of the five fruit puree samples analyzed (Table 5, B), a high concentration of BPS was found along with lower concentrations of BPE and BPB. In one of the samples BPA was detected at 12.72 µg/kg.

In the complex food matrices such as ravioli (Table 5, C), all samples contained similar concentrations of BPAF, BPA, BPB and BPS with a similar pattern being shown with the vol-au-vents. With the tuna sample, analysis revealed the presence of BPE, BPA, BPS and BPM.

In beverages (Table 5, D), only BPA, TMBPF and BPS were detected. BPA and BPS were found at concentrations below the LOQ. Only TMBPF was found at concentrations above the LOQ in the two beer samples.

In order to test the robustness of the QuEChERS extraction method, a confirming test was conducted changing the instrument and method of analysis and involving the use of the LC-MS/MS. The samples that showed higher positive concentrations, especially of BPA, were the ones re-analyzed with the LC- MS/MS. Two types of different ravioli and the farce for vol-au-vent were subjected to analysis with GC-MS and LC-MS/MS. The goal is to compare the results obtained with the GC-MS analysis and specifically the BPA concentration in the samples that resulted positive. Samples which showed higher positive concentrations of BPs, especially BPA, were reanalyzed using this method. Ravioli and vol-au-vents were subjected to analysis with both GC-MS and LC-MS/MS with the aim of comparing results (Table 6).

The data obtained from LC-MS/MS for the ravioli was comparable to that found using GC-MS. However, for the vol-au-vents BPA concentration deviated with a concentration of 27.35 µg/kg using LC-MS/MS and 37.47 µg/kg with GC/MS analysis.

The small difference between the values obtained for GC-MS and LC-MS/MS confirms that the use of GC after QuEChERS extraction is also a valid method of analysis of bisphenols in complex matrices.

When compared with the migration limit set out by the EU and Switzerland for food in contact with coated cans, all samples were in accordance with the regulations with BPs present at concentrations <50 µg/kg. Results for the concentrations of BPA found in canned pineapple concurred with results of a similar study conducted by Cunha et al. with concentration ranges between 4 and 10 μg/kg [58].

The fruit puree samples were not collected from a can yet, high concentrations of BPs, in particular BPS were found in the range of 6.65-11.11 µg/kg with one sample revealing BPA at 12.72 µg/kg. Migration from the epoxy resin was not possible in these samples lending weight to the theory of contamination during the food production chain [59].

BPA was also found in ravioli samples in the range of 21.67-36.80 µg/kg and from 37.47-40.65 µg/kg in the vol-au-vents. A 2010 study focused on the presence of BPA in the Belgium market, finding a concentration of 73.1 µg/kg for ravioli and 29.3µg/kg for cream of chicken soup, [60] a sample which is chemically comparable to the vol-au-vents analyzed. A concentration of 10.82 µg/kg of BPA was found in the tuna which was comparable with a study of both canned and non-canned tuna which has concentrations in the range of 1.0-99.9 µg/kg [61]. Based on the results obtained, it appears that the presence of fat or a matrix increased the migration and retention of bisphenols inside foods. Interestingly, a lower concentration of bisphenols was found in canned fruit and vegetable soup rather than in food matrices such as ravioli, vol-au-vent and tuna which could be linked to the high-octanol-water partition coefficient presented by most of the bisphenol analogues.

4. Conclusions

The aim of this study was to develop an extraction and analytical method which was suitable for the detection of bisphenol analogues in complex food matrices. We have successfully demonstrated the acceptability of QuEChERS-GC-MS, also compared to QuEChERS LC-MS/MS. An important point to note is the number of bisphenol analogues included in this study. Sixteen in total were investigated with the method developed allowing for simultaneous detection of the concentration of these compounds in canned foods. The results obtained from the samples highlighted the presence of a variety of bisphenol analogues in canned foods, not only BPA. The BPS, BPAF, BPE, BPF and even TMBPF found in canned foods may be due to the replacement of BPA with these analogues. However, they may also be present due to contamination of the food production chain.

Contamination can be produced during any stage of the production food chain and often the same industries that output BPA products also make use of other bisphenols. More detailed studies are needed to investigate the presence of bisphenol analogues in the canned food industry, not only in the final stages of production of the finished product but throughout the food production chain with a focus on European and Swiss markets. This will provide a more complete picture of exposure to these compounds and may provide a boost to toxicity studies on this still understudied class of compounds.