Quantification of acrylamide in infant formula via a modified QuEChERS protocol coupled with high-performance liquid chromatography-tandem mass spectrometry

Abstract

The accurate quantification of acrylamide within complex infant formulas remain a significant challenge. This study validates a novel method for analyzing acrylamide in infant formula, which combines the quick, easy, cheap, effective, rugged, safe (QuEChERS) pretreatment approach with the isotope dilution HPLC-MS/MS technique. The optimized QuEChERS procedure effectively eliminates co-extracted matrix interferences and achieves high extraction efficiency. Notably, this approach simultaneously accomplishes a recovery rate of 103 % to 105 %, a relative standard deviation of less than 1.6 %, and a limit of quantification at 0.90 μg/kg. The high accuracy of this proposed method is further confirmed using commercially certified reference material (CRM, No. 108–02-006). Importantly, when 20 real samples were analyzed, five were found to be contaminated with acrylamide. These results demonstrate that the proposed method is rapid and reliable for determining acrylamide levels in infant formula, suggesting its potential as a primary detection method for acrylamide in complex matrices.

1. Introduction

Acrylamide (2-propenamide) is an α, β-unsaturated amide with a low molecular weight (Omar et al., 2015). It is a heat-induced food contaminant commonly found in food matrices. Previous studies have revealed that acrylamide could be easily observed in protein- and carbohydrate-rich foods, which could be formed via the Maillard reaction between the amino group of the asparagine and the carbonyl group of sugars at temperature exceeding 120 °C (Bertuzzi et al., 2020; Mottram et al., 2002; Stadler et al., 2002). Acrylamide has been classified as a probable Group 2 A human carcinogen by the International Agency Research on Cancer (IARC), and is treated as one of the neurotoxic and genotoxic compounds (IARC, 1994). The European Commission has also set a benchmark level of 40 μg/kg for the acrylamide content in infant foods (EU Commission Regulation, 2017).

The accurate quantification of acrylamide at extremely low concentrations remains challenging due to its low molecular weight and highly polar characteristics, especially within complex food matrices (Desmarchelier et al., 2020; Omar et al., 2015). Several highly selective detection methods, including liquid chromatography (LC) and gas chromatography (GC) coupled with mass spectrometry (MS), have been developed for acrylamide detection (Ghiasi et al., 2022; Lee et al., 2024; Omar et al., 2014). Given the complexity of food matrices, an effective sample pretreatment process is typically required for acrylamide determination. For GC-related techniques, a lengthy derivatization step in sample pretreatment is necessary (Ghiasi et al., 2022), and the high injection temperature may lead to the formation of acrylamide from co-extracted precursors. Furthermore, occasional false results have been reported (Desmarchelier et al., 2020; Omar et al., 2015; Sebastià et al., 2023; Sun et al., 2022). On the other hand, acrylamide can be analyzed directly without derivatization using LC-MS/MS-based techniques, but interferences are prominent in the chromatography, and the signal intensity of acrylamide is influenced by both the clean-up procedures and LC conditions (Desmarchelier et al., 2020; Sun et al., 2022). High-resolution mass spectrometry has shown its ability to identify trace analytes in complex matrices with improved mass accuracy and resolution (Shi et al., 2022), but the high cost of related instruments hinders widespread implementation. Therefore, it is crucial to develop efficient extraction and clean-up procedures, i.e., sample pretreatment techniques, for complex food matrices. Current validated methods for extracting acrylamide from food matrices encompass solid phase extraction (SPE) (Jiao et al., 2005; Wang, Cai, et al., 2022), dispersive liquid-liquid microextraction (DLLME) (Ghiasi et al., 2022), and QuEChERS (Sebastià et al., 2023). The latter has been innovatively utilized for acrylamide extraction in food matrices, employing various sorbents such as primary secondary amine (PSA), C18, and basic alumina (Al₂O₃) (Sebastià et al., 2023). These methods have achieved commendable limits of detection (LOQ) and repeatability across diverse food matrices, including French fries (Gao et al., 2021), potato chips (Jung et al., 2021), potato cutlets (Kumari et al., 2023), chocolate (Kruszewski & Obiedziński, 2020), crisp bread (Wei et al., 2020), pastry products (Cheng et al., 2021), and other ready-to-eat high-fat food products (Huang et al., 2019; Zhu et al., 2021). Notably, prior results suggest that QuEChERS-based extraction techniques offer potential advantages for determining acrylamide in complex food matrices.

Infant formula is a complicated matrix, containing protein, fat, carbohydrates, and other multiple dietary additives. It is highly desirable to develop an accurate detection method for acrylamide in infant formula since the matrix critically interferes with the quantification of analysts. Inspired by the merits of QuEChERS-based extraction techniques (Li et al., 2024; Sebastià et al., 2023; Yang et al., 2024), we report the validation of acrylamide in infant formula samples by the combination of a modified QuEChERS extraction method and UPLC-MS/MS technique. Efficient removal of matrix interference was achieved in infant formula samples via the optimization of sorbents used in the QuEChERS pretreatment step. Moreover, the isotope-labeled internal standard method provided the reliable quantification of acrylamide in infant formula. The detection results also verified that the proposed method could be suitable for the screening and determination of acrylamide in infant formula samples and milk powders.

2. Material and methods

2.1. Reagents and chemicals

Acrylamide certified reference material (CRM, GBW10057) was obtained from the National Institute of Metrology, China (NIM, China), while ¹³C₃-acrylamide (with a purity of ≥99 %) was purchased from Cambridge Isotope Laboratories (Tewksbury, Massachusetts, USA). Acrylamide in infant formula (CRM No. 108–02-006) was obtained from the Korea Research Institute of Standards and Science (KRISS, Daejeon, Republic of Korea). Anhydrous magnesium sulphate (MgSO₄) was purchased from Macklin (Shanghai, China). Methanol (MeOH) and acetonitrile (ACN) of HPLC grade were purchased from Merck (Darmstadt, Germany). Formic acid (FA) and n-hexane were obtained from Honeywell (Shanghai, China). The study utilized an ultra-pure water system (Milli-Q, Darmstadt, Germany). Sodium chloride (NaCl), basic alumina (Al₂O₃) and aminopropyl sorbents columns were all purchased from Sigma-Aldrich (Darmstadt, Germany). Primary secondary amine (PSA) and C18 sorbents columns were obtained from Agilent Technologies (California, USA). Twenty real samples were obtained from supermarkets in Beijing, encompassing nine types of infant formula, three types of skim milk powder, five types of whole milk powder, one camel milk powder, one donkey milk powder, and one goat milk powder.

2.2. Preparation of standard solutions

The stock solutions (1.00 mg/g) of acrylamide and ¹³C₃-acrylamide were prepared in methanol, and then stored at −20 °C. The acrylamide standard solutions with concentration ranging from 0.5 to 100 μg/kg were prepared from the stock solutions. The stock solutions were further diluted using water. After that, the ¹³C₃-acrylamide solution was added into the acrylamide standard solutions to obtain solutions with different isotope ratios (acrylamide/¹³C₃-acrylamide). The ratio of acrylamide/¹³C₃-acrylamide changed from 1:10 to 20:1. Finally, the calibration curve was plotted by the ratio of response (acrylamide/¹³C₃-acrylamide) vs. the ratio of content (acrylamide/¹³C₃-acrylamide) in water.

2.3. Pretreatment of infant formula samples

The modified QuEChERS pretreatment process in the study was described as follows: The concentration of acrylamide in the infant formula sample was preliminarily estimated. 1.00 g of infant formula was added into a polypropylene centrifuge tube (50 mL), and then, a certain amount of solution of ¹³C₃-acrylamide was spiked into the tube to reach an estimated ratio between acrylamide and ¹³C₃-acrylamide of 1:1. The mixture was dissolved in 10 mL of water, and 15 mL of acetonitrile (containing 0.1 % FA) was added. After stirring for 1 min, 5 mL of n-hexane was added, and the tube was stirred for one more 1 min. Next, a mixture of 3.00 g of anhydrous MgSO₄ and 1.00 g NaCl was added into the tube. The sample was vibrated immediately for 15 min, and then centrifuged at 9000 rpm for 6 min. The top layer in the tube was hexane, which was removed using a Pipette. Then, 4 mL of acetonitrile layer was transferred into another centrifuge tube, which contained 200 mg of basic Al₂O₃ and 150 mg of MgSO₄. The mixture in the tube was then stirred for 1 min, and centrifuged at 7000 rpm for 6 min. Finally, the liquid layer was transferred into a tube using a Pipette and dried under the nitrogen atmosphere with a sample concentrator (RayKol Auto EVA 80). The as-obtained solid was dissolved in water (1 mL). Prior to injection into LC-MS/MS equipment, the solution was filtered using the PTFE syringe filters (0.22 μm).

2.4. LC-MS/MS analysis

The analysis of acrylamide was carried out using the Waters ACQUITY™ UPLC system coupled with a Waters Xevo™ TQ-S tandem triple quadrupole mass spectrometer (Manchester, UK). An Atlantis dC18 column (150 mm × 3.0 mm, 5 μm) was utilized in the LC system. The mobile phase was a mixture of water (A) and methanol (B). The flow rate was 0.4 mL/min. The gradient elution program started with 1 % B and then ran for 6 min. After that, the proportion of B increased linearly to 20 % within 1 min. After keeping 20 % B for 0.5 min, the gradient elution changed to 1 % B at 7.51 min and kept isocratic for 3.5 min. The total injection volume was 5 μL.

The detection of MS was conducted in multiple reaction monitoring mode (MRM) using positive electrospray ionization (ESI⁺). The capillary voltage was adjusted to 3.0 kV. The temperature of the ion source and desolvation were maintained at 150 °C and 550 °C, respectively. Nitrogen served as both the cone gas and the desolvation gas, with flow rates of 150 L/h and 800 L/h, respectively. For the analysis of acrylamide, the precursor ion [M + H]⁺ with a mass of 72.0 was fragmented using a collision energy of 10 V. The resulting product ions with masses of 55.0 and 44.1 were then observed. For ¹³C₃-acrylamide, the precursor ion [M + H]⁺ with a mass of 75.4 was fragmented at collision energies of 9 V and 15 V. The corresponding product ion masses were 58.1 and 45.0, respectively.

2.5. UHPLC-QE plus-orbitrap MS

The identification of acrylamide in procedural blank was performed using a UHPLC QE plus Orbitrap MS system (Thermo Fisher Scientific). Optimal parameters were as follows: spray voltage: 3.50 kV; capillary temperature: 270 °C; sheath gas flow rate: 25.0 μL/min; aux gas flow rate: 14.0 μL/min; sweep gas follow rate: 0 μL/min; S-lens RF level: 50.0 %; aux gas heater temperature: 280 °C. The data was collected under parallel reaction monitoring (PRM) mode. This mode included a targeted scan (scanning range: m/z 70–80; resolution: 35000 FWHM) and a full MS² scan with applied fragmentation energy of 10 V. The isolation window was set to be 1.0 m/z with a resolution of 175,000 FWHM. The automatic gain control (AGC) target was controlled to be 1.0e⁶ with a maximum injection time of 200 ms.

2.6. Matrix effect

The evaluation of the matrix effect (ME) of infant formula on acrylamide quantification was conducted utilizing the UPLC-MS/MS method. Firstly, blank matrices were prepared from the infant formula samples, which were pretreated with the same procedure as those utilized in the QuEChERS method. Then, a matrix-matched calibration solution at 5 μg/kg was prepared by adding a certain amount of acrylamide into the blank matrices. In parallel, the standard calibration solutions of acrylamide were prepared using water as solvent. It should be noted that the concentration of acrylamide in the matrix-matched calibration solution was equivalent to that of the standard calibration solution (Wang, Li, et al., 2022). Two factors, namely, ME_α and ME_β, were used to estimate the ME on acrylamide and ¹³C₃-acrylamide quantification in MS, respectively. The correction factor (θ) of the isotope dilution mass spectrometry-related methods was the ratio between ME_α and ME_β, which generally reflects the estimated degree of ME. ME_α, ME_β and θ can be calculated from the following (1), (2), (3), respectively.

$${\mathit{ME}}_{\alpha }=\frac{A_M}{A_S}$$ (1)

$${\mathit{ME}}_{\beta }=\frac{A_{M,\mathit{IS}}}{A_{S,\mathit{IS}}}$$ (2)

$$\theta =\frac{{\mathit{ME}}_{\alpha }}{{\mathit{ME}}_{\beta }}$$ (3)

where A_M, A_M,IS refers to the peak area of acrylamide and ¹³C₃-acrylamide in the matrix-matched calibrants, and A_S, A_S,IS refers to the peak of acrylamide and ¹³C₃-acrylamide in the standard calibrants.

2.7. Method validation

The validation of the proposed method was evaluated by assessing its performance parameters such as linearity, recovery, reproducibility, limit of detection (LOD) and limit of quantification (LOQ). The linearity was evaluated using the standard solutions with concentrations ranging from 0.5 to 100 μg/kg. The standard solutions were prepared by diluting the acrylamide stock solution, and the concentration of the internal standard in the standard solutions was set to 5 μg/kg. The LOD is defined as the lowest detectable concentration of acrylamide, which was calculated on the basis of signal-to-noise ratio at 3. Moreover, the recovery of the method was estimated at three different spiked levels, including 9, 40, 90 μg/kg. The inter-day precision (reproducibility) was evaluated using the relative standard deviation (RSD, %). In addition, the proposed method was also verified using the commercial acrylamide certified reference material (CRM, No. 108–02-006, KRISS).

2.8. Statistical analysis

In the study, the mass fraction of acrylamide in infant formula was quantified using single-point isotope dilution mass spectrometry (SP-IDMS). According to a previous report (Wang, Li, et al., 2022), the mass fraction of acrylamide can be calculated on the basis of the following eq. (4).

$$M=\frac{A_1/A_1^{\prime }}{A_2/A_2^{\prime }}\times \frac{{M_1}^{\prime }}{{M_2}^{\prime }}\times \frac{M_2}{M_S}$$ (4)

where M is the mass fraction of acrylamide in the infant formula sample (μg/kg), A₁ is the peak area of acrylamide measured in the infant formula sample, A₁^’ is the peak area of ¹³C₃-acrylamide measured in the infant formula sample, A₂ is the peak area of acrylamide measured in the standard solution, A₂^’ is the peak area of ¹³C₃-acrylamide measured in the standard solution, M₁^’ is the mass of ¹³C₃-acrylamide added into the infant formula sample (μg), M₂^’ is the mass of ¹³C₃-acrylamide added into the standard solution (μg), M₂ is the mass of acrylamide in the standard solution (μg), M_s is the mass of the infant formula sample (kg).

Moreover, five replicates were carried out in each experiment, including the pretreatment process of modified QuEChERS, method validation, and the quantification of acrylamide in real samples. The statistical tests were implemented for the whole data analysis, and the outliers in the data were checked by Dixon statistical method. The final detection values are expressed by the average value with the corresponding relative standard deviation.

3. Results and discussion

3.1. Optimization of LC-MS/MS parameters

The parameters of mass spectrometry and chromatography were optimized to achieve high sensitivity for acrylamide detection. First of all, each standard solution was analyzed using an infusion experiment, and a full mass scan was carried out in the range m/z of 20–100. The precursor ion was subsequently fragmented in a daughter ion scan, yielding the corresponding product ions. For acrylamide, the identified precursor ion was [M + H]⁺ an m/z value of 72.0. Its quantitative fragment ion had an m/z value of 55.0 ([(H₂C = CH-C=O)] ⁺), stemming from the loss of the NH₃ group from the protonated acryl amide (Kumari et al., 2023; Riediker & Stadler, 2003; Sun et al., 2022). Moreover, the ion with an m/z value of 44.1 was chosen as the qualitative ion. For ¹³C₃-acrylamide, the precursor ion exhibited an m/z value of 75.4, and ions with m/z of 58.1 and 45.0 were selected as the product ions (Kumari et al., 2023).

Acrylamide is well known to co-elute with polar interfering compounds present in food matrices, potentially hampering the resolution and sensitivity of liquid chromatography (Kim et al., 2015; Kumari et al., 2023). To identify an optimal column for the separation of acrylamide, we evaluated peak shape, separation efficiency, and resolution in three distinct columns: HSS T3, BEH amide, and Atlantis dC18 (Fig. 1a). Given acrylamide's high polarity and low molecular weight, mixtures of ACN/water and MeOH/water were considered mobile phases (Kumari et al., 2023; Wang, Cai, et al., 2022). Preliminary experiments revealed that acrylamide's retention time in the ACN phase was marginally shorter than in the MeOH phase. Furthermore, the response of acrylamide in the MeOH mobile phase was double that in the ACN phase, corroborating a previous report (Sun et al., 2022). Thus, subsequent trials employed MeOH as the mobile phase. Although incorporating organic acid into the mobile phase has been reported to enhance the ionization performance and stability of acrylamide during LC-MS/MS analysis (Kumari et al., 2023; Roach et al., 2003), our study observed a marked decrease in response upon adding 0.1 % FA to the MeOH/water mobile phase. Consequently, a mixture of MeOH/water was chosen as the mobile phase for acrylamide analysis. Prior research suggests that the HSS T3 column is tailored for the separation of polar small molecules, while the BEH amide column is also apt for polar compounds (Bertuzzi et al., 2020; Jung et al., 2021). In our investigation, acrylamide was eluted in 1.7 and 2.2 min using the BEH amide and HSS T3 columns, respectively. This suggests potential significant matrix interference due to the brief retention time. To improve the retention time of acrylamide, further chromatographic separation was conducted using a column with larger particle sizes and inner diameter. The Atlantis dC18 column (150 mm × 3.0 mm, 5 μm) yielded a retention time of 4.0 min for acrylamide at an optimal flow rate of 0.4 mL/min. Consequently, the Atlantis dC18 column was selected for use in this study.

Fig. 1.

The corresponding MRM chromatography of acrylamide obtained using different columns, including BEH amide, HSS T3 and Atlantis dC18 (a); the recovery of acrylamide, detected under various conditions, including volumes of the extraction solvent (b), amounts of NaCl (c) and MgSO₄ (d).

3.2. Optimization of QuEChERS method

The QuEChERS technique was utilized for the pretreatment of infant formula matrices, with a detailed optimization of extraction solvent, sorbents, partition salts, and procedural blank (Kumari et al., 2023; Perestrelo et al., 2019). The process of liquid-liquid extraction requires an appropriate solvent. Typically, ACN is employed as the sample pretreatment solvent due to its suitable polarity and high extraction yields for polar and medium polar compounds (Yang et al., 2022). Furthermore, ACN effectively removes lipid-soluble compounds from various matrices (Kumari et al., 2023). The extraction yield of analytes from food matrices can be significantly enhanced by incorporating a suitable organic acid into the extraction solvents (Kumari et al., 2023). In this study, ACN and ACN containing 0.1 % FA were evaluated. The results showed that the recovery at 40 μg/kg was approximately 45 % for both extraction solvents. Additionally, the relative standard deviation (RSD) decreased from 5.0 % to 3.1 % upon the addition of 0.1 % FA to ACN. These changes indicated improved precision and good reproducibility when using ACN containing 0.1 % FA. Consequently, ACN containing 0.1 % FA was selected as the extraction solvent for acrylamide extraction from infant formula matrices.

Infant formula is a complex matrix that contains a certain quantity of fat, which could potentially interfere with the detection of acrylamide. Typically, this fat is miscible with n-hexane, offering a straightforward method for removing less polar, lipophilic matrix co-extractives (Li et al., 2019; Li et al., 2023). In this study, 5 mL of n-hexane was used as the upper extraction buffer. Consequently, a mixture of acetonitrile (ACN) containing 0.1 % formic acid (FA) and n-hexane was employed for defatting. Additionally, the study examined various volumes of ACN containing 0.1 % FA extraction solutions, specifically 10, 15, and 20 mL. As illustrated in Fig. 1b, the corresponding recoveries were 54 %, 63 %, and 63 %, respectively. Therefore, the optimal volume of extraction solvent (ACN containing 0.1 % FA) was determined to be 15 mL.

NaCl and anhydrous MgSO₄ were also involved in the extraction step. The function of anhydrous MgSO₄ has been considered to adsorb water contained in the solvents, and NaCl can accelerate the phase separation of the solvents. (Gao et al., 2021; Santana-Mayor et al., 2019). The addition of MgSO₄ and NaCl promoted the separation of the ACN-water phase, and the partition of the analyte was dispersed in the ACN layer (Gao et al., 2021). Therefore, the amounts of anhydrous MgSO₄ and NaCl were optimized accordingly. Firstly, the amount of MgSO₄ was set to 4.0 g, and the dependence of recovery on the amount of NaCl (from 0 to 3.0 g) is shown in Fig. 1c. The recovery presented the highest value of 71 % when the amount of NaCl was 1.0 g. Subsequently, the amount of anhydrous MgSO₄ was optimized with the amount of NaCl of 1.0 g. The corresponding acrylamide response is shown in Fig. 1d. The maximum value of recovery reached 80 % when 3.0 g of anhydrous MgSO₄ was used. Therefore, 1.00 g of NaCl and 3.00 g of anhydrous MgSO₄ were used in the salting-out extraction process.

3.3. Optimization of sorbent in dSPE unit

The sorbents employed in the dSPE unit, such as PSA, C18, and basic Al₂O₃, are typically used in the QuEChERS step to determine acrylamide in various food matrices (Kumari et al., 2023; Omar et al., 2015； Sebastià et al., 2023). In this study, four sorbents—basic Al₂O₃, PSA/C18, PSA, and aminopropyl—were optimized. Fig. 2a illustrates that the recovery was only around 60 % when PSA and aminopropyl sorbents were utilized. Conversely, the recovery reached approximately 80 % with the basic Al₂O₃ or PSA/C18 sorbent. Additionally, the matrix effects (MEs) of these four sorbents were assessed. Fig. 2a demonstrates that the highest ME value of 0.81 for acrylamide was observed following the clean-up process using basic Al₂O₃, suggesting its effectiveness in eliminating matrix interference. This is likely attributed to the food matrices containing amino acids and short-chain fatty acids, which are adsorbed by basic Al₂O₃ (Omar et al., 2015). On the other hand, the ME of acrylamide over the other three sorbents was significantly lower, averaging 0.60, indicating a substantial ion suppression effect across these sorbents (Matuszewski et al., 2003; Wang, Li, et al., 2022).

Fig. 2.

(a) Effect of the sorbents on the recovery and matrix effect of acrylamide in the QuEChERS pretreatment process, (b) the MRM chromatography of the quantifier transitions for acrylamide in the procedural blank solvents, treated with basic Al₂O₃ and PSA sorbents, (c) the relation between recovery of acrylamide and the amount of Al₂O₃, and (d) the MRM chromatography of acrylamide, which were obtained using the different pretreatment methods, including QuEChERS, LLE-SPE and LLE.

The procedural blank can contribute to an overestimation of acrylamide during LC-MS/MS analysis (Desmarchelier et al., 2020). An examination of the procedural blank was conducted using both PSA and basic Al₂O₃ sorbents (Fig. 2b). Notably, no blank interference was observed at the retention time (RT) of 4.0 min with the basic Al₂O₃ sorbent. However, a peak was detected at the 4.0 min RT with the PSA sorbent, which coincided exactly with that of acrylamide. Subsequent identification of this peak using UHPLC QE plus Orbitrap MS revealed a precursor ion with an m/z value of 72.04431 and a fragment ion with an m/z value of 55.01801, suggesting the compound was acrylamide with a mass deviation of less than 5 ppm. Thus, the peak at 4.0 min was attributed to acrylamide in samples treated with the PSA sorbent. The use of PSA sorbent in the clean-up stage will lead to an overestimation of acrylamide mass content. In comparison to other sorbents, basic Al₂O₃ exhibited high recovery, minimal matrix effect (ME), and a negligible procedural blank, making it the most suitable sorbent for the dSPE process. The optimal amount of basic Al₂O₃ for the clean-up procedure was determined to be 200 mg, at which point the recovery rate reached its peak of 81 % (Fig. 2c). Therefore, 200 mg of basic Al₂O₃ sorbent was used in the clean-up procedure.

3.4. Evaluation of different pretreatment steps

The results discussed above confirm the suitability of the proposed QuEChERS method for quantifying acrylamide in infant formula. However, the interference from food matrices can lead to a decrease in signal and disrupt the selectivity of the analytes (Crawford & Wang, 2019). Consequently, this study conducted three pretreatment methods on infant formula samples: liquid-liquid extraction (LLE), LLE combined with solid phase extraction (SPE), and the proposed QuEChERS method (Fig. 2d). The extraction step of the LLE pretreatment closely mirrored that of the QuEChERS method, except without the d-SPE step. It's crucial to note that the LLE was coupled with a subsequent SPE, HLB, and Bond Elut-AccuCAT, as per the Chinese national standard (National Health and Family Planning Commission of the people's Republic of China (NHFPC), 2014). When the LLE method was used to pretreat the infant formula sample without any cleanup step, it resulted in an unstable baseline in the liquid chromatography and a significant interference peak at a retention time (RT) of approximately 2.1–2.7 min (Fig. 2d). This large peak indicated the low efficiency of matrix removal. Even when the LLE was combined with the successive SPE techniques, SPE-HLB and SPE-Bond Elut-AccuCAT, a substantial peak still appeared at an RT of 2.1 min, suggesting serious interference with the sensitivity of acrylamide detection. However, the application of the QuEChERS step led to a significant reduction in the peak at an RT of 2.1 min. These results confirmed that the co-extraction of acrylamide was effectively restrained using the QuEChERS-based approach. Therefore, the proposed QuEChERS method allows for both satisfactory separation and enhanced sensitivity of acrylamide detection in infant formula matrices.

3.5. Method validation

The performance of the proposed approach was subsequently confirmed through the evaluation of multiple performance metrics, including ME, linearity, LOD, LOQ, recovery, and precision. The corresponding parameter values are detailed in Table 1. Five replicates (n = 5) were executed for each experiment. The ME values obtained for acrylamide and ¹³C₃-acrylamide were 0.811 and 0.807 respectively, leading to an associated θ value of 1.00. This suggests that the matrix effect was effectively mitigated using the isotopically labeled internal standard method. Furthermore, an impressive linearity was observed for acrylamide within the range of 0.50 to 100.00 μg/kg, with an R² value exceeding 0.9999. The LOD and LOQ of the proposed method were found to be 0.30 and 0.90 μg/kg, respectively. Recovery assessments, conducted by spiking the acrylamide at concentrations of 9.00, 40.0, and 90.0 μg/kg in the blank matrix, revealed a relative recovery varying from 103 % to 105 %, with an RSD of less than 1.6 % (Table 1). Additionally, reproducibility assessments yielded an RSD lower than 2.5 %, reinforcing the precision of the proposed approach for detecting acrylamide. The method's validity was further corroborated using commercial matrix-certified reference material (CRM, No. 108–02-006). The measured value of 53.4 ± 1.9 μg/kg obtained via the proposed QuEChERS approach closely aligned with the certified value of this CRM (55.7 ± 2.1 μg/kg), thereby confirming the high accuracy of acrylamide quantification in infant formula using the proposed QuEChERS method.

Table 1.

The linearity, LOD, LOQ, precision and accuracy of the as-developed QuEChERS-HPLC-MS/MS method.

Parameters	Results			Parameters	Results
Matrix effect	acrylamide	13C3-acrylamide	θ	LOD	0.30 μg/kg
	0.811	0.807	1.00	LOQ	0.90 μg/kg
Recovery	spiked value	Mean value	RSD	Linearity	y = 0.4532x - 0.0091	R2>0.9999
	9.00 μg/kg	105 %	1.6 %	Reproducibility	Mean value	RSD
	40.0 μg/kg	103 %	1.3 %		18.5 μg/kg	2.5 %
	90.0 μg/kg	105 %	1.4 %		52.4 μg/kg	2.3 %

Compared with previous reports, the present study provided an efficient and reliable sample preparation technique (Table 2). Good LOD and LOQ for the detection of acrylamide in infant formula have been achieved. Traditional LLE-SPE and DLLME pretreatment methods were also used for the detection of acrylamide in infant formula matrices. However, recoveries of over 85 % with RSD less than 5.1 % were obtained in their studies. The proposed QuEChERS method possessed a recovery of 103 %–105 % and RSD < 1.6 %. Moreover, as summarized in Table 2, compared with QuEChERS methods used for the detection of acrylamide in other food matrices, including potato, heat-processed foods and dark chocolate, the proposed QuEChERS method also demonstrated both good sensitivity and accuracy (Cheng et al., 2021; Gao et al., 2021; Huang et al., 2019; Kruszewski & Obiedziński, 2020; Kumari et al., 2023). Consequently, the proposed method is suitable for detecting trace amounts of acrylamide in infant formula matrices because it could substantially reduce the sample-matrix interference.

Table 2.

Comparison of different methods for determination of acrylamide in various food matrices.

Sample matrix	Extraction methods	Clean up Sorbent	LOD (μg/kg)	LOQ (μg/kg)	Recovery (%)	RSD (%)	Reference
Infant formula	DLLME	Carrez reagent	0.60	1.98	>85	2.9	Ghiasi et al., 2022
Infant powder milk Baby food Baby food Cereal-based infant food Baked and fried food	LLE-SPE LPE-dSPE LPE + dSPE+SPE QuEChERS LLE + dSPE	HLB Al2O3 PSA + SCX PSA + C18 PSA	0.30 5.00 10.0 2.71 0.47	1.00 20.0 20.0 10.0 1.52	86–97 100–108 107–110 101–103 97–112	2.2–5.1 7.0–10.0 2.5–8 2.7–3.6 3.4–6.1	Jiao et al., 2005 Prata et al., 2023 Petrarca et al., 2017 Dekić et al., 2024 Cheng et al., 2021
Potato cutlet	QuEChERS	PSA + C18	0.70	2.00	91–109	1.8–10.6	Kumari et al., 2023
Heat-Processed foods	QuEChERS	EMR-lipid	2.50	12.5	89–103	1.9–6.8	Huang et al., 2019
Heat-Processed foods	QuEChERS	EMR-lipid	10.0	30.0	97–105	4.4–8.4	Gao et al., 2021
Dark chocolate	QuEChERS	PSA + C18	6.70	20.3	78–101	2.0–4.5	Kruszewski et al, 2020
Infant formula	QuEChERS	Basic Al2O3	0.30	0.90	103–105	1.3–1.6	This study

Furthermore, the proposed QuEChERS approach could be applied to establish the correlation between acrylamide levels and formula categories, predict the formation mechanisms of acrylamide during the production process, and monitor the levels of acrylamide under different storage conditions. Besides infant formula, the proposed approach could also be suitable for other dairy products.

3.6. Identification and quantification of acrylamide in real samples

To assess the universality of the developed method, we analyzed acrylamide concentrations in 20 real samples. These samples were purchased from supermarkets in Beijing. The findings are summarized in Table 3, and the associated chromatograms are presented in Fig. 3. Out of the 20 samples, acrylamide mass fractions in 15 were below the LOQ, suggesting minimal acrylamide presence. The concentrations for the remaining five samples—S3, S7, S14, S19, and S20—were found to be 12.38, 8.70, 31.18, 65.61 and 53.70 μg/kg, respectively. Notably, the acrylamide levels in samples S19 and S20 surpassed the EU's minimum benchmark for acrylamide in infant and children's products, set at 40 μg/kg. This suggests potential health risks associated with these two samples.

Table 3.

The measured values with the corresponding SD of acrylamide (μg/kg) contained in the commercial real samples, detected using the proposed approach.

Number of samples#	Acrylamide (μg/kg)	Number of samples	Acrylamide (μg/kg)
S1	<LOQ	S11	<LOQ
S2	<LOQ	S12	<LOD
S3	12.38 ± 0.54	S13	<LOD
S4	<LOQ	S14	31.18 ± 0.71
S5	<LOQ	S15	<LOQ
S6	<LOQ	S16	<LOQ
S7	8.70 ± 0.73	S17	<LOQ
S8	<LOQ	S18	<LOQ
S9	<LOQ	S19	65.61 ± 0.37
S10	<LOQ	S20	53.70 ± 0.95

^#S1, S3, S14, S15, S19: five brands of whole milk powder; S2: donkey milk powder; S4, S5, S6: three brands of skim milk powder; S7: goat milk powder; S8: camel milk powder; S9, S10, S11, S12, S13, S16, S17, S18, S20: nine brands of infant formula for three age groups.

Fig. 3.

The typical chromatography of acrylamide contained in real samples obtained from supermarket, detected using proposed QuEChERS-HPLC-MS/MS approach.

4. Conclusion

In conclusion, we established a modified QuEChERS pretreatment technique combined with a stable isotope dilution LC-MS/MS approach for the analysis of acrylamide in infant formula samples. This method effectively eliminates interference in complex matrices through the simple QuEChERS pretreatment step. As a result, the proposed approach offers several advantages: a LOQ of 0.90 μg/kg, an LOD of 0.30 μg/kg, an RSD ranging from 1.3 to 2.5, and a recovery rate of 103 %–105 %. Importantly, the reliability of this method was confirmed using a commercial matrix CRM (No. 108–02-006). The application of the proposed method to real infant formulas revealed that contamination by acrylamide is extremely rare. This cost-effective QuEChERS method allows for high-throughput analyses of acrylamide in food testing laboratories, providing essential technical support for regulatory implementation.

CRediT authorship contribution statement

Jiahui Ma: Validation, Formal analysis, Writing – original draft, Investigation. Xiaomin Li: Resources, Writing – review & editing, Conceptualization, Supervision, Methodology. Shuangxia Luo: Validation, Formal analysis. Xianjiang Li: Writing – review & editing. Qinghe Zhang: Resources, Supervision. Xiong Yin: Resources, Writing – review & editing, Supervision.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Data availability

Data will be made available on request.