Abstract :
[en] [en] BACKGROUND: Mineral oil aromatic hydrocarbon (MOAH) analysis in foods is a major analytical challenge. Quantification is associated with a high uncertainty. The sources of uncertainty are multiple, but the major one is related to data interpretation and integration, which is partially derived from insufficiently efficient sample preparation. Recently, an updated ISO method for the analysis of mineral oil in fats and oils and a standard operating procedure for infant formula analysis have been published. Both methods reported significantly different (up to 1.25) distributions of the internal standards used for quantification (i.e., tri-tert-butyl benzene (TBB) and 2-methyl naphthalene (2-MN)) over the different solvent phases used in the saponification step.
RESULTS: In this work, a microwave-assisted saponification and extraction method was optimized for MOAH analysis to solve the problem related to the MOAH internal standards partition. The paper examines the impact of the solvent mixture used, the concentration of KOH on the partition of TBB and 2-MN, and the effect of the matrix and the washing step to extract the unsaponifiable fraction containing the mineral oils.
SIGNIFICANCE: The optimized procedure achieved a TBB/2-MN ratio of 1.05 ± 0.01 tested in five different fats and oils, namely, sunflower, rapeseed, coconut, palm, and extra virgin olive oils. The method can significantly contribute to reducing the uncertainty of the MOAH quantification when saponification is applied.
Funding text :
The use of the FID, a nonspecific detector, requires that only highly purified fractions reach the detector for reliable quantification; additionally, enrichment steps are often needed to achieve the required sensitivity. Therefore, extensive sample preparation steps are often applied before the LC-GC-FID analysis to enrich the extract (e.g., saponification) and remove interferent compounds (e.g., epoxidation or purification on aluminum oxide) [9\u201312].Despite the very wise design of the method, which uses several ISs to control the different steps of the analytical workflow and to quantify the MOSH and the MOAH, the uncertainty of the results remains, indeed, rather high in certain class of samples, mainly characterized by high fat content or high amount of interferences (e.g., edible oils and spices). The main source of variability is due to data interpretation and integration, which has been estimated to account for more than 20 % of the total variability [9\u201312]. Several tools have been developed and made available to support analysts, among which the use of GC \u00D7 GC, which is gaining attention to support the interpretation [13\u201315] and a guidance to support chromatogram interpretation (at least on infant formula) was published by the Joint Research Center (JRC) [16]. However, interpretation and integration variability are tightly linked to the significant variability associated with the sample preparation step (i.e., extraction/enrichment and purification \u2013 typically epoxidation for MOAH). In 2019, and updated in 2023, the JRC published a Guidance on sampling, analysis, and data reporting aiming for harmonization [8,17]. A decision tree was reported in the attempt of harmonizing the analytical workflow and support analysts in deciding whether to apply or not the enrichment or purification step. Nevertheless, it appeared evident that this was not enough, as high variability among laboratories was still present, as shown in some interlaboratory trials organized by the JRC for infant formula and edible oils reporting even up to 60 % of intermediate precision [18,19], and that standardization of the analytical method was necessary. A very limited number of official methods are indeed available. The UNI-EN 17517 for MOSH and MOAH in animal feeding stuff was approved in 2022. In 2017, the first official method (i.e., EN 16995:2017 for the determination of MOSH and MOAH in oils and vegetable fats [17]) was released. This was recently updated to lower the quantification limit (LOQ) from 10 mg/kg to 3 mg/kg for the MOSH fraction and 2 mg/kg for the MOAH [20]. The JRC also validated a standard operating procedure (SOP) for the determination of MOAH content in infant formula (IF) products [21,22]. In these methods, to reach the sensitivity required, an enrichment step is mandatory, and the most efficient procedure involves the use of saponification to simultaneously extract the MOSH and MOAH and remove the triglycerides that represent the limiting factor in increasing the sensitivity. In fact, the presence of triglycerides limits the amount of sample that can be injected in the LC-GC-FID system to 20 mg of lipids [23]. The removal of triglycerides by saponification before injection allows for an increase of up to 5-times in the sample that can be injected, thus reducing the LOQ proportionally [24]. Although highly necessary, the saponification step has been shown to contribute additional variability related to an uneven distribution of the MOAH internal standards (namely TBB and MNs, which should have a ratio of 1.0) in the different solvent phases [8,25]. This issue was highlighted by Menegoz Ursol et al. [25] (who suggested correcting for the recovery or using the average of the two standards) during the recent interlaboratory trials performed for updating the EN 16995 and standardizing the SOP for the IF. In the JRC report for the infant formula validation study [26], the authors discussed the high variability observed among the laboratories in the TBB/MN ratio (in the 0.9\u20131.4 range), having an average \u223C13 % lower MOAH results using TBB. They also mentioned that similar data were observed in the interlaboratory trial performed to validate the update of the EN16995 method. Finally, the recently updated JRC Guidance discusses that the TBB and MNs ratio is not respected when saponification is applied, having an average ratio around 1.15; and suggestes the use of TBB rather than MNs for the quantification of the MOAH (as claimed to be better extracted than the MNs), without explicitly setting a maximum acceptable ratio [8]. In the updated EN 16995 method (DGF/ISO-20122) the suitability test requires that TBB/2-MN ratio should be \u2264 1.25. The rationale behind preferring TBB over MN was that TBB has a higher degree of alkylation and thus better mimics the MOAH extraction [20]. This situation does not contribute to reducing the uncertainty of the results, first due to the different \u201Ctolerance\u201D reported in the official documents, and secondly, because an intrinsic uncertainty of up to 25 % on top of the variability due to the interpretation and integration of the chromatograms aforementioned is accepted in the sample preparation step when saponification is applied.Using Sap 2 and wash 3, the ratio TBB/2-MN was consistently around 1.05 \u00B1 0.01 (average and standard deviation in the 5 oils tested, n = 15), which means significantly lower than the 1.15 or 1.25 reported by the JRC Guidance (as a general average) and the DGF/ISO-20122 method (for quality control), respectively. For the method Sap 1 and wash 3 an overall average of 1.14 \u00B1 0.04 was obtained, not significantly different (p > 0.6 performing a t-test) from the one our laboratory obtained during the interlaboratory trial for the validation of the method DGF/ISO-20122. The difference in the ratio between the two method was confirmed for each oil by performing a t-test. A p-value largely below 0.05 was found for all the different oils, except EVOO, due to the high variability obtained with the Sap 1 wash 1 method. To further support the finding the variability in the TBB/2-MN ratio within each laboratory participating in the interlaboratory trial for validating the DGF/ISO-20122 method was calculated. Fig. S3 reports the average TBB/2-MN ratios for each laboratory on the 8 samples showing that the large majority of the laboratory obtained ratios far above the targeted value of 1.0. We can conclude that using the Sap 2 and wash 3 conditions, the intrinsic error is reduced from \u223C25 % to \u223C5 % (Fig. 4). Nevertheless, the robustness of the method needs to be tested in an interlaboratory trial to further confirm the improvement. Moreover, it would be interesting to further test it in infant formula to see if the changed saponification mixture contributes to the improvement of the ratio also in that case.This work is partially supported by Fonds de la Recherche Scientifique Belgique (FNRS) (CDR projects-MOHPlatform J.0170.20 and PDR projects-ToxAnaMOH T.0187.23). The authors thank LECO, Restek, and Milestone for their support. This article is based upon work from the Sample Preparation Study Group and Network, supported by the Division of Analytical Chemistry of the European Chemical Society.
Scopus citations®
without self-citations
0
