Acrylamide Determination in Infant Formulas: A New Extraction Method Authors: Sumeyra Sevim, Rosalia Lopez-Ruiz, Antonia Garrido-Frenich, Angelo Antonio D’Archivio Published: 2025 ABSTRACT AbstractInfant formulas are specialized foods designed for babies and toddlers who cannot be exclusively breastfed. However, acrylamide (AA) may form during the thermal processing involved in their production. Although chromatographic techniques offer high sensitivity and detection capability for AA analysis, their application remains limited due to the complexity of diverse food matrices, high operating costs, time requirements, and environmental concerns. A new validated liquid chromatography–mass spectrometry (LC-MS) protocol for AA detection in infant formula was developed using sequential hydration, acetonitrile (ACN) precipitation, and dual-sorbent clean-up, which minimized matrix effects and ensured clarity and high reproducibility. The validated method demonstrated excellent linearity (R2 = 0.9985, solvent-based; 0.9903, matrix-based), a pronounced matrix effect (−67%), satisfactory sensitivity (limit of detection, LOD: 10 µg/kg; limit of quantification, LOQ: 20 µg/kg), and consistent recovery (82–99%) with less than 15% variation. AA analysis was performed on 31 infant formula samples. The highest individual AA level (268.2 µg/kg) was detected in an amino acid-based formula intended for infants under one year of age while the highest mean concentration was found in cereal-based samples (188.1 ± 100.8 µg/kg), followed by goat’s milk-based (52.7 ± 25.67), plant-based (48.8 ± 31.68), and cow’s milk-based (27.5 ± 29.62) formulas (p < 0.001). The wide variability in AA concentrations among infant formulas can be attributed to differences in formulation, ingredient composition, manufacturing processes, and analytical methodologies. These findings highlight the need for continuous monitoring of AA levels in infant foods to ensure their safety.Keywords: acrylamide, infant formula, LC, sample preparation, MS, milk 1. Introduction Commercially available infant formulas used as partial or complete substitutes for breast milk are consumed worldwide to feed infants and young children aged 0–36 months [1]. There are two main forms of infant formula: powdered, which must be reconstituted with water, and ready-to-feed liquid. Among these, powdered formula is the most widely available, shelf-stable, and commercially distributed type [2]. As infant formulas play a critical role during the first months of life, they must be free from potentially hazardous toxic compounds from a food safety perspective [3]. Infant formulas are manufactured using a wide range of processing techniques, incorporating both thermal and non-thermal technologies, which result in distinct final product compositions [2]. Conventional thermal processes are essential for producing powdered products and for achieving pasteurization or sterilization effects. These methods effectively reduce microbial loads and extend product shelf life; however, maintaining an appropriate balance between microbial safety and the prevention of undesirable chemical changes is crucial [4,5]. Moreover, thermal treatments have a pronounced impact on the physicochemical properties of infant formulas due to complex interactions among proteins, carbohydrates, and lipids. During production, the Maillard reaction can occur when protein- and sugar-containing ingredients are exposed to heat, leading to the formation of harmful compounds such as acrylamide (AA) [6]. AA is an α,β-unsaturated carbonyl compound formed in certain foods processed at temperatures above 120 °C through the reaction between amino acids and reducing sugars such as glucose and fructose [7]. According to the International Agency for Research on Cancer (IARC), AA is classified as a Group 2A compound, meaning it is “probably carcinogenic to humans” [8]. Over the past two decades, studies have shown that dietary exposure to AA can cause carcinogenic, neurotoxic, genotoxic, and reproductive or developmental effects in both humans and animals [9,10,11,12]. Reflecting its recognized carcinogenic potential, AA was listed among the high-priority agents for re-evaluation in the IARC Monographs Programme for 2020–2024 [13]. AA in food poses a particularly high risk to infants and young children due to their specific physiological characteristics, including higher dietary intake per kilogram of body weight, elevated resting metabolic and respiratory rates, and a larger body surface area relative to adults [3]. The European Commission has established “indicative values” for AA in various food categories based on monitoring data from Member States between 2007 and 2011. Although these values are not safety thresholds, exceeding them requires manufacturers to take corrective measures to limit AA formation. In the baby food category, reference levels of 50 µg/kg for non-cereal-based products and 80 µg/kg for foods containing plums have been set [14]. A recent meta-analysis reported that the mean AA content in baby foods (156.3 µg/kg) exceeded the reference limits (40–150 µg/kg), which is concerning given children’s greater susceptibility compared to adults [15]. “Baby foods other than processed cereal-based” were identified as the largest contributors to infants’ total AA exposure according to European data, with infants estimated to ingest three times more AA per day than adults [16]. In recent years, liquid and gas chromatographic techniques (LC-MS/MS and GC-MS) have become the standard analytical approaches for AA determination due to their high sensitivity and detection capability. Several extraction and clean-up methods, such as solid–liquid extraction (SLE), solid-phase extraction (SPE), and QuEChERS, have been employed to isolate AA from food matrices. However, the application of these techniques remains limited by the complexity of food matrices, high operational costs, time-consuming procedures, and environmental impact [9,17,18,19]. Therefore, there is a growing need for simple, robust, and broadly applicable analytical methods compatible with widely available detection systems. To address these limitations, this study proposes a simplified extraction strategy designed to reduce sample-handling steps and matrix interferences, particularly for powered infant formula matrices. In this context, the objective of the study was to establish a sensitive and selective analytical approach for AA determination in powdered infant formulas. For this purpose, an optimized SLE-based extraction combined with a dispersive solid-phase extraction (d-SPE) clean-up and liquid chromatography–triple quadrupole mass spectrometry (LC-QqQ-MS) detection was proposed. 2. Results and Discussion 2.1. Optimization and Validation of the Extraction and LC-MS Method The MS characterization and chromatographic separation were previously optimized following the procedure described by López-Ruiz et al. [18]. The extraction procedure was further refined based on a previously developed method that employed SLE with water and a clean-up step using C18 and Primary Secondary Amine (PSA) sorbents. In that protocol, an infant formula sample (1 g) was hydrated with 5 mL of water and directly subjected to purification. However, this approach exhibited significant drawbacks when applied to powdered infant formula, as reliable AA detection could not be achieved under these conditions. Pronounced matrix effects were observed, mainly due to the high protein and lipid content of the matrix. Moreover, filtration of the extracts was problematic, as they could not be readily prepared for LC-MS injection. To address these issues, several extraction solvents and mixtures such as dichloroethane, ACN/water/formic acid, and ACN/water were tested. None of these conditions yielded satisfactory results: AA was either undetectable or exhibited inconsistent recoveries (below 45%). These unsuccessful attempts highlighted the challenge of efficiently extracting AA from infant formulas without incorporating an effective protein precipitation step. To overcome this limitation, ACN was introduced following an initial hydration phase with water. This modification improved protein precipitation, reduced matrix complexity, and enhanced AA recovery to 85%. For method validation, the parameters established in the SANTE/11312/2021 guidelines [20], were evaluated, including linearity, matrix effect (ME), limits of quantification (LOQ) and detection (LOD), as well as trueness and precision (intra- and inter-day). Linearity and matrix effects were assessed using calibration curves prepared both in solvent (ACN/water) and in matrix-matched extracts at concentrations of 20 (LOQ), 50, 100, 200, 300, 400 and 600 µg/kg. Relative peak areas were calculated as the ratio of the AA peak area to that of the internal standard (AA-d3). The determination coefficients (R2) demonstrated excellent linearity, with values of 0.9985 for the solvent-based calibration curve and 0.9903 for the matrix-based curve. The matrix effect, calculated according to Equation (1), indicated strong signal suppression for AA (−67%). Considering the strong matrix effect observed, AA concentration in all samples were quantified using the matrix- based curves to ensure accurate quantification. The LOQ was defined as the lowest concentration providing acceptable recovery (70–120%) and precision (RSD < 20%), which was established at 20 µg/kg for AA. The LOD, determined as the lowest concentration yielding a signal-to-noise (S/N) ratio below 3, was set at 10 µg/kg. Trueness, expressed as recovery (%), and precision, expressed as relative standard deviation (RSD), were evaluated at two concentration levels and 5 replicates each level: One sample of each type of main ingredient (as representative groups) were prepared according to Section 3.3, were spiked with the analytical standard of AA at the LOQ (20 µg/kg) and 250 µg/kg. Recovery rates were 82% at the LOQ and 99% at 250 µg/kg, while intra- (same day) and inter-day (different days n = 4) precision values remained below 15% at both levels. 2.2. Samples Analysis The analyzed infant formula samples were classified according to age group, main ingredient, and formula type to ensure representativeness across different product categories, and this distribution is shown in Table 1. Detailed information regarding the 31 infant formula samples analyzed, including age category, main ingredient (e.g., cow’s milk-based), formula type, brands codes (B1–12), primary protein and carbohydrate source and the AA concentration found in each sample, is provided in Supplementary Table S1. Among the analyzed products, most (n = 14; 45.2%) were intended for infants aged 0–6 months, followed by formulas for infants < 1 year (n = 8; 25.8%), >6 months (n = 3; 9.7%), >1 year (n = 3; 9.7%), 0–36 months (n = 2; 6.5%), and 6–12 months (n = 1; 3.2%). Based on the main ingredient, cow’s milk-based formulas represented the majority (n = 21; 67.7%), followed by goat’s milk-based (n = 5; 16.1%), plant-based (n = 2; 6.5%), cereal-based (n = 2; 6.5%), and amino-acid based (n = 1; 3.2%) formulations. In terms of functional classification, products were categorized as either standard formulas (n = 18; 58.1%) or therapeutic/special formulas (n = 13; 41.9%) according to product labeling. Table 1Summary of infant formula samples (n = 31) grouped by age category, main ingredient, and formula type.Age GroupMain IngredientFormula TypeSample (Brand Code)0–6 monthsCow’s milk-basedStandard formula9 (B1, B10, B2, B3, B7, B8) Cow’s milk-basedTherapeutic/Special2 (B2, B4) Goat’s milk-basedStandard formula3 (B12, B6, B9)6–12 monthsCow’s milk-basedStandard formula1 (B2)<1 yearCow’s milk-basedTherapeutic/Special5 (B1) Plant-basedTherapeutic/Special2 (B1, B2) Amino-acid basedTherapeutic/Special1 (B1)>6 monthsCereal-basedStandard formula1 (B1) Cow’s milk-basedTherapeutic/Special1 (B4) Goat’s milk-basedStandard formula1 (B12)>1 yearCow’s milk-basedStandard formula2 (B2, B5) Goat’s milk-basedStandard formula1 (B11)0–36 monthsCereal-basedTherapeutic/Special1 (B4) Cow’s milk-basedTherapeutic/Special1 (B2) Summary of infant formula samples (n = 31) grouped by age category, main ingredient, and formula type. Table 2 structure allows comparison of AA concentration levels across different brands. The measured AA concentrations in the analyzed infant formulas showed substantial variability across age groups and main ingredients (Table 2). The highest mean AA concentration was observed in the amino-acid-based formula of brand B1 intended for infants under one year of age (mean = 268.2 µg/kg), followed by the same brand’s cereal-based formula for infants older than six months (mean = 259.4 µg/kg) and the cereal-based therapeutic formula of brand B4 for infants aged 0–36 months (mean = 116.8 µg/kg). In contrast, AA levels were considerably lower in cow’s milk- and plant-based formulas produced by the same brand (B1) for different age groups. Table 2AA concentrations in infant formulas by age group, main ingredient, and brand (μg/kg).Age GroupMain IngredientBrand CodenMeanMinMax0–6 monthsCow’s milk-basedB1211.56 monthsCereal-basedB11259.4259.4259.4 Cow’s milk-basedB4124.024.024.0 Goat’s milk-basedB12180.080.080.0 Mean 3121.124.0259.4>1 yearCow’s milk-basedB210.0 0.05). Accordingly, cow’s milk-based formulas showed the lowest average AA levels (0–77.5 µg/kg; mean = 27.5 ± 29.6 µg/kg), while goat milk-based formulas showed approximately double the concentration (23.7–80.0 µg/kg; mean = 52.7 ± 25.7 µg/kg), but there was no statistically significant difference. Plant-based formulas displayed moderate AA levels (26.4–71.2 µg/kg; mean = 48.8 ± 31.7 µg/kg), whereas cereal-based formulas contained significantly higher concentrations (116.8–259.4 µg/kg; mean = 188.1 ± 100.8 µg/kg) compared to the other groups (p < 0.001). Figure 1 shows an extracted ion chromatogram (EIC) for AA detected in infant formulas derived from different main ingredients. Extracted ion chromatogram of AA in infant formula samples. (a) goat milk (Sample 12); (b) oatmeal (Sample 24); (c) soybean (Sample 23); (d) Rice (Sample 30). According to the European Food Safety Authority [14], AA concentrations in baby foods and infant formulas vary significantly depending on the product type and ingredients. The mean AA concentration reported in non-cereal-based infant formulas was 24 µg/kg. Notably, formulas containing prunes showed much higher mean levels (101 µg/kg) compared with those without prunes (20 µg/kg), primarily due to the elevated sugar and free asparagine content in dried fruits, which are key precursors for AA formation. AA concentrations in infant formulas vary widely worldwide due to differences in formulation and composition, production and processing parameters, and analytical detection methods [3]. A study conducted in Iran reported extremely high AA levels in infant formulas, ranging from 48 to 5835 µg/kg. The authors concluded that among all factors, protein composition had the greatest impact on AA formation in powdered formulas, while both formulation and ingredient composition also showed significant effects [17]. Similar findings were observed in infant formula powders analyzed in Colombia [21]. The results of this study also support these findings. In the two samples with the highest AA content, the protein source consists of free amino acids and the carbohydrate source consists of glucose syrup. The presence of free amino acids and reducing sugars, which increase acrylamide formation in the Maillard reaction, explains the high AA levels observed in these products. Conversely, French “Follow-on formula” and “Infant formula” products contained relatively low AA levels (0.14/2.2 µg/kg for the lower-bound assumption and 0.60/2.9 µg/kg for the upper-bound assumption) compared with previous studies [22]. More recently, a study conducted in Türkiye reported AA levels ranging from 12 months. The highest level (251 µg/kg) was detected in a brand 4 formula intended for children > 12 months [23]. 2.3. Comparison of the Method Used with Alternative Techniques for Determination of AA in Infant Formula Recent advances in analytical methodologies—such as simplified derivatization, SPE, and advanced purification techniques—have substantially improved detection limits and analytical reliability [9]. SPE, using either normal-phase or reversed-phase sorbents, enables further refinement of extracts [24]. This sample preparation technique is widely used in analytical chemistry to concentrate and isolate target compounds from complex matrices prior to analysis. The solid-phase sorbent material selectively retains analytes while removing unwanted matrix components [25]. Due to its versatility and adaptability, SPE is applicable to a broad range of samples, including biological, food, and environmental matrices [26]. Accordingly, the newly developed method in this study incorporates an SPE-based extraction step specifically optimized for infant formula matrices to achieve enhanced sensitivity and reliability. The method demonstrated excellent analytical performance, with a linear range of 20–600 µg/kg (R2 = 0.9905), LOD and LOQ values of 10 μg/kg and 20 μg/kg, respectively, and recovery rates ranging from 70% to 120%. The method precision (RSD) was below 15% across all concentration levels, confirming satisfactory repeatability. The comparison of alternative techniques for AA detection in infant formulas is summarized in Table 3. Lee et al. [27] employed the isotope dilution–liquid chromatography/mass spectrometry (ID–LC/MS) method to develop and validate a robust analytical procedure aimed at establishing a reliable reference material for accurately quantifying AA in infant formula. The ID-LC/MS approach demonstrated excellent accuracy and reliability, achieving low repeatability (<1.1%) and intermediate precision (<1.4%). The combined use of chloroform–methanol (MeOH) extraction and Oasis PRiME SPE clean-up effectively minimized matrix effects, while the use of the 13C3-AA internal standard further enhanced quantification accuracy. However, despite its high analytical performance, this method requires costly instrumentation and consumables, and its complexity demands highly trained personnel. These factors limit their routine application in standard analytical laboratories [27]. Basaran and Aydın [23] confirmed that LC-MS/MS is a reliable and accurate technique for AA determination in infant formula. Their validation results and high analytical performance demonstrated that the method is suitable and effective for quantitative analysis of AA in such matrices. Similarly, Ghiasi et al. [17] developed a dispersive liquid–liquid microextraction-based microextraction (ME)-GC/MS method that exhibited excellent performance for AA determination in powdered infant formulas. This approach enabled rapid and efficient extraction, providing high sensitivity with a low detection limit (0.6 ng/g). The method also showed high selectivity, acceptable accuracy and reproducibility (RSD 2.9%), a strong enrichment factor (140), and improved recovery. Its user-friendly design and the use of low-cost GC analysis enhance its practicality. Additionally, the two-step sample preparation process (hydrolysis followed by microextraction) further improved sensitivity, selectivity, and recovery. Mastovska et al. [28] developed a method allowing rapid and simple sample preparation for AA determination in diverse food matrices, which was later effectively applied to infant formulas by Pacetti et al. [21]. The method proved reliable in terms of accuracy and reproducibility, while offering high sample throughput, low operational cost, reduced contamination risk, and the use of an isotopic internal standard (AA-d3). The combination of ACN-based extraction, hexane-assisted oil removal, and MgSO4−NaCl salt addition ensured efficient phase separation, whereas dispersive d-SPE with PSA sorbent effectively reduced matrix effects and removed interfering compounds, particularly beneficial in LC-MS/MS analysis [28]. Table 3 highlights that the current method efficiently eliminates matrix interactions, reduces solvent consumption and shortens sample preparation time compared to previous methods, emphasizing the performance of the analytical method. However, its analytical sensitivity was comparatively lower than that of several previously reported methods, as reflected by the higher LOQ (20 µg/kg) shown in Table 3. Importantly, these detection limits were obtained using a streamlined, time-efficient SLE followed by d-SPE procedure specifically optimized for infant formula matrices. The combined use of PSA and DSC-18 sorbents during the clean-up stage ensured efficient removal of both polar and non-polar matrix components. PSA effectively eliminated polar substances such as organic acids, pigments, and certain sugars, while C18 retained lipids and hydrophobic co-extractives. This workflow reduced matrix interferences, minimized solvent consumption, and shortens sample preparation time. Therefore, the proposed method provides a reliable and effective substitute with acceptable recovery (82–99%) and precision (RSD < 15%) for determining AA in powdered infant formulas by combining analytical robustness with practical applicability. Table 3Evaluation of the alternative techniques for detecting AA in baby formulas.ReferenceExtractionCleanupAnalytical InstrumentResultsLOD & LOQThis study0.75 g sample + 10 μL AA-d35 mL water5 mL ACN100 mg DSC-18100 mg PSALC-QqQ-MS0–268.2 μg/kg in infant formulas10 & 20 µg/kg[27]1 g sample + AA-13C310 mL water10 mL of a chloroform–MeOH (2:1, v:v)Oasis PRiME HLB (6 cc, 200 mg) SPE cartridgesOasis PRiME MCX (6 cc, 150 mg) SPE cartridgesID-LC/MS55.7 ± 2.1 μg/kg in infant formulas0.03 & 0.1 µg/kg[23]1 g sample + 1 mL AA-d39 mL water5 mL n-hexaneFiltration through a 0.45 μm PVDFOasis HLB cartridgesA Bond Elut Accucat SPE cartridgesLC-MS/MS