Determination of Acrylamide in Commercial Baby Foods by LC-QqQ-MS/MS: a Simple Method for Routine Analyses

A method based on liquid chromatography-triple quadrupole tandem mass spectrometry was developed and validated for the analysis of acrylamide in baby foods. The sample preparation involves a simple extraction process using a mixture of acetonitrile:water:formic acid (69:30:1, v/v/v) in combination with dispersive solid-phase extraction (d-SPE) using alumina as sorbent. The method shown good linearity within the range 20–250 μg kg−1 through matrix-matched and solvent calibrations. The recovery rates for acrylamide ranged from 100 to 108% with coefficients of variation below 10%, under repeatability and reproducibility conditions (within-laboratory). The obtained limit of quantification (20 μg kg−1) complies with the values indicated by the European Union for acrylamide analysis in baby foods. The developed and validated method was applied to 50 ready-to-eat baby foods available in the Brazilian market. Acrylamide was detected in 13% of samples primarily composed of fruits, whereas it was found in approximately 37% of baby food containing meat and/or vegetables. Additionally, in 2 samples, the detected levels exceeded the benchmark value established by the EU (40 μg kg−1). The study showcases the applicability of this method for routine analysis of acrylamide detection.


Introduction
According to the International Agency for Research on Cancer (IARC), acrylamide has been classified as "probably carcinogenic to humans" in group 2A (IARC 1994). It can be found in various heat-treated foods such as potato chips, dried fruits, biscuits, pastry, chocolate, bread, and coffee (Khan et al. 2017(Khan et al. , 2018. This compound is primarily present in carbohydrate-rich foods that have undergone heat processing at temperatures exceeding 120 °C under lowmoisture conditions (Fernández et al. 2022;Boyaci-Gunduz 2022). The Maillard reaction is a significant pathway for the formation of this contaminant in foods, being asparagine the main precursor. Additionally, other pathways, including acrylic acid, acrolein, 3-aminopropionamide, and wheat gluten, have also been involved in the generation of acrylamide (Capuano and Fogliano 2011). It is well-known that glycidamide, one of the metabolites formed after acrylamide intake, represents the major route of genotoxicity and carcinogenicity associated with this contaminant (EFSA 2015).
Baby foods have been identified as the main contributor to the total exposure of infants, followed by other products made with potatoes and cereal-based ingredients (EFSA 2015). Furthermore, the exposure to acrylamide in infants, toddlers, and other children is higher compared to adults, owing to the higher daily food intake relative to body weight (Boyaci-Gunduz 2022). Therefore, this contaminant generated during thermal processes poses a significant risk to children's health, as it is commonly present in the diet of this vulnerable group of consumers 1 3 (Boyaci-Gunduz 2022). Hence, it is needed to stablish measures for controlling and mitigating the presence of acrylamide in food, as well as setting maximum limits in foodstuffs.
The regulatory benchmark level for the presence of acrylamide in food intended for infants and young children is set at 40 μg kg −1 by the European Commission (European Commission, 2017). However, the acrylamide content in foods can vary significantly depending on the processing technique, ingredients used, storage methods, moisture content, and pH levels (Timmermann et al. 2021). Potato products, such as French fries, have traditionally been the main sources of acrylamide in the human diet (Elias et al. 2017). However, purple and red varieties of potatoes are more prone to acrylamide formation compared to yellow potatoes due to the higher content of sugar (Orsák et al. 2022). Consequently, strategies can be used to reduce acrylamide formation in children's food by substituting ingredients and adjusting thermal processing conditions.
In the literature, there are limited reports on acrylamide in baby foods, and the existing methods involve complex sample preparation steps due to the nature of these matrices. For the extraction of acrylamide and other processing contaminants from fruit-based baby foods, a combination of acetonitrile, as extraction solvent, with dispersive primary secondary amine (PSA) and cation-exchange solid-phase extraction (SPE) cleanup was used (Petrarca et al. 2017). The QuEChERS (quick, easy, cheap, effective, rugged, safe) method followed by dispersive solid-phase extraction (d-SPE) cleanup step using PSA as sorbent was employed for the analysis of meat and/or vegetables and fruit-based baby foods (Elias et al. 2017). Another study applied water/nhexane partitioning, followed by clarification using Carrez reagents and SPE cleanup, for the extraction of acrylamide in baby food containing meat and vegetables/cereals processed (Mojska et al. 2012). In the analysis of cereal-based baby food with fruit purees, methanol/water mixture was used as the extracting solvent, followed by defatting with n-hexane, freezing at − 18 °C, and SPE cleanup with Oasis HLB cartridges (Michalak et al., 2013). Additionally, for the analysis of acrylamide in meat-based baby foods, defatting with petroleum ether, extraction with an aqueous solution of sodium chloride, liquid-liquid extraction with ethyl acetate, and cleanup by Oasis HLB SPE cartridges were employed (Jiao et al. 2005). Fohgelberg et al. (2005) reported an extraction method using water, followed by filtration through an SPE column (Multimode, 1 g), and the filtered extract was loaded onto an SPE column (ENV+) for determination of acrylamide in meat, vegetables, and fruit-based baby foods. However, many of these methods reported employed SPE technique, which is time-consuming, expensive, and not easily applicable for sample preparation (Orlando and Simionato 2013;Tuzimski et al. 2016).
In the present study, a simple extraction approach followed by liquid chromatography triple quadrupole tandem mass spectrometry (LC-QqQ-MS/MS) was proposed for the detection of acrylamide in baby foods with different compositions. The analytical method fulfills the requirements for quantitative analyses of acrylamide in food matrices established by the Commission Regulation (EU) 2017/2158 (European Commission 2017). Furthermore, the feasibility of the developed method was demonstrated by analyzing 50 commercially available ready-to-eat baby food samples from the Brazilian market.

LC-QqQ-MS/MS Analysis
The LC-QqQ-MS/MS conditions were based on a previous study described by Ferrer-Aguirre et al. (2016). However, several modifications were made to enhance the analytical signal obtained for acrylamide. Chromatographic analysis was performed using a liquid chromatograph 1290 RRLC instrument (Agilent, Santa Clara, CA, USA) equipped with a Jet Stream electronic spray ionization (ESI) source (G1958-65138) and connected to an Agilent triple quadrupole mass spectrometer (6460 A). The MassHunter Workstation Quantitative Analysis for QQQ (Agilent) software was employed for data acquisition. The MS conditions were as follows: gas flow rate of 5 L min -1 ; source temperature of 300 °C; sheath gas temperature of 400 °C; nebulizer pressure set at 45 psi; sheath gas flow rate of 11 L min -1 ; capillary voltage of 3500 V; and nozzle voltage of 500 V. The dwell time for all selected reaction monitoring (SRM) transitions was set to 0.15 s. Nitrogen was used as nebulizing and collision gas. Chromatographic separation was achieved on an ACE Excel 3 SuperC18 column (150 × 4.6 mm, 3.0-μm particle size), maintained at 30 °C (Advanced Chromatography Technologies, Aberdeen, Scotland). The mobile phase consisted of an aqueous solution containing 0.1% of formic acid (v/v) (eluent A) and methanol (eluent B). The flow rate was set at 0.4 mL min −1 , and the gradient profile was as follows: 95% eluent A for 3 min, followed by a linear decrease to 0% in 4 min, and then held for 11 min. Subsequently, eluent A was restored to 95% within 1 min and maintained for 1 min. The total running time was 20 min, and the injection volume was 5 μL.
Data acquisition was performed in SRM mode using ESI in positive mode ( Table 1). Acquisition of ion transitions was established for acrylamide and acrylamide-d3, and the protonated molecular ion [M + H] + were selected for each compound. A retention time (t R ) of 6.5 min (± 0.1 min of a maximum tolerance) was employed for the positive identification of acrylamide in baby food samples. The IS was added to the samples at a concentration of 100 μg kg −1 prior to the extraction procedure.

Sampling
Two groups of baby foods were obtained in the city of Campinas, SP, located in the South-Eastern region of Brazil. One group primarily consisted of baby foods made with fruits (n = 23), while the other group comprised meat-and/ or vegetable-based baby foods (n = 27). A total of 50 commercial baby food samples were randomly collected in local retail supermarkets and stored in their original packaging, such as glass jars (up to 170 g each) or plastic pots (99 g each), at room temperature until analysis.

Sample Extraction
For the extraction, 1 g of baby food (spiked with 100 μg kg −1 of IS), previously homogenized, and 9 mL of acetonitrile:water:formic acid (69/30/1, v/v/v) solution were added into a centrifuge tube (polypropylene). The mixture was agitated on a rotatory shaker for 60 min at 100 rpm, followed by centrifugation at 3061×g for 20 min. Then, the supernatant (1.5 mL) and 50 mg of alumina sorbent were vortexed for 1 min and then centrifuged at 3061×g for 5 min. The extract was filtered through a 0.2-μm nylon syringe filter and transferred to a glass vial for subsequent LC-QqQ-MS/ MS analysis.

Validation Procedure
The analytical method was in-house validated according to the following documents: Eurachem Guide recommendations criteria (Magnusson and Örnemark, 2014) and Commission Regulation (EU) 2017/2158 (European Commission, 2017). Therefore, analytical selectivity, accuracy, limit of detection (LOD), limit of quantification (LOQ), linearity in solvent and matrix-matched calibration curves, and matrix effects were evaluated using a representative baby food matrix. Accuracy was evaluated through recovery and precision trials, which were carried out under repeatability and within-laboratory reproducibility conditions.
The LOD and LOQ were determined by analyzing blank baby food samples spiked at different concentration levels. For this, the LOD was defined as the concentration that gives a signal-to-noise ratio of 3:1, while the LOQ was set at the concentration that produces a signal-to-noise ratio of 10:1, The matrix effect was estimated by comparing the slopes obtained from solvent and matrix-matched calibration curves within the range of 20-250 μg kg -1 , using Eq. 1: Linearity was evaluated using solvent and matrixmatched calibration curves within the concentration range of 20-250 μg kg −1 , including the LOQ as the lowest concentration level. To assess repeatability, recovery (%) and precision (relative standard deviation-RSD, %) were determined by analyzing five replicates of spiked baby food samples at two levels (20 and 100 μg kg −1 ). These replicates were extracted within a single day. To evaluate reproducibility, five replicates of spiked baby food samples at each concentration level (20 and 100 μg kg −1 ) were analyzed on five different days.

Optimization of the Extraction Procedure
To minimize the steps involved in the determination of acrylamide, a conventional solid-liquid extraction (SLE) was employed using a mixture of acetonitrile, water, and formic acid (69/30/1, v/v/v) as extraction solvent. This SLE procedure was based on a previous study (Tölgyesi and Sharma 2020), with some modifications considering the effectiveness of these solvents in extracting acrylamide at low levels from complex matrices such as high-sugarcontent foods, bread, and fried potato (Tölgyesi and Sharma 2020). The use of acetonitrile minimizes the co-extraction of common matrix components like highly non-polar fats while also aiding in protein precipitation (Petrarca et al. 2017). Additionally, the acidified acetonitrile has proven to be more effective for the extraction of acrylamide from different types of food samples (Tölgyesi and Sharma 2020).
During the optimization of the extraction solvent volume, different amounts of acetonitrile, water, and formic acid solution (69/30/1, v/v/v) were tested, ranging from 3 to 9 mL. A baby food sample consisting of meat and vegetables (potato, carrot, tomato, and onion) was spiked at 200 μg kg −1 . The results are shown in Fig. 1, and it was observed that as the volume of the extraction solvent increased, the recovery of acrylamide also increased. When a one-factor analysis of variance was used, significant difference was observed, observing similar results when an extraction volume higher than 5 mL was used. Moreover, a decrease in the matrix effect was observed when the extraction solvent (1) Matrix Effect (%) = (matrix slope − solvent slope) solvent slope x 100 increase, likely due to the dilution of co-extractives. Consequently, the largest volume tested was selected for the extraction of acrylamide from baby foods. Due to the significant amount of matrix interferences that can be co-extracted with acrylamide when water is used as the extraction solution (Ferrer-Aguirre et al. 2016), a cleanup step based on the dispersive solid-phase extraction (d-SPE) technique was studied. Thus, 1.5 mL of the extract obtained from a baby food sample containing meat and vegetables was submitted to two different cleanup procedures: (i) d-SPE with 50 mg of PSA and (ii) d-SPE with 50 mg of alumina (Fig. 1). Both tested sorbents provide suitable recoveries. However, a slightly higher recovery was observed when alumina was used as the cleanup sorbent in the d-SPE process, although when a two sample t-test was carried out, significant difference were not observed between the results obtained for both sorbents. Additionally, a low matrix effect of approximately 4% was also observed using this sorbent (Fig. 1). Consequently, 50 mg of alumina was selected for the cleanup step, effectively reducing the occurrence of co-extractives in the final extract, which could potentially affect the reproducibility and sensitivity of the analytical method. Moreover, it should be noted that alumina sorbent can adsorb amino acids (Omar et al. 2015). Considering that valine has been identified as a critical interference in the analysis of acrylamide in food matrices by LC-MS (Şenyuva and Gökmen 2006), the use of alumina sorbent can prevent the presence of this matrix interference.

Method Validation
Analytical selectivity was verified by analyzing a "blank" baby food and spiked baby food samples, and no interfering peaks were observed at the retention time of the acrylamide in the SRM chromatograms, allowing an unequivocal identification of the target compound (Fig. 2). Moreover, adequate linearity was observed in both solvent and matrix-matched calibration curves within the range 20-250 μg kg −1 , with coefficients of determination (R 2 ) ≥ 0.9932. The LODs and LOQs were 5 and 20 μg kg −1 , respectively. The achieved LOQ meets the requirements indicated by the Commission Regulation (EU) 2017/2158 (European Commission 2017), which mandates a minimum LOQ of 20 μg kg −1 for analytical methods used for the determination of acrylamide in baby foods, bearing in mind that the benchmark level is lower than 125 μg/kg (40 μg/kg).
The extraction efficiency of the developed method was evaluated through recovery and precision studies. The average recoveries ranged from 100 and 108%, which fall within the acceptable range of 75-110% for analytical methods intended for acrylamide analysis (European Commission 2017). Furthermore, the precision of the method meets the criteria set by Commission Regulation (EU) 2017/2158 (European Commission, 2017). Under repeatability conditions, the RSD values ranged from 7 to 10%, while under within-laboratory reproducibility conditions, the RSD was 8% for both analyzed concentration levels. All these values were lower than the coefficient variation (CV) estimated at the level of 100 μg kg −1 (22.6%), which was calculated using the Horwitz equation: CV = 2( 1-0.5logC ), where C is the mass fraction expressed as a power (exponent) of 10. For levels below 100 μg kg −1 , it is recommended to have CV values as low as possible (European Commission 2002).
Additionally, the extraction efficiency was also assessed using three other baby food samples with different formulations: (i) a fruit purée-based baby food composed of apple, peach, banana, and apricot; (ii) a vegetable purée-based baby food composed of carrot, tomato, potato, onion, spinach, and chickpea; and (iii) a baby food sample composed of fish and vegetables. These samples were spiked at 20 μg kg −1 (LOQ level) and extracted following the procedure described in the "Sample extraction" section. The matrix effect and recovery were evaluated. The results showed a low matrix effect and high recoveries, ranging from 80 to 109% (Fig. 3), indicating that our proposed method is robust and suitable for the analysis of acrylamide in baby foods with diverse compositions.

Analysis of Commercial Samples of Baby Foods
To assess the suitability of the proposed method, a total of 50 commercial baby food samples available in the Brazilian market were analyzed but acrylamide was only detected in 13 samples as it was indicated in Table 2. The concentration of acrylamide varied depending on the ingredients of the sample.
Among the fruit-based baby foods, acrylamide was detected in 13% of the samples, with concentration ranging from < LOQ to 37 μg kg −1 . Furthermore, acrylamide was also detected in baby foods containing plum, albeit at levels < LOQ. In the case of baby foods containing apples and strawberries, the contaminant was detected at a concentration of 20 μg kg −1 . Figure 4 displays the SRM chromatograms of the samples naturally contaminated with acrylamide. Fig. 1 Effect of the extraction solvent volume on a recovery (%) for acrylamide in a spiked matrix of baby food composed of meat and vegetables (potato, carrot, tomato, and onion) at 200 μg kg −1 level using alumina or PSA as sorbents and b matrix effect for acrylamide in a spiked matrix at 200 μg kg −1 after performing the clean-up with alumina or PSA 1 3 The acrylamide levels found in this study were in accordance with those values reported in the literature. Similar content was reported in plum-based baby food from Brazil with a concentration of 35 μg kg −1 (Petrarca et al. 2017). In cereal-based baby foods containing fruit purees from Poland, acrylamide levels ranged from 10.8 and 15.7 μg kg −1 (Michalak et al. 2013). Fruit-based baby foods from Estonia had a mean acrylamide content < 30 μg kg −1 (Elias et al. 2017). Commission Regulation (EU) 2017/2158 has established measures to minimize the presence of acrylamide in In baby food samples containing meat and/or vegetables, higher levels of acrylamide were observed compared to fruit-based baby foods. Approximately 37% of the analyzed samples contained detectable amounts of acrylamide.
Among them, the highest acrylamide level was found in a sample of stroganoff and rice (90 μg kg −1 ) followed by chicken risotto (56 μg kg −1 ) ( Table 2). Both samples contained a high proportion of potatoes in addition to meat and rice. Cereal grains (such as rice and wheat) and potatoes are known to be rich in the precursors that contribute to the formation of acrylamide, which could explain the higher levels of the contaminant in these samples (Tateo et al. 2007). The formation of acrylamide is also influenced by processing conditions and food seasonings (Tateo et al. 2007). Regarding the samples made solely with vegetables, acrylamide was detected in one sample at a concentration < LOQ (20 μg kg −1 ).
Acrylamide was also reported in meat and/or vegetablebased baby foods from Poland, with an average level of 55 μg kg −1 (Mojska et al. 2012). In Sweden and China, levels up to 25 and 124.93 μg kg −1 , respectively, were also observed (Fohgelberg et al. 2005;Jiao et al. 2005). Acrylamide levels in samples from Estonia were similar to those obtained in the present study (Elias et al. 2017).

Conclusions
The combination of LC-QqQ-MS/MS with a simple and easy acidified aqueous acetonitrile-based extraction achieves adequate performance characteristics for monitoring acrylamide in baby foods with complex and diverse compositions. An attractive feature of the proposed method is its efficient removal of matrix co-extractives, as evidenced by the low matrix effect (approximately 4%), achieved through dispersive alumina sorbent cleanup, without compromising the analytical sensitivity and accuracy. Moreover, the sample preparation approach eliminates the need for expensive and time-consuming SPE techniques typically used in acrylamide analysis in food samples. The results obtained in this Purple sweet potato purée, ora-pro-nobis, and corn <LOQ 13 Vegetables and meat <LOQ Fig. 4 Chromatograms obtained in the selective reaction monitoring (SRM) mode of baby food samples of A chicken risotto and B apple and plum (plastic bag), both containing naturally acrylamide at a concentration of 56 and 37 μg kg −1 , respectively study further support previous findings that the acrylamide content in baby foods is influenced by their composition. Therefore, by selecting baby foods with ingredients that exhibit lower acrylamide formation during thermal processing, it is possible to achieve reduced dietary exposure to acrylamide in infants.