Comparison between Different Extraction Methods for Determination of Primary Aromatic Amines in Food Simulant

The primary aromatic amines (PAAs) are food contaminants which may exist in packaged food. Polyurethane (PU) adhesives which are used in flexible packaging are the main source of PAAs. It is the unreacted diisocyanates which in fact migrate to foodstuff and then hydrolyze to PAAs. These PAAs include toluenediamines (TDAs) and methylenedianilines (MDAs), and the selected PAAs were 2,4-TDA, 2,6-TDA, 4,4′-MDA, 2,4′-MDA, and 2,2′-MDA. PAAs have genotoxic, carcinogenic, and allergenic effects. In this study, extraction methods were applied on a 3% acetic acid as food simulant which was spiked with the PAAs under study. Extraction methods were liquid-liquid extraction (LLE), dispersive liquid-liquid microextraction (DLLME), and solid-phase extraction (SPE) with C18 ec (octadecyl), HR-P (styrene/divinylbenzene), and SCX (strong cationic exchange) cartridges. Extracted samples were detected and analyzed by HPLC-UV. In comparison between methods, recovery rate of SCX cartridge showed the best adsorption, up to 91% for polar PAAs (TDAs and MDAs). The interested PAAs are polar and relatively soluble in water, so a cartridge with cationic exchange properties has the best absorption and consequently the best recoveries.


Introduction
Primary aromatic amines (PAAs) are food contaminants. Migration of PAAs into food is through colored plastics, printed paper, cooking utensils, and flexible packaging. One of the main sources of PAAs is polyurethane (PU) adhesives which are used extensively in lamination of multilayer films. PU adhesives might contain unreacted aromatic diisocyanates coming from the imperfect polymerization process of polyurethane; after packaging, water present in foods hydrolyzes residual aromatic diisocyanates, thus leading to PAA formation [1].
Toluene diisocyanate (TDI) and methylene diphenyl isocyanate (MDI) are used in the production of PU adhesives.
PAAs have genotoxic, carcinogenic, and allergenic effects. Epidemiological studies of the 4,4′-MDA indicate a risk of bladder cancer in humans, and the 2,4-TDA and 4,4′-MDA are listed as possible human carcinogens [3].
In the European Union (EU) Regulation (10/2011), plastic material and articles shall not release PAAs in detectable quantity (DL � 0.01 mg/kg) in food or food simulants. Chromatography method is recommended for individual identification and quantification of migrated PAAs [4]. e extraction/preconcentration of PAAs is required in order to reach the detection limit, prior to analysis. Sample preparation is commonly done for LLE or SPE method [5]. Each one of these two general approaches has advantages and disadvantages; also there has been a new development in the two methods in order to save time, labor, chemicals, and other materials used. is process has led to the creation of new methods such as solid-phase microextraction (SPME), liquid-phase microextraction (LPME), and dispersive liquidliquid microextraction (DLLME) [6]. DLLME technique has been developed and often used for the determination of organic compound in water samples. DLLME is a very simple and rapid extraction method which is used for preconcentration of organic and inorganic compounds from aqueous samples. is method is based on the fast injection of the appropriate mixture of extraction (high-density solvent) and dispersive solvents (polar) into the aqueous solution to form a cloudy ternary component solvent. Extraction process is being completed with centrifugation step, and the enriched analyte is collected in the shape of droplets in the sediment phase [7][8][9].
Several methods are provided in the literatures for determination of TDIs and MDIs or TDAs and MDAs in the urine, blood, PU foam, water, and waste water samples. ese methods are often used with or without hydrolysis of samples, extraction/preconcentration in case of liquid-liquid extraction (LLE) or solid-phase extraction (SPE), and with or without derivatization for GC or HPLC analysis. e selection of extraction methods depends on the type and condition of the sample [10][11][12][13][14].
In this study, two different methods of liquid-phase extraction (LLE, DLLME) and solid-phase extraction (Chromabond C18 ec, Chromabond HR-P, and SCX) were investigated for the extraction of interested PAAs, which were spiked in a 3% acetic acid as food simulant [15]. Use of 3% acetic acid which is accepted by the FDA offers the best condition for simulation and a worst-case scenario for this study, and also it is easy to evaluate [16,17]. e selection of aromatic amines was based on the type of diisocyanates used to produce PU adhesives. e polyurethane adhesives are often produced from toluene diisocyanates (2,4-TDI and 2,6-TDI) or methylene diphenyl isocyanate (4,4′-MDI); thus, the PAAs presented are 2,4-TDA, 2,6-TDA, 4,4′-MDA, 2,4′-MDA, and 2,2′-MDA, and the relative recoveries of the PAAs of interest are determined in the 3% acetic acid matrix.

Standards and Reagents.
e reagents used in this work were all of analytical grade with no further purification. All the reagents were obtained from Merck KGaA, Darmstadt, Germany.

Apparatus and Conditions.
HPLC system was an Agilent 1200, equipped with a quaternary pump (G1311A), a column thermostat (G1316A), a degasser unit (G1322A) an autosampler (G1329), and a diode array detector (G1315D). e HPLC system was controlled, and data were analyzed by a computer equipped with LC software (Agilent Chem Station). e chromatographic conditions were as follows: HPLC analysis was performed with methanol (solvent A) and an ammonium acetate buffer 10 mM (solvent B) as mobile phase. Separation with solvent programming was accomplished on a Phenomenex C18 column (250 × 4.6 mm i.d., 5 μm particle sizes); flow rate was 1 ml·min −1 , detection wavelength was 235 nm, and temperature was adjusted to 25°C. e gradient elution process was performed as follows in order to achieve the optimum separation: solvent A was kept at 15% and solvent B was kept at 85% for 2 min, solvent A percentage was increased to 65% along a linear gradient curve for 24 min, and then, solvent A concentration was again increased from 65% to 100% along a linear gradient curve for further 2 min and kept steady at 100% for another 2 min and then was decreased to 15% in the same manner in 1 min and continued for a further 2 min, giving the total run time of 30 min. e sample solutions were prepared from standard mix solutions 10 mg·l −1 with three concentrations 7.5, 15, and 30 ng·ml −1 in 3% acetic acid (w/v).

Liquid-Liquid Extraction Method.
For the extraction of PAAs by solvents, the optimized method was applied [12,13] and modified as follows: an aliquot sample solution of 8 ml was transferred to a test tube, the pH was adjusted to above 10 with sodium hydroxide, and the solution was saturated with sodium chloride. e sample was extracted three times with 5 mL portion of mixed dichloromethane/petroleum benzene (3 :1) and followed by twice with 5 mL of mixed dichloromethane/methyl tert-butyl ether (1 :1). e organic solvent fractions were combined and evaporated to dryness by a stream of nitrogen gas. Finally, the extract was dissolved in 0.8 mL of methanol/sodium citrate solution 0.1 M (25/75) and filtered through a 0.22 μm filter for HPLC analysis.

Dispersive Liquid-Liquid Microextraction Method.
e method used in DLLME was obtained from Wang et al. [8] and Zhou et al. [9], which was modified by the description given below.
A volume of 5 mL sample solution was placed in a conical bottom tube, 1 g sodium chloride was added, and the pH was adjusted to 12 using a 5 N NaOH solution.
A mixture of 500 μL acetonitrile and 90 μL toluene was rapidly injected into the sample using a syringe, and a cloudy solution was obtained. e solution thus prepared was then shaken for 2 min and was then centrifuged at 6000 rpm for 5 min. e dispersed fine droplets of toluene were sedimented at the bottom of the test tube. e upper layer was removed, and the residual phase was blown to almost dryness with lowpressure nitrogen gas. Finally, the extract was dissolved in 0.5 mL of methanol/sodium citrate solution 0.1 M (25/75) and filtered through a 0.22 μm filter for HPLC analysis.

SPE Chromabond ® C18 ec.
SPE cartridges were conditioned based on the method explained by Oostdyk et al. [5] as the procedure with 5 ml methanol and 5 ml distillated water. en, 50 ml sample solution was adjusted to pH 10 with 10 M sodium hydroxide solution and was passed slowly through the cartridges. e cartridges were eluted three times with 1 ml of ethyl acetate. Eluents were evaporated in a stream of nitrogen. e remaining residue was dissolved in 5 mL of methanol/sodium citrate solution 0.1 M (25/75) and collected into a 5 ml graduated tube. e solutions were filtered through a 0.22 μm filter for HPLC analysis.

SPE
Chromabond HR-P. SPE cartridges were conditioned as advised by Macherey-Nagel Company [11], with 2 × 5 ml methanol, 2 × 5 ml acetonitrile, and 2 × 5 ml 10 −5 M sodium hydroxide solution. en, 50 ml sample solution was adjusted to pH 9 with 10 M sodium hydroxide solution and was passed slowly through the cartridges. e cartridges were eluted by 3 × 1.5 ml methanol/acetonitrile (1 : 1). Eluents were evaporated in a stream of nitrogen. e remaining residue was dissolved in 5 mL of methanol/sodium citrate solution 0.1 M (25/75) and collected into a 5 ml graduated tube. e solutions were filtered through a 0.22 μm filter for HPLC analysis.

SPE SCX.
Aznar et al. [16] applied SCX cartridges for PAAs, and the method was modified as below: cartridges were conditioned with 2 × 3 ml methanol and 2 × 3 ml 3% acetic acid (w/v). en, 50 ml sample solution was passed slowly through the cartridges. e cartridges were eluted with 5 × 1 ml methanol/sodium citrate solution 0.1 M (25/75) and collected into a 5 ml graduated tube. e solutions were filtered through a 0.22 μm filter for HPLC analysis.    (Table 1). ere was a good linearity in the range of 0.05-0.8 mgl −1 for all five target compounds.

Accuracy.
In order to investigate the recoveries of the method including SPE procedures, blank food simulants (3% acetic acid) spiked with five PAAs at three different concentrations 7.5, 15, and 30 μgl −1 were performed with three replications (n � 3). e recoveries are presented for the five aromatic amines ( Table 2). ese data confirm not only the accuracy of the method, but also the integrity of the SPE procedure.

Precision. Precision calculation was done based on the repeatability criterion in one day (intraday) and in 3 consecutive days (interday).
Intraday precision was evaluated with spiked blank food simulant (3% acetic acid) at three different concentrations 7.5, 15, and 30 μgl −1 including SPE procedures. e RSD of intraday was obtained (Table 3).
Interday precision was evaluated with spiked blank food simulant (3% acetic acid) at three different concentrations 7.5, 15, and 30 μgl −1 including SPE procedures on three different days. e RSD of interday was obtained (Table 3).

Sensitivity.
Relative standard deviation (RSD) of the method was performed with spiked blank food simulant (3% acetic acid) by five PAAs at concentration 7.5 μgl −1 , including the SPE procedure. LOD and LOQ of the method were calculated (Table 4).

Results and Discussion
e migration of primary aromatic amines (PAAs) from flexible food packaging represents a serious risk to public health as these compounds are potentially carcinogenic substances. e source of PAAs is from the residues of aromatic diisocyanates (2,4-TDI, 2,6-TDI, and 4,4′-MDI) arising from incomplete curing of the main polyurethane (PU) adhesive, and also other aromatic diisocyanates such as 2,4′-MDI and 2,2′-MDI are present in adhesive inappropriately. e five PAAs (2,4-TDA, 2,6-TD, 4,4′-MDA, 2,4′-MDA, and 2,2′-MDA) used in this study were selected on the basis of their origin of PU adhesives used in food packaging. Toxicity of PAAs was evaluated and classified into three groups based on their levels of toxicity with 2,4-TDA, 2,6-TDA, and 4,4′-MDA being in the high toxicity class [18]. According to the EU Regulation (EU 10/2011), detection limit of the released PAAs must be below 10 µg/kg of food or food simulant. e method was developed and designed based on the worst-case scenario for the migration of PAAs from packaging material into a food simulant which in this study was 3% acetic acid. e aim of the research was to design and select an extraction method, liquid and/or solid phase, of analysis for these PAAs (TDAs and MDAs) with high sensitivity and also to make possible to understand the origin of the PAAs detected. e method was designed for 3% acetic acid (as food simulant) in purified water, since this simulant was considered the most restrictive, that means it uses the worst-case scenario for the migration of TDAs and MDAs from food packaging. e direct analysis of the simulant by HPLC-UV did not provide enough sensitivity, as the sample amount injected into the system was very low. For this reason, the LLE and SPE experiments were needed prior to the HPLC detection.
TDAs and MDAs have two functional groups (-NH 2 ) that are polar and partly soluble in water; as a result, the  extraction of these compounds from aqueous solution is difficult [11]. TDAs (single ring) in comparison with MDAs (two rings) have more polarity and are more soluble in water. erefore, the mean recoveries for the extraction of TDA derivatives (2,4-TDA and 2,6-TDA) should be less than that of MDA derivatives (4,4′-MDA, 2,4′-MDA, and 2,2′-MDA).
In liquid extraction methods, in general data obtained for mean recoveries of extraction for both LLE and DLLME were not very satisfactory. e mean recoveries of LLE is better than those of DLLME, because the number of extraction steps could be increased in the LLE method, while there is just one time dispersion in the DLLME method. e mean recoveries for the extraction of MDAs were greater than that of TDAs in both methods. DLLME were used in extraction of aromatic amines in aqueous matrix with acceptable recoveries [8,9], but in our research, the interested aromatic amines have two functional group (-NH 2 ) which are more soluble in aqueous solutions, and consequently, recoveries are lower than LLE and other methods.    In SPE methods, the mean recoveries of TDAs and MDAs were in the order of SCX > HR-P > C18 ec. Octadecyl sorbents (nonpolar) are not suitable for TDAs and MDAs in the C18 cartridge. In HR-P columns, the cross-linking in SDVB leads to better adsorption than C18 for the polar PAAs. e mean recoveries of the five extraction methods were compared, and the SPE method with SCX cartridge was the best (more than 90%). e result of all applied methods are compared and shown in Figure 1.
SCX cartridge with the mechanism of action being based on ion exchange could properly adsorb polar PAAs and so has the best mean recovery among the studied cartridges. Figure 2 shows the chromatogram of the standard solution of five PAAs (100 ng·ml −1 ). e chromatogram of 15 ng·ml −1 spiked food simulant, in which the final concentration of the extracted and preconcentrated solution by SCX cartridge is 150 ng·ml −1 , is shown in Figure 3. Also the chromatograms of each method are compared for the performance of the extraction and shown in Figures 4-7.  HPLC method combined with SCX cartridge was validated and developed for the separation and quantitation of five primary aromatic amines in 3% acetic acid as food simulant. After optimizing all separation parameters, the good separation of five PAAs in food simulant was feasible within 30 min. Additionally, the method was validated for linearity, LOD, precision, and intra-and interday variation.

Conclusions
It is important to determine and control the migration of contaminants from materials in contact with food into the food. Packaging materials provided this route for migration of contaminants such as PAAs from PU adhesives which are used in laminated multilayer films; some PAAs can migrate to the food. Analysis of PAAs is very valuable in the European regulations, and for this purpose, EU has established regulations for PAAs. Simultaneous extraction and analysis of PAAs (2,4-TDA, 2,6-TDA, 4,4′ MDA, 2,4′ MDA, and 2,2′ MDA) lead to the process of food safety assessment and maintenance, also a means to monitor the use of authorized adhesives and prevention of unauthorized ones being used. Solid-phase extraction SCX is introduced as a suitable method for extraction of polar primary aromatic amines (PAAs) which might have migrated from the laminated multilayer films of packaging materials into 3% acetic acid as food simulant.