One-step assisted extraction of patulin—a new LC-MS/MS method for testing apples and apple products Authors: Adrian Olszewski, Grażyna Chwatko Published: 2026 ABSTRACT Patulin is a secondary metabolite produced by various fungi, often identified in apples and apple-based products. It is included in group three of mycotoxins classified by the International Agency for Research on Cancer. Therefore, the development of sensitive and accurate methods for the determination of patulin in food is crucial for consumer safety. In this study, a simple and sensitive LC-MS/MS method was developed for determining patulin concentrations in food products. Foods from local Polish markets were analyzed, and the toxin was not found in any of them. The method uses ethyl acetate for single-step extraction, achieving greater than 90% effective patulin extraction. After extraction, the organic phase was evaporated to dryness, and the residue was reconstituted in a suitable solvent and further diluted. In addition, in accordance with the standards in the regulation, the method was tested based on fortification on each matrix. The tests were performed in two laboratories using proficiency test materials with assigned target values and acceptable ranges as reference materials, demonstrating excellent sensitivity and reproducibility while supporting method performance and verification.Supplementary InformationThe online version contains supplementary material available at 10.1038/s41598-026-56715-y. Introduction Secondary metabolites produced by filamentous fungi, known as mycotoxins, occur naturally in food and animal feed products1–3. One of these mycotoxins is 4-hydroxy-4 H-furo[3,2c]pyran-2(6 H)-one, also known as patulin (PAT). PAT is an unsaturated heterocyclic lactone. This compound can be produced by many different fungal species, including Penicillium, Aspergillus, and Byssochlamys4–6, but mainly by Penicillium patulum and Penicillium expansum. PAT was first isolated in 1941 from the fungus Penicillium patulum. PAT biosynthesis consists of several steps, as demonstrated by comprehensive biochemical studies (see Supplementary data, Figure S1). From these studies, it was discovered that numerous mutations arose at different stages of the PAT synthesis pathway7. Initially, the substance was characterized and found to have antimicrobial properties; for this reason, it was used as a cold remedy8. Shortly thereafter, studies showed that PAT was also characterized by toxic, carcinogenic, and teratogenic effects2,9,10 and was classified by the International Agency for Research on Cancer as group 3, not classifiable as a human carcinogen11. Some acute symptoms such as convulsions, pulmonary congestion, edema, intestinal hemorrhage, and epithelial cell degeneration12. Food spoilage and PAT generation can occur in fruit prior to processing, most commonly caused by the P. expansum fungus. The formation of so-called soft rot, which spreads very rapidly, can consequently lead to the production of characteristic blue spore rings. Because of its appearance, this disease is commonly referred to as “blue rot”13. PAT is mainly found in apples and apple-based products. The mycotoxin can also be found in other fruits, e.g., apricots, pears, grapes, strawberries, blueberries, and peaches10,14–16. However, Poland is one of the largest apple producers in Europe and a world leader, ranking fourth behind China, Turkey, and the USA. Apple yields in Poland are growing steadily. In the period 2021–2022, an exceptionally high harvest of more than 4 million tons was achieved (see Supplementary data, Figure S2)17,18. Thus, PAT concentration determination is a key concern in fresh apples and apple-based products. The presence of PAT in food is a global concern, consequently leading to the creation of international guidelines and regulations for maximum levels in consumer products16. Thus, within the European Union, the European Commission in Commission Regulation (EU) 2023/915 has set the following PAT levels: fruit juices and alcoholic beverages made from apples or juice can contain up to 50 µg/kg, solid apple products including apple compote and apple puree can contain up to 25 µg/kg, while foods made from apples for infants and young children can contain up to 10 µg/kg. The US FDA has set maximum levels for PAT in fruit juices at 50 µg/kg19. Due to its toxicological relevance and its occurrence in food at low concentrations, sensitive and selective analytical methods are required for accurate detection and quantification of PAT.Various sample purification methods have been described in the literature, but most are still based on traditional liquid-liquid extraction, QuEChERS-type extraction, and solid-phase extraction20. Studies on the presence of PAT in food products have used, among others, thin-layer chromatography21–23, high-performance liquid chromatography (LC) coupled to a spectrophotometric detector14,24,25, or a mass spectrometer7,24,26, gas chromatography14,22,23, and capillary electrophoresis27. However, the sensitivity obtained with LC coupled to a tandem mass spectrometer (MS/MS) has made it one of the most advanced techniques for the determination of mycotoxins, including PAT20. As PAT is a very polar compound with a low molecular weight and often low concentration, numerous analytical challenges have been reported in the literature for LC-MS/MS methods4,7,20. Due to these physicochemical properties, PAT analysis requires carefully optimized analytical conditions, as they may result in reduced recovery and/or sensitivity7. In particular, MS detection of PAT, especially in electrospray ionization (ESI), is not straightforward and may present an analytical challenge4. It has been reported that PAT may exhibit relatively low ionization efficiency under both ESI and atmospheric pressure chemical ionization conditions, particularly at low ppb levels or under non-optimized conditions, which may lead to reduced sensitivity13. Furthermore, significant ion suppression is frequently observed when co-extracted matrix components are present when ESI is applied, which may adversely affect validation parameters4,7. Additionally, due to its high polarity, PAT exhibits low retention in reversed-phase liquid chromatography and may co-elute with other polar compounds, resulting in signal suppression and further affecting analytical performance28. Although it is often found when using a multistep extraction process, these procedures are often validated on matrices where PAT is unlikely to be present28–30. At the same time, it should be emphasized that, when properly optimized, LC-MS/MS methods can achieve high sensitivity for PAT determination, as demonstrated in the literature20. Therefore, the challenges associated with PAT analysis are not related to the inadequacy of the technique itself, but rather to the need for careful optimization of ionization conditions and effective mitigation of matrix effects. In addition, it has been noted that sampling itself can pose a problem, mainly affecting solid samples (mycotoxins have a mostly heterogeneous distribution) rather than liquid samples such as fruit juices, which are more homogeneou24. The objective of this study was to develop a sensitive, simple, and reliable method for the determination of PAT using LC-MS/MS with an ESI source, intended for routine monitoring of this compound in apple and apple-based products. Particular emphasis was placed on addressing the analytical gap associated with currently available methodologies. Many established methods, originating from the AOAC procedure (1996)31, rely on multiple extraction steps with ethyl acetate, often requiring two or three successive extractions, pH adjustment, and additional cleanup procedures, which increase both the complexity and duration of the analysis. Alternative approaches, such as salting-out techniques, have also been reported; however, they are often associated with longer and more complex sample preparation workflows. In this context, the main novelty of the proposed method lies in the application of a single-step extraction using ethyl acetate in the presence of ascorbic acid. Ascorbic acid enhances PAT extraction efficiency through a salting-out effect. Increasing the ionic strength of the aqueous phase reduces water activity and decreases the solubility of PAT, thereby shifting the partition equilibrium toward the organic phase. Additionally, reduced hydration of PAT molecules contributes to a more favorable partition coefficient, promoting more efficient transfer into ethyl acetate. As a result, high extraction efficiency (approaching 95–100%, depending on the matrix) can be achieved in a single step. This approach eliminates the need for repeated extractions, pH adjustment, and additional cleanup steps, thereby significantly simplifying and accelerating the sample preparation procedure while minimizing analyte loss. The developed method was designed as a universal analytical tool applicable to a wide range of apple and apple-based products, including infant and young child foods (e.g., purées, fruit-based products, and freeze-dried snacks). This is particularly important, as such products may be introduced into the diet as early as 4–6 months of age, requiring highly sensitive and reliable analytical methods. The method combines simplicity and rapid execution with the potential for miniaturization, making it well-suited for routine laboratory use. Furthermore, method performance and reliability were confirmed through internal validation using reference materials, including materials obtained from proficiency testing schemes with assigned target values and acceptable ranges, as well as through participation in external proficiency testing. By selecting an appropriate sample preparation strategy and optimized chromatographic conditions, matrix effects were effectively minimized, enabling the reliable determination of PAT at low concentrations. Results and discussion This paper presents a method based on salt-assisted extraction. The study used reference material to test the developed method, which is to be applied for daily PAT analysis in accordance with EU standards. To the best of our knowledge, no method has been found that uses ascorbic acid and yields very good results with a single extraction. Sample preparation It is widely recognized that sample preparation is a critical step in the analytical workflow, and errors introduced at this stage cannot be corrected later. The developed method applies not only to liquid samples but also to solid ones, for which preparation is associated with greater variability. Therefore, before analysis, all food samples were homogenized to ensure representativeness and improve the reliability of the analytical results, particularly in heterogeneous matrices such as apples and apple-based products. Homogenization was performed using specialized laboratory mills, resulting in high particle size reduction and excellent sample uniformity. It is well established that the degree of comminution directly affects the rate of mycotoxin release from the matrix at the extraction stage. Finer and more homogeneous particles increase the surface area available for extraction, thereby improving both reproducibility and extraction efficiency. The reference point for our method under development was the method first adopted by the Association of Official Agricultural Chemists, in which the extraction with ethyl acetate is repeated twice, then the fractions are combined, desalted by the addition of sodium carbonate and sodium sulfate, followed by filtration and evaporation to dryness31. The residue was dissolved in a suitable solution. Optimization of sample preparation for analysis began with selecting an extractant additive to increase extraction efficiency. Therefore, the effect of adding salt and acid, i.e., sodium chloride and ascorbic acid, on the yield after the first extraction was checked. For this purpose, 0.15, 0.25, and 0.35 g of salt or acid were added during the extraction of 2.5 g of apple mousse fortified at 10 µg/L. The samples were processed as described in section “Application of the method”, and the results are shown in Fig. 1. Fig. 1Extraction efficiency without and with sodium chloride or ascorbic acid. Extraction efficiency without and with sodium chloride or ascorbic acid. The extraction efficiency after a single extraction of the sample alone with ethyl acetate was 50%, while the addition of sodium chloride or ascorbic acid increased the efficiency. The best results were obtained for a 0.15 g weight of ascorbic acid. Increasing the weight of additives by 0.1 g and 0.2 g resulted in a decrease in extraction efficiency. In the next step, the effect of shaking time on extraction yield was evaluated over 5–30 min. There were no significant differences in extraction yields over the time interval studied (see Supplementary data, Figure S3), and 5 min was selected as the optimal extraction duration. The last factor optimized was the mixture used for dissolving the sample after evaporation. Solubility was tested in water, acetonitrile, methanol, and their mixtures acidified with acetic acid. From a practical point of view, methanol alone proved to be the best because it did not adhere to the Falcon walls after shaking and was compatible with the mobile phase. Acidification of methanol with 0.1% or 1% acetic acid did not affect the PAT assay. On the other hand, diluting the sample with water provides symmetric peaks. Considering this, the residue was dissolved in methanol and finally diluted with water. Chromatographic and MS/MS conditions During the experiments, three chromatographic columns and various mobile phase compositions were evaluated to select the optimal separation conditions for PAT. The following columns were evaluated: ZORBAX Eclipse Plus C18 (1.8 μm, 50 × 2.1 mm; column 1), InfinityLab Poroshell 120 SB-C18 (2.7 μm, 100 × 2.1 mm; column 2), and Reprospher 100 C18-Aqua (1.8 μm, 100 × 2.0 mm; column 3). Various mobile phase compositions were tested, including pure solvents such as water (phase A) and methanol or acetonitrile (phase B), as well as mixtures containing additives such as formic acid, ammonium formate, ammonium fluoride, acetic acid, and ammonium acetate. Columns 1 and 2 were excluded at the initial stage of the study due to suboptimal performance. Column 1 consistently produced tailing peaks, regardless of mobile phase composition and gradient used, making integration difficult and negatively affecting analytical accuracy. Achieving acceptable peak symmetry would have required rapid gradient changes, potentially compromising column longevity and resulting in elevated system pressure. In the case of column 2, PAT determination was only possible at high concentrations, in contrast to column 3, indicating insufficient sensitivity for trace-level analysis. Furthermore, interference from matrix components masked the PAT signal, making identification and reliable quantification impossible. Column 3 proved to be the most suitable for PAT analysis. It yielded symmetrical peaks, and the corresponding peak areas and heights enabled reliable quantification of PAT at low concentrations, as detailed below. The selection of chromatographic conditions was further supported by the structural properties of PAT, which, as a small and polar molecule containing hydroxyl and lactone groups, shows a strong tendency to ionize in negative electrospray ionization mode (ESI−) via the formation of deprotonated molecular ions [M − H]⁻ at m/z 153. The study of mobile phase effects on PAT retention began with a mixture of pure solvents: water and methanol. Subsequently, various additives were added, including ammonium fluoride, acetic acid, and ammonium formate. Improved results were observed using a combination of phase A: 5% methanol in water with 0.125 mmol/L ammonium fluoride and phase B: 5% water in methanol with 0.125 mmol/L ammonium fluoride, but better results were obtained on a phase consisting of 1 mmol/L acetic acid and 2 mmol/L ammonium acetate in water (A) and methanol (B), which is consistent with the reports of Gab-Allah et al.32. Improved separation was attributed to the buffering capacity of ammonium acetate and acetic acid, which maintained a stable pH and ionic environment during elution. The use of mobile phase A (water with ammonium acetate and acetic acid) was motivated by the need to ensure stable and reproducible ionization conditions, as ammonium acetate, being a volatile buffer, stabilizes the ionic environment and supports the formation of [M − H]⁻ ions, while acetic acid enables precise control of mobile phase pH. Mildly acidic conditions help suppress unwanted ion forms while maintaining a stable signal in ESI− mode. Mobile phase B (methanol) was selected for its favorable effect on droplet desolvation in the ion source, leading to more efficient gas-phase ion formation and improved detection sensitivity. To best separate the analyte from the components of the different matrices while maintaining optimal peak heights and areas, the most suitable elution profile was developed (section “Chromatographic and MS/MS conditions”). Preliminary studies for MS/MS conditions started by checking PAT fragmentation in both positive and negative ionization modes using Optimizer software (version 10.1, build 10.1.67, Agilent Technologies). The PAT molecule undergoes better ionization in negative mode, as confirmed in the literature26. The analysis was performed in ESI− mode, monitoring MRM transitions corresponding to m/z 153→109 and 153→81 for PAT, and 160→115 and 160→86 for the isotopically labeled internal standard (¹³C₇-PAT). The first transition corresponds to a fragment formed via loss of CO₂ (decarboxylation/lactone-related fragmentation) [M − H−CO2]− and the second one to further fragmentation of the ring structure with additional neutral losses, [M − H−CO2 − CO]−. Both MRM transitions were consistently monitored and recorded throughout the entire analytical procedure, including calibration standards, quality control samples, and all analyzed samples. The identity of PAT was confirmed based on the following criteria: (i) agreement of the retention time with that of the reference standard within a tolerance of ± 0.1 min, and (ii) consistency of the ion ratio (quantifier/qualifier) within an acceptable deviation of ± 20% relative to the corresponding ratio obtained for the calibration standard. In the next step, the source parameters were selected by the Source Optimizer program (version 10.1, build 10.1.67, Agilent Technologies). The results of the source optimization were unsatisfactory, so some source parameters were manually selected based on data provided by the software (Supplementary data, Table S1). The study was started by verifying the sheath gas temperature at 250, 300, 350, and 400 °C. It was noted that the highest temperature yields the highest PAT signals; however, this is the device’s maximum operating temperature. Therefore, 350 °C was chosen for technical reasons. The dry gas temperature was tested from 150 to 300 °C in 50 °C increments, and 200 °C was optimal, 50 degrees higher than the optimizer indicated. The last factor analyzed was dry gas flow rate, which was verified from 6 to 12 L/min. The best results were obtained at a flow rate of 8 L/min, two units higher than those obtained using the optimizer. The capillary voltage was examined from 3000 to 5000 V in 1000 V increments. A voltage of 3000 V was selected instead of 2000 V. With the changes applied, the peak heights and areas increased. During manual optimization, an appropriate amount of time was waited after the applied changes to stabilize the selected source operating parameter. Internal standard It is well known that an internal standard (IS) is used to correct variability in sample preparation, instrument response, and injection volume during quantitative analysis. It improves the precision and accuracy of the method by compensating for matrix effects and potential analyte losses during sample processing. The IS should be chemically similar to the analyte but must be clearly distinguishable by the detector to allow reliable quantification. In MS/MS analysis, an isotopically labeled (13C) IS is often used. It co-elutes with the analyte and exhibits nearly identical physicochemical properties, providing the most accurate correction for matrix effects and ion suppression. For this reason, 13C7-PAT was selected as the IS. The impact of IS concentration on the slope of the curve and analytical results was systematically evaluated. The series of IS solutions at concentrations of 100, 250, 500, and 1000 µg/L were prepared and added to a calibration curve ranging from 0.25 to 200 µg/L, and the reference material. It turned out that concentrations of 100 and 250 µg/L were insufficient to maintain a stable and consistent calibration curve, in contrast to the higher concentrations. Taking into account both analytical performance and the high cost of the labeled compound, a final IS concentration of 500 µg/L was selected as optimal. Validation of the method To assess the validity of the developed method for quantifying PAT in food, complete validation was conducted based on the International Council on Harmonisation33. According to these guidelines, we determined specificity, linearity, the limit of quantification (LOQ), precision, accuracy, and stability. The influence of the matrix effect and carry-over effect was also evaluated. Linearity The linearity of the method over the range of 0.25 µg/L to 200 µg/L was obtained from ten different concentrations of PAT. Calibration curves were prepared in triplicate on the same day and in a single repetition for three consecutive days. Each sample was processed according to the recommended procedure. The peak area ratios of the analyte to IS were used to construct the calibration curve (see Supplementary data, Figure S4). The equation for the linear regression line and the coefficient of determination for intra-day data were y = 0.051314x − 0.003236 and R2 = 0.9989, respectively. The corresponding equation for inter-day data was y = 0.047130x − 0.003908, with R2 = 0.9997. For both curves, the correlation coefficients are greater than 0.999, indicating a linear relationship between the detector signal and the concentration of the tested PAT. The method was considered linear over the tested concentration range when the coefficient of determination (R²) was ≥ 0.9933. Precision, accuracy Samples for precision and accuracy were prepared in triplicate on the same day and in one repetition over three consecutive days to determine repeatability within and between days. According to the ICH guide, precision and accuracy quality control samples should be determined at a minimum of four concentration levels within the calibration curve range. The first is the lowest point of the curve, the second is within the bottom 25% of the curve, the next is at about 30–50% of the curve’s range, and the last is at least 75% of the highest concentration33. To verify the method, the accuracy and precision of calibration standards were checked at each concentration spanning the calibration curve, and for food samples at three different concentration levels. The precision of the method was expressed using the coefficient of variation (CV) of repeatability and accuracy as a percentage of analyte recovery. For standard PAT solutions, intra-day and inter-day precision ranged from 0.1 to 9.3% and 1.1 to 8.8%, respectively. At the same time, intra-day and inter-day accuracy in the studied range varied from 96.5 to 101.7% and 94.5 to 103.8%, respectively. The obtained data meet the criteria and demonstrate that the developed method is precise and accurate. Additional data are provided in the Supplementary data, Table S2. The intra-day precision for food samples was generally satisfactory across all tested matrices, with CV not exceeding 10% at any concentration level (Table 1). Intra-day accuracy for these samples was between 81.3 and 104.6%, indicating acceptable recoveries. Notably, slightly lower accuracy was observed in some juice-based products at mid- and high concentrations, which may reflect matrix-related effects, such as partial signal suppression. In accordance with applicable standards, the coefficient of variation CV [%] was used to assess the precision of the method, the value of which should not exceed 15% for the tested concentration range, excluding the lowest point of the curve for which a CV value of 20% is acceptable. The method’s accuracy was assessed based on recovery [%]. In the tested concentration range, the accuracy should be between 85% and 115% for all points, excluding the lowest concentration on the calibration curve, where the accuracy is between 80% and 120%33. Table 1Evaluation of intra-day precision for different food matrices (n = 6).Product groupMaximum permittedlevel * [µg/kg]ProductConcentration[µg/kg]Precision [%]Accuracy [%]addedfoundFruit juices, fruit juices fromconcentrate, concentratedfruit juices and fruit nectars50Juice21.9 ± 0.14.196.95041.4 ± 1.02.482.710081.3 ± 1.51.881.3Apple concentrate21.87 ± 0.14.993.55050.0 ± 1.42.8100100102.5 ± 2.01.9102.5Smoothie - fruit juicesfrom concentrated juice22.0 ± 0.29.8100.35043.3 ± 1.63.886.710084.8 ± 1.31.684.8Nectars21.9 ± 0.14.395.1108.2 ± 0.66.881.910081.7 ± 1.11.481.7Spirit drinks, cider andother fermented drinksderived from apples orcontaining apple juice50Vodka22.0 ± 0.29.8101.25048.1 ± 1.42.896.3100101.9 ± 1.01.0101.9Beer22.1 ± 0.29.5104.6547.9 ± 1.63.495.8100102.1 ± 0.90.9102.1Solid apple products placedon the market for the finalconsumer25Apple22.1 ± 0.11.4103.45048.6 ± 0.81.697.210094.9 ± 2.12.294.9Freeze-dried bar22.1 ± 0.12.0104.65048.9 ± 0.71.597.810096.7 ± 1.51.696.7Apple juice and solid appleproducts for infants andyoung children and labelledand placed on the marketas such10Juice22.0 ± 0.12.697.75043.7 ± 2.04.587.310084.2 ± 1.21.484.2Children’s puree22.0 ± 0.13.9102.0109.4 ± 0.10.994.110092.6 ± 2.32.592.6Baby food10Baby dessert21.9 ± 0.13.995.45052.0 ± 1.42.7104.1100101.3 ± 1.31.3101.3* In accordance with the regulation35. Table 1Evaluation of intra-day precision for different food matrices (n = 6).Product groupMaximum permittedlevel * [µg/kg]ProductConcentration[µg/kg]Precision [%]Accuracy [%]addedfoundFruit juices, fruit juices fromconcentrate, concentratedfruit juices and fruit nectars50Juice21.9 ± 0.14.196.95041.4 ± 1.02.482.710081.3 ± 1.51.881.3Apple concentrate21.87 ± 0.14.993.55050.0 ± 1.42.8100100102.5 ± 2.01.9102.5Smoothie - fruit juicesfrom concentrated juice22.0 ± 0.29.8100.35043.3 ± 1.63.886.710084.8 ± 1.31.684.8Nectars21.9 ± 0.14.395.1108.2 ± 0.66.881.910081.7 ± 1.11.481.7Spirit drinks, cider andother fermented drinksderived from apples orcontaining apple juice50Vodka22.0 ± 0.29.8101.25048.1 ± 1.42.896.3100101.9 ± 1.01.0101.9Beer22.1 ± 0.29.5104.6547.9 ± 1.63.495.8100102.1 ± 0.90.9102.1Solid apple products placedon the market for the finalconsumer25Apple22.1 ± 0.11.4103.45048.6 ± 0.81.697.210094.9 ± 2.12.294.9Freeze-dried bar22.1 ± 0.12.0104.65048.9 ± 0.71.597.810096.7 ± 1.51.696.7Apple juice and solid appleproducts for infants andyoung children and labelledand placed on the marketas such10Juice22.0 ± 0.12.697.75043.7 ± 2.04.587.310084.2 ± 1.21.484.2Children’s puree22.0 ± 0.13.9102.0109.4 ± 0.10.994.110092.6 ± 2.32.592.6Baby food10Baby dessert21.9 ± 0.13.995.45052.0 ± 1.42.7104.1100101.3 ± 1.31.3101.3* In accordance with the regulation35. Evaluation of intra-day precision for different food matrices (n = 6). Fruit juices, fruit juices from concentrate, concentrated fruit juices and fruit nectars Smoothie - fruit juices from concentrated juice Spirit drinks, cider and other fermented drinks derived from apples or containing apple juice Solid apple products placed on the market for the final consumer Apple juice and solid apple products for infants and young children and labelled and placed on the market as such * In accordance with the regulation35. The inter-day precision and accuracy were evaluated for two food matrices: solid apple and fruit juices (Table 2). In both cases, inter-day precision did not exceed 5% at the tested concentration levels, and accuracy varied from 94.6 to 102.5%. These results demonstrate high repeatability and confirm that the method is suitable for routine analysis of PAT in a wide range of apple and apple-based products. Table 2Evaluation of inter-day precision in different food samples (n = 6).Product groupMaximum permittedlevel * [µg/kg]ProductConcentration[µg/kg]Precision [%]Accuracy [%]addedfoundSolid apple products placedon the market for the finalconsumer50apple5047.3 ± 1.22.594.650.9 ± 1.52.9101.850.1 ± 1.02.1100.2Fruit juices, fruit juices fromconcentrate, concentratedfruit juices and fruit nectars50juice10095.1 ± 4.14.395.1102.5 ± 2.22.1102.5100.6 ± 2.22.2100.6* In accordance with the regulation35. Table 2Evaluation of inter-day precision in different food samples (n = 6).Product groupMaximum permittedlevel * [µg/kg]ProductConcentration[µg/kg]Precision [%]Accuracy [%]addedfoundSolid apple products placedon the market for the finalconsumer50apple5047.3 ± 1.22.594.650.9 ± 1.52.9101.850.1 ± 1.02.1100.2Fruit juices, fruit juices fromconcentrate, concentratedfruit juices and fruit nectars50juice10095.1 ± 4.14.395.1102.5 ± 2.22.1102.5100.6 ± 2.22.2100.6* In accordance with the regulation35. Evaluation of inter-day precision in different food samples (n = 6). Solid apple products placed on the market for the final consumer Fruit juices, fruit juices from concentrate, concentrated fruit juices and fruit nectars * In accordance with the regulation35. Limit of detection and limit of quantification To estimate the LOD and LOQ for PAT, food samples free of endogenous PAT were spiked with decreasing concentrations of the analyte and analyzed. The experimentally determined values were 0.125 µg/L (0.125 µg/kg) for the LOD and 0.25 µg/L (0.25 µg/kg) for the LOQ. For dried apples and freeze-dried bars, slightly higher detection limits were observed due to the more complex matrix. The limit of detection (LOD) was 0.33 µg/kg, while the limit of quantification (LOQ) was 2 µg/kg. At these concentrations, the corresponding signal-to-noise ratios were 3 and 10, respectively33. The obtained LOQ value is better than that reported by others: 42 µg/kg20, 4.5 µg/kg29, and 10 µg/kg34. But some methods have better or similar LOQ 0.0704 µg/L, 0.76 µg/kg12 or 1.15 µg/kg7. Additional details are provided in the supplementary materials (Table S3). For further validation and application studies, a concentration level of 2 µg/kg was selected. This level was chosen to reflect the intended application of the method as a tool for routine monitoring of PAT in food, where analytical results are primarily interpreted in relation to regulatory limits (i.e., compliance vs. non-compliance). With respect to the EU and US FDA regulations mentioned above, the selected level of 2 µg/kg remains significantly below the established limits, ensuring that the method provides sufficient sensitivity for reliable detection of contamination well before regulatory thresholds are exceeded. It should be noted that the higher LOQ observed for dry matrices (2 µg/kg) is still fully adequate for regulatory purposes and consistent with the intended use of the method for routine analysis. Although the method allows determination of PAT at lower concentrations (particularly in liquid matrices), the primary goal of this study was not ultra-trace analysis, but rather the development of a robust and practical method for everyday food safety monitoring. Carry-over effect In analytical methods, an undesirable effect is carry-over, i.e., the transfer of part of the analyte to the next sample, which can falsify the result of a given analysis by increasing the concentration. In this study, the effect of carry-over was assessed by setting up a zero sample (blank sample spiked with IS) after the samples of the. highest concentration of the calibration curve (200 µg/L). On this basis, it was concluded that the developed method is free of this effect, as confirmed by Fig. 2, which shows no PAT signal. Fig. 2MRM chromatograms for a zero sample injected after the highest concentration of calibration standard. (A) m/z PAT 153→109 (B) m/z PAT 153→81 (C) m/z 13C7-PAT 160→115 (D) m/z 13C7-PAT 160→86. MRM chromatograms for a zero sample injected after the highest concentration of calibration standard. (A) m/z PAT 153→109 (B) m/z PAT 153→81 (C) m/z 13C7-PAT 160→115 (D) m/z 13C7-PAT 160→86. Matrix effect To evaluate the matrix effect, signal suppression was assessed based on the response obtained from matrix extracts prior to correction with the IS. As presented in Table 3, noticeable variability in signal suppression was observed both between different matrices and across analytical days. The suppression reached up to 43.2% for apple concentrate and up to 20.9% for apple samples, confirming the presence of matrix effects typical of PAT analysis. This behavior is consistent with the known properties of PAT, particularly its high polarity and tendency to co-elute with matrix components, which contributes to ion suppression. To compensate for the signal variability of PAT, the use of an isotopically labeled IS was necessary. Table 3Day-to-day variation of signal in apples and apple concentrate extracts in analyses without an internal standard.ExtractsDay 1Day 2Day 3Day 4Apple concentrate27.9 ± 5.5%43.2 ± 4.7%32.1 ± 3.2%39.3 ± 3.1%Apple20.9 ± 1.4%16.1 ± 2.7%7.3 ± 3.4%19.4 ± 2.0% Table 3Day-to-day variation of signal in apples and apple concentrate extracts in analyses without an internal standard.ExtractsDay 1Day 2Day 3Day 4Apple concentrate27.9 ± 5.5%43.2 ± 4.7%32.1 ± 3.2%39.3 ± 3.1%Apple20.9 ± 1.4%16.1 ± 2.7%7.3 ± 3.4%19.4 ± 2.0% Day-to-day variation of signal in apples and apple concentrate extracts in analyses without an internal standard. The influence of IS use on the minimization of matrix effects was assessed using food samples, listed in the EU regulation35. The effect was tested by analyzing various apples and apple-based products. Thirteen products from the Polish market, including freeze-dried bars, mousses, beers, strong alcohols, baby foods, dried foods, juices, and others, were used in the study. These products were fortified with PAT at each concentration level, ranging from 2 to 100 µg/L, and prepared according to the procedure described in section “Sample preparation”. For this purpose, the slope of each matrix’s curve was compared with that of the solvent curve, and the CV was determined to be 9.73% (Table 4). Therefore, it can be concluded that this method is not entirely free from the matrix effect, since the coefficient of variation is greater than 0. However, its impact on the analysis is negligible, as it remains below 15%, which is consistent with ICH guideline33. The influence of the sample matrix on analyte retention was also investigated by comparing PAT retention times across different samples (Supplementary data, Figure S5). Comparison of PAT retention time obtained from the calibration curve, food samples enriched with PAT at the same concentration as the calibration curve, and authentic food samples, demonstrated that the matrix effect did not significantly impact the determination of PAT in the tested concentration range with the use of isotopically labeled IS. It should be emphasized that the use of an IS does not eliminate matrix effects themselves, but effectively corrects their impact on quantification. Table 4Evaluation of the matrix effect.Source numberMatrix typeSlopeAverageSDCV [%]1Freeze-driedbar0.0078440.008380.000829.732Apple0.0081683Vodka 10.0082834Mousse0.0081995Mousse forchildren0.0096436Fruit infusion0.0070227Nectar0.0088648Children’s puree0.0088399Smoothie0.00804110Juice0.00850211Dried apple0.00785212Vodka 20.00846913Beer0.008555Ppure solventWater: methanol 1:10.009050 Table 4Evaluation of the matrix effect.Source numberMatrix typeSlopeAverageSDCV [%]1Freeze-driedbar0.0078440.008380.000829.732Apple0.0081683Vodka 10.0082834Mousse0.0081995Mousse forchildren0.0096436Fruit infusion0.0070227Nectar0.0088648Children’s puree0.0088399Smoothie0.00804110Juice0.00850211Dried apple0.00785212Vodka 20.00846913Beer0.008555Ppure solventWater: methanol 1:10.009050 Evaluation of the matrix effect. Freeze-dried bar Mousse for children Stability testing A stability study was performed to ensure that none of the sample preparation and analysis steps, as well as the storage conditions of the standard solutions and tested food products, influenced the results. The stability of the analytical standards was evaluated by testing three series of curves stored in an autosampler at 4 °C at several time points up to 7 days. The results demonstrated that the standards remained stable across a broad concentration range (0.25–200 µg/L), with acceptable deviations observed after 7 days of storage. At the lowest tested concentration, the standard showed a slight decrease from 0.25 to 0.21 µg/L, corresponding to 16.0%. The 1.0 µg/L standard remained relatively stable, showing only a minor lowering of the concentration from 1.01 to 0.90 µg/L (-10.9%). A comparable downward trend was also observed at higher concentrations. For instance, the 10 µg/L standard decreased from 10.83 to 9.73 µg/L, corresponding to a 10.2% reduction. At the upper concentration level of 100 µg/L, the observed decrease was from 97.85 to 90.89 µg/L (-7.1%). The stability of PAT in a real matrix sample (proficiency test materials with assigned target values and acceptable ranges as reference materials of apple concentrate) was evaluated at storage temperatures of -20, 4, and 20 °C over a period of 7 days. For apple concentrate, analyte stability was evaluated by comparing recovery to day 0. At -20 °C, recoveries ranged from 92.9% to 100.2%; at 4 °C, from 91.3% to 99.7%; and at 20 °C, from 91.0% to 99.4%. These results indicate that storage temperature did not have a significant impact on the short-term stability of PAT in the tested matrix. The long-term stability of PAT in proficiency test materials with assigned target values and acceptable ranges, serving as reference materials, was also evaluated (Fig. 3). Table 5 shows the long-term stability of apple puree tested two weeks apart. A downward trend was observed in average PAT concentration from 20.05 ± 0.37 to 16.93 ± 0.08 µg/kg (84.4%) after two weeks of storage at -20 °C. The increase in variance could be due to small differences in storage conditions. Fig. 3Control chart for proficiency test materials with assigned target values and acceptable ranges as reference materials (apple concentrate) within 8 months showing long-term sample stability. Control chart for proficiency test materials with assigned target values and acceptable ranges as reference materials (apple concentrate) within 8 months showing long-term sample stability. Table 5PAT analysis results for proficiency test materials with assigned target values and acceptable ranges as reference materials.Proficiency test materials with assignedtarget values and acceptable rangesas reference materialsScope of the certificateand targetResults obtainedDate testingCider (degassed)38–116 µg/kg,69.97 µg/kg20.03.2025Bipea 28 /9-0428-0021Target = 77 µg/kg63.65 µg/kg20.03.2025Apple puree8.56–22 µg/kg,20.47 µg/kg13.11.202419.88 µg/kg13.11.202419.79 µg/kg13.11.2024Test Material 1685Target = 15.3 µg/kg16.98 µg/kg27.11.202416.87 µg/kg27.11.2024Apple juice (cloudy) TestMaterial 168610.2–26.2 µg/kg,14.46 µg/kg13.11.202414.17 µg/kg13.11.2024Target = 18.2 µg/kg15.00 µg/kg13.11.2024 Table 5PAT analysis results for proficiency test materials with assigned target values and acceptable ranges as reference materials.Proficiency test materials with assignedtarget values and acceptable rangesas reference materialsScope of the certificateand targetResults obtainedDate testingCider (degassed)38–116 µg/kg,69.97 µg/kg20.03.2025Bipea 28 /9-0428-0021Target = 77 µg/kg63.65 µg/kg20.03.2025Apple puree8.56–22 µg/kg,20.47 µg/kg13.11.202419.88 µg/kg13.11.202419.79 µg/kg13.11.2024Test Material 1685Target = 15.3 µg/kg16.98 µg/kg27.11.202416.87 µg/kg27.11.2024Apple juice (cloudy) TestMaterial 168610.2–26.2 µg/kg,14.46 µg/kg13.11.202414.17 µg/kg13.11.2024Target = 18.2 µg/kg15.00 µg/kg13.11.2024 PAT analysis results for proficiency test materials with assigned target values and acceptable ranges as reference materials. Apple juice (cloudy) Test Material 1686 Application of the method After validation, the PAT assay method was used to analyze food that could contain this toxin and is listed in the EU regulation35. Testing using each matrix started by checking for the presence of the toxin. If a given matrix was free of PAT, then it was fortified at three levels. The lowest (2 µg/kg) and highest (100 µg/kg) levels were common to all food samples, while the middle level depended on the value given in the regulation35, i.e., for juices and nectars − 50 µg/kg, apple fermented drinks − 50 µg/kg, solid apple products − 25 µg/kg, baby food − 10 µg/kg. Verification of the developed method was based on testing proficiency test materials with assigned target values and acceptable ranges as reference materials of various matrices, including cider (degassed) (38–116 µg/kg, target = 77 µg/kg), apple puree (8.56–22 µg/kg, target = 15.3 µg/kg), apple juice (cloudy) (10.2–26.2 µg/kg, target = 18.2 µg/kg) and apple concentrate (33–115 µg/kg, target = 74 µg/kg) (Table 5 and Figure S5). Additionally, the developed method was also used by an external laboratory that participated in the Food Chemistry Proficiency Tests (1686 – Apple Juice (Cloudy); z-score = 0.2, 1688 – Apple Purée; z-score = -0.2, and 1685 – Apple Purée; z-score = 0.2, and 1683 – Apple Juice (Cloudy); z-score = 0.6). The first laboratory site obtained a PAT concentration of 18.81 µg/kg, which is close to the expected (target) value of 18.02 µg/kg. The second laboratory site obtained a PAT concentration of 10.05 µg/kg while the target was 10.5 µg/kg. In the third proficiency test, the result was 16 µg/kg, while the expected value was 15.3 µg/kg. However, in the latest FAPAS (Food Analysis Performance Assessment Scheme) study, a value of 25 µg/kg was reported for juice, whereas 22.2 µg/kg was expected. In addition, the usefulness of the method was also verified in BIPEA (Bureau International des Problèmes d’Eau et d’Alimentation). Stewed apples were tested, and the PAT content was 116 µg/kg (z-score = 0.04), while the target was 115 µg/kg. All studies demonstrated the usefulness of the developed method for determining PAT in food and the ease of translation across different laboratories. Results from proficiency test materials with assigned target values and acceptable ranges, used as reference materials, confirmed the reliability and applicability of the developed method for routine PAT determination in food. Ascorbic acid in ethyl acetate effectively and with very good efficiency (approximately 95%) extracts PAT from the matrix relevant in food analysis. The research paper presents fortification results for all matrices specified in the EU regulation, and no PAT was detected in commercially purchased food. However, studies reported in the literature have confirmed the presence of PAT in commercially available apples and apple-based products. Table 6 presents a summary of data on PAT occurrence and the analytical parameters of the methods used across various countries. The data clearly demonstrate that PAT contamination levels in apples and apple-based vary widely by region and matrix, ranging from below the limit of quantification (< LOQ) to several hundred µg/kg, and in extreme cases, even exceeding 1000 µg/kg (e.g., Brazil). At the same time, significant differences in analytical sensitivity are visible between the published methods. Reported LOQ values range from 0.3 µg/kg to 42 µg/kg, depending on the method and matrix. In this context, the method proposed in the present study, with LOQs of 0.25 µg/kg (for liquid products) and 2 µg/kg (for dried apples and freeze-dried bars), provides sensitivity equal to or better than most of the reported approaches. Importantly, many studies report PAT levels below LOQ or within relatively low concentration ranges (e.g., Poland, Belgium, Spain), consistent with the results of this work: no PAT was found in commercially purchased food in Polish shops. This result indicates that the absence of detectable PAT in the analyzed samples is not unusual but rather reflects current trends in improved quality control and processing practices. The herein described method is simple and does not require specialized equipment, which is confirmed by the results of inter-laboratory studies. The results presented in this work support the view that our method is a simple yet effective way to determine PAT concentration in food products. Table 6Occurrence of patulin and the analytical parameters of the methods used in various countries.Country / regionMatrix (Product type)PAT range [µg/kg]LOD *[µg/kg]LOQ *[µg/kg]AnalyticalmethodSourcePolandJuices, baby food< LOQ – 14.725LC-DAD 37 BelgiumJuices, cider, puree< LOQ – 35.90.41.2LC-MS/MS 38 BrazilFresh apples, juices< LOQ – 757.0< LOQ –129812.542LC-MS/MS 20 USAJuices, ciders, puree, baby food< LOQ – 56.61.144.0LC- MS/MS 39 TunisiaJuices, compote< LOQ – 167.00.10.3LC-UV 40 ChinaApple juice concentrates6.3 – >500.822.72LC-DAD 41 SpainJuices, apple products< LOQ – 252.086.25LC-UV 42 * Lowest values. Table 6Occurrence of patulin and the analytical parameters of the methods used in various countries.Country / regionMatrix (Product type)PAT range [µg/kg]LOD *[µg/kg]LOQ *[µg/kg]AnalyticalmethodSourcePolandJuices, baby food< LOQ – 14.725LC-DAD 37 BelgiumJuices, cider, puree< LOQ – 35.90.41.2LC-MS/MS 38 BrazilFresh apples, juices< LOQ – 757.0< LOQ –129812.542LC-MS/MS 20 USAJuices, ciders, puree, baby food< LOQ – 56.61.144.0LC- MS/MS 39 TunisiaJuices, compote< LOQ – 167.00.10.3LC-UV 40 ChinaApple juice concentrates6.3 – >500.822.72LC-DAD 41 SpainJuices, apple products< LOQ – 252.086.25LC-UV 42 * Lowest values. Occurrence of patulin and the analytical parameters of the methods used in various countries. < LOQ – 757.0 < LOQ – 1298 * Lowest values. Conclusion The developed LC-MS/MS method represents a robust, efficient, and practically applicable approach for the determination of PAT in apples and apple-based products. A key innovation of the method is the use of a single-step extraction with ethyl acetate supported by ascorbic acid, which enables highly efficient analyte transfer (often approaching 95%) without the need for repeated extraction cycles or additional cleanup procedures commonly reported in the literature. This solution significantly simplifies and accelerates sample preparation while reducing solvent consumption and the risk of analyte loss. In contrast to many existing methods that rely on multistep extraction and matrix purification, the proposed approach minimizes procedural complexity while maintaining high analytical performance. The achieved limit of quantification (0.25 µg/kg) is substantially lower than the maximum levels established by both European Union regulations (50 µg/kg for juices and 10 µg/kg for infant food) and the US FDA (50 µg/kg), ensuring reliable detection of PAT at levels far below regulatory thresholds. Compared to literature data, the method demonstrates equal or superior sensitivity relative to commonly applied techniques. Although significant matrix effects were observed, as expected for PAT due to its physicochemical properties, the use of an isotopically labeled IS (¹³C-PAT) enabled effective compensation, ensuring accurate and reproducible quantification across different matrices. Importantly, no detectable levels of PAT were found in the analyzed real samples. When interpreted in the context of the method’s low LOQ and current regulatory limits, this result confirms both the method’s appropriate sensitivity and the compliance of the tested products with food safety requirements. Methods Reagents and materials The conducted research studies used commercially available reagents with a purity not lower than the analytical purity for the following reagents: ethyl acetate (141-78-6, Sigma-Aldrich), L (+)-Ascorbic acid (50-81-7, J.T. Baker, Deventer, Holland), sodium chloride (7647-14-5, Merck). The LC-MS purity reagents were water (7732-18-5, Supelco, Merck), methanol (67-56-1, Supelco, Merck), acetonitrile (75-05-8, Supelco, Merck), acetic acid 100% (64-19-7, LiChropur®, Merck), formic acid 98–100% (64-18-6, LiChropur®, Merck), ammonium acetate (631-61-8, Sigma-Aldrich), ammonium formate (540-69-2, Sigma-Aldrich), ammonium fluoride (12125-01-8, Sigma-Aldrich). Patulin (100 µg/mL, 149-29-1, BiopureTM) and 13C7-patulin (25.02 µg/mL, 149-29-1, BiopureTM). Instrumentation The experiment used an Agilent 1260 Infinity II liquid chromatograph equipped with a pump coupled to a model G7104C vacuum degasser, column compartment model G7116B, automated sample injector model G7129C connected with temperature-controlled sample storage, and a triple quadrupole G6470B mass spectrometer system with Agilent JetStream ionization technology (Agilent Technologies, Waldbronn, Germany). Instrument control, data acquisition, and analysis were performed using MassHunter software version 10.2. Analytes were separated on Reprospher 100 C18-Aqua (1.8 μm, 100 × 2.0 mm) column from Dr. Maisch, High-Performance LC GmbH (Entringen, Germany). Samples and stock solutions were stirred in a fridge or freezer (Panasonic Healthcare Co., Ltd., Sakata, Japan). The Multi Speed Vortex MSV-3500 (Biosan, Latvia), CLASSIC Vortex Mixer (VELP Scientifica), and shaker (Collomix GmbH Daimler) were used for PAT extraction, while sample concentration was done in an automatic system with a nitrogen stream (Techne DB100/3 Dri-block) or vacuum concentrator (LABCONCO 181064986 G). Lastly, solid samples were ground using Blixer®3. Standard solutions The PAT and 13C7-PAT standard stock solutions 200 and 500 µg/L, respectively, were prepared by diluting stock solutions with a mixture of methanol/water (1/1, v/v). To prepare a PAT calibration curve in the range of 0.25 to 200 µg/L, the intermediate solutions were prepared by dilution using the methanol/water solution (1/1, v/v). All solutions were prepared using LC-MS grade solvents. Sample collection In this work, Polish apples and apple-based products (freeze-dried bar, three types of puree, nectar, children’s puree, smoothie, juice, dried apple) or as an ingredient in alcoholic beverages (two kinds of vodka, beer) were studied. The products were purchased in September 2024 at local markets. Sample preparation A 2.500 g ± 0.005 g sample was weighed into 15 mL polypropylene flacons (to make a fortification at a given level, add the appropriate volume of standard to the sample), and 125 µL of 13C7-PAT at a concentration of 500 µg/L, 0.15 g of L(+)-ascorbic acid and 5 mL of ethyl acetate were added. The sample was extracted for 5 min in the vortex and then centrifuged at 5000g for 5 min at room temperature. Then, 4 mL of supernatant was collected and evaporated to dryness (about 45 min) using a rotary evaporator LABCONCO system at 50 °C36. The residue was dissolved in 0.889 mL of methanol and filtered with 0.45 μm nylon filter. Then, 50 µL of the filtrate was transferred into a glass dark vial with an insert, diluted with 50 µL of water. The solution was mixed and subjected to chromatographic analysis. When preparing beer or a carbonated beverage, samples were degassed under air for 5 min before weighing. For dry and strong alcohol samples, 2.5 mL of water was added before extraction. Chromatographic and MS/MS conditions PAT analysis was performed using LC-MS/MS equipped with an ESI source. The chromatographic separation was achieved by C18-Aqua column, kept at 40 °C. A mobile phase with a gradient program was used as follows: 5% B (at 0 min), 20% B (at 1.5 min), 35% B (at 4.5 min), 70% B (at 6 min), and 5% B (at 6.5 min) with a post-time 5 min. Mobile phase A was water containing 1 mmol/L acetic acid and 2 mmol/L ammonium acetate, and phase B was methanol. Elution was obtained at a flow rate of 0.2 mL/min and an injection volume of 5 µL. Mass spectrometry analysis was carried out in negative ESI mode. The optimal ionization source conditions were set in Table 7. Table 7Ion source and data acquisition settings in MS/MS mode.ParametersValuesSourceDry gas temperature [°C]200Dry gas flow [L/min]8Nebulizer pressure [psi]60Sheath gas temperature [°C]350Sheath gas flow [L/min]10Capillary [V]3000Nozzle voltage [V]0AcquisitionQuantifierQualifierFragmentor100100Collision energy [V]511Cell accelerator voltage [V]44m/z PAT153→109153→81m/z 13C7-PAT160→115160→86 Table 7Ion source and data acquisition settings in MS/MS mode.ParametersValuesSourceDry gas temperature [°C]200Dry gas flow [L/min]8Nebulizer pressure [psi]60Sheath gas temperature [°C]350Sheath gas flow [L/min]10Capillary [V]3000Nozzle voltage [V]0AcquisitionQuantifierQualifierFragmentor100100Collision energy [V]511Cell accelerator voltage [V]44m/z PAT153→109153→81m/z 13C7-PAT160→115160→86 Ion source and data acquisition settings in MS/MS mode. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1