Chemical Contaminants in Plant Based Protein Survey

Initial method development work was carried out to determine the best extraction solvents and sample loading volume and sample mass equivalent for the IACs. The aim was to maximise the amount loaded on the IAC to ensure the target LOQs were achieved. The initial test used 80% methanol as extraction solvent, with the addition of salt. The extract was diluted with PBS and a relatively large volume (50 mL) was loaded on the IAC. This was equivalent to 0.625 g sample loaded. The approach resulted in acceptable recoveries for aflatoxins, HT-2, T-2 and zearalenone, but recoveries for other analytes were lower than the acceptable minimum of 70%. 

A series of different combinations of solvent mixes (ratios and solvent combinations) and different sample loading amounts were assessed. The results (not shown) found that the combination of acetonitrile / methanol / water as extraction solvent and a sample loading equivalent to 0.25 g sample resulted in the most consistent, acceptable recovery of all analytes across all three concentrations (1x, 2x and 10x target LOQ), while still allowing the target LOQs to be achieved. This method was used in a validation batch consisting of three blank samples, and six spikes each at 1x, 2x and 10x target LOQ as required by Fera in-house procedure for Application of Flexible Scope accreditation. 

The results of the in-house validation are presented in Annex C, Table C1. Average recoveries are between 70-120% for all mycotoxins at all concentrations, and for the overall average with one exception. This was for the 1xLOQ spike average for ochratoxin A where the average recovery was 68.8%. The repeatability was also high at this concentration with an RSDr (CV%) at 28.2%. The sample used as the blank material contained an average concentration of 0.33 µg/kg OTA, therefore the background level was greater than the spike addition level, which can cause variability and lower recovery results. The higher concentration spikes both gave average recovery values greater than 70% and the overall average was 73.2% with acceptable precision. One result was excluded from the data set for the 10xLOQ replicates for ochratoxin A. This samples contained a lower concentration than the blank, unspiked sample, so it ether was not injected properly on the LC-MS, or more likely was not spiked prior to extraction. 

In all other cases the target LOQ spikes gave acceptable recovery and precision. Therefore, the method was deemed to be acceptable. The calibration range used for each analyte extended below the target LOQ, to allow spiked samples with recovery less than 100% to be accurately quantified. In all cases two calibration solutions lower than the target LOQ were used in the calibration curves, a total of 9 calibration points were used. In all cases calibration curves met QC requirements for linearity and residuals from in-house document FSG 002. This has allowed results in survey samples below the target LOQ to be reported where they also met other reporting criteria, i.e. confirmation by ion ratio. 

The results for the mycotoxins in protein products from Phase 1 are given in Table C2 to Table C5. 

Table presents results for the six protein powder samples. Most results (69 of 78 data points, 88%) were below the LOQ for the method. Two samples contained no detectable residues of mycotoxins. Sterigmatocystin, DON, HT-2, T-2 and fumonisins B1, B2 and B3 were not detected in any protein powder samples. Three samples contained OTA, one pea protein powder at 9.33 µg/kg (S24-069134), and two samples of soy protein powder contained 4.15 and 3.22 µg/kg (S24-069137 and S24-069135). For samples S24-069134 and S24-069135, a pea and soy respectively, OTA was the only mycotoxin detected. Sample S24-069137 (soy) also contained aflatoxins B1, B2, G1 and G2 and zearalenone. The aflatoxin B1 level was 2.24 µg/kg, and the sum of the four aflatoxins was 4.94 µg/kg. The sample also contained 1.13 µg/kg zearalenone. 

There are GB MLs for aflatoxins and ochratoxin A in Commission Regulation (EC) No 1881/2006 in some foods but there are no MLs for mycotoxins currently in force for soy, pea or other plant proteins (1, 2) but there are MLs in force in GB for a range of products such as cereals (3 µg/kg) and dried vine fruit (10 µg/kg). There is also a ML for wheat gluten not placed on the market for the final consumer of 8 µg/kg (1, 2). There is an ML in force in the EU (Commission Regulation (EU) 2023/915) for soybeans (as well as sunflower seeds, pumpkin seeds and, watermelon seeds) of 5 µg/kg (4). The levels of aflatoxins (AFB1 and total) and OTA found in sample S24-069134 exceed the GB MLs for other foods such as cereals and wheat gluten. They also exceed the EU ML for soybean. The OTA level in S24-069137 and S24-069135 exceed GB MLs for other foods such as cereals for direct consumption.

Table C3 presents results for soy protein products, these included tofu and meat substitutes. Three of the nine products contained at least one mycotoxin, while the other six contained no mycotoxins above the LOQ. The three samples that contained mycotoxins, all soy mince products, contained different mycotoxins, but at low concentrations. S24-069322 contained low levels of AFB1 and OTA, S24-0698890 contained 1.85 µg/kg OTA, and sample S24-069321 contained FB1, FB2 and FB3, 31.8 µg/kg in total.

Table C4 presents results for pea protein based products, these were all meat substitute products. As with the other samples there was a low incidence of low concentrations of a small number of mycotoxins, only 4 residues of mycotoxins were found above the LOQ. These were detected in four of the six samples, two samples did not contain any mycotoxins. Three samples contained OTA at levels of 0.26 µg/kg (S24-069888), 0.53 µg/kg (S24-069886) and 0.54 µg/kg (S24-069889). One sample (S24-069885) contained 1.75 µg/kg ZON. None of these concentrations are close to or above any MLs.

Table C5 presents the results for the other samples that were based on combinations of soy, pea and wheat proteins. One sample contained no detectable mycotoxins, the other four samples contained at least one mycotoxin. DON was most frequently detected, it was found in four out five samples at levels from 2.85 to 7.35 µg/kg (all of these samples contained wheat protein). For one of the soy and wheat mixed products (S24-069325) DON was the only mycotoxin detected. Sample S24-069891 also contained OTA and S24-069328 also contained ZON. Sample S24-069326, a pea, wheat and soy mixed product contained DON, OTA and STG at low concentrations. This was the only sample that contained STG. 

Table C6 summarises the QC data that was obtained during the analysis of the Phase 1 survey samples. All QC criteria for LC-MS analysis were met, and the results show acceptable recovery across three concentrations for all the different types of products tested. The lowest recovery was found for the 1xLOQ OTA spike in protein powder due to the background concentration in this sample as described above. 

The results for the mycotoxins in protein products from Phase 2 are given in Table C7 to Table C9. 

Table C7 presents the results of the nine protein powder products analysed in Phase 2. As in Phase 1 OTA was the most frequently detected mycotoxin, it was detected in 8 of the 9 samples at concentrations ranging from 0.37 to 2.27 µg/kg. This was lower than the maximum level found in Phase 1. The two highest levels were found in pea protein powders. The next most frequently detected mycotoxin was AFB1, 4 samples contained concentrations from 0.11 to 0.63 µg/kg. None of these levels were at or above GB MLs set for OTA or AFB1 in other foods such as wheat gluten, nuts or cereals and are well below the EU ML for OTA in soybeans. 

Three samples contained ZON, levels were 1.06 to 2.80 µg/kg, all three samples contained pea protein. One of these samples also contained fumonisin B1 and B2 at 14.91 and 5.89 µg/kg. 

Sterigmatocystin was detected in 2 samples (0.20 and 1.55 µg/kg), the higher level was found in the sample that contained pea and rice protein. 

Table C8 presents the results of the meat replacement products. Again OTA was most frequently detected, it was found in 4 out of 10 samples. The levels measured were all below 0.5 µg/kg (range 0.13 to 0.32 µg/kg). One wheat and pea based product contained DON at 11.22 µg/kg. ZON was detected in two samples, one at 1.88 µg/kg, the other residue of 1.28 µg/kg was not confirmed, the uncorrected value was just at the LOQ of the method. Both products were soy based. 

Table C9 presents the results of ready to drink protein shakes and protein and cereal bars. DON was the most frequently detected mycotoxin, it was found in 2 out of 3 shakes and 2 out of 4 protein bars. The levels measured ranged from 2.32 to 14.55 µg/kg. Three samples contained OTA (all protein and cereal bars), levels were all below 1 µg/kg. Three samples contained ZON, one shake (2.08 µg/kg) and 2 protein bars (1.67 and 4.85 µg/kg). 

The results of the QC for these analyses are presented in Table C10. Spike recoveries met the requirements for Commission Regulation EC (No) 401/2006 for all analytes in all products (19). Sterigmatocystin had slightly lower average recoveries of 60-66%, but these are still within the range of 50-130% deemed acceptable in some cases, e.g. where low concentrations or no residues are measured or the mycotoxins are not regulated (19, 21, 22).

In-house validation according to the protocol Application for Flexible scope was carried out for mixed plant protein products. The results are given in Table C11. No analytical standard is available for β-ergocryptine, therefore there are no results for this during validation. However, this compound occurs in naturally contaminated samples and is separated from α-ergocryptine, where present it is quantified using α-ergocryptine and is included in the sum value with the other 11 epimers to calculate the sum of ergot alkaloids. 

The validation data show acceptable recovery and precision for all analytes included in the sum parameter at all three spiking levels. Samples were spiked at 1x, 2x and 10x target LOQ. In this case the target LOQ was 0.5 µg/kg for each epimer. 

Five samples were analysed for ergot alkaloids, these were the samples that included wheat protein as an ingredient (Table C12). Two samples were found to contain ergot alkaloids, one contained nine epimers giving a sum of ergot alkaloids of 6.90 µg/kg, eight epimers were detected in the second sample giving a sum of ergot alkaloids of 5.71 µg/kg. The reporting limit for individual compounds was reduced to 0.25 µg/kg, as three compounds were detected between this level (the lowest calibration point) and the validated LOQ (0.5 µg/kg). The residues of ergocorninine, ergocristine and α- and β-ergocryptinine were at concentrations between 0.25 and 0.5 µg/kg, all three residues were confirmed by ion ratio so were accepted as quantitative results. 

All samples from Phase 2 were analysed for ergot alkaloids. 

Table C13 presents the results for the ready to drink shakes and protein and cereal bars. No residues of ergot alkaloids were detected in any of the three ready to drink shakes analysed. 

Two of the four protein bars contained ergot alkaloids, one containing a sum total of 3.41 µg/kg and the other 11.01 µg/kg. Both products contained wheat (flour and or gluten) as ingredients, so it is expected the ergot alkaloid content originated from that. 

Table C14 contains the results of the plant protein powder products. Of the nine samples tested only one sample, S25-037301, contained detectable ergot alkaloids. Trace levels (between the lowest calibration standard and the lowest validated level, or reporting level) were found for 2 ergot alkaloids (ergocristine and ergosine). The sum ergot alkaloids (sum of 12 compounds) value for this sample was 0.56 µg/kg. This is below the normal RL but the residues were confirmed by mass spectrometry and so the result has been reported.

Table C15 contains the results of the meat replacement products. One of the ten samples analysed contained detectable levels of 6 ergot alkaloids, 2 of these were above the reporting level (RL), and 4 were between the RL and the lowest calibration standard. The sum value for ergot alkaloids found in this sample was 2.81 µg/kg. This product’s ingredients listed vegetable protein (17%) (wheat, pea) as the second ingredient, wheat gluten is also listed in the ingredients this. Again the inclusion of wheat is the most likely source of the ergot alkaloids. 

There are no MLs for ergot alkaloids in GB. There are MLs in force in the EU (4), which also apply in Northern Ireland, but there are no specific MLs not for plant protein products. The lowest ML in force is 20 µg/kg for processed cereal based food for infants and young children. The levels found in the products in this study were well below this. 

The in-house method FSG 827 for tropane alkaloids is accredited for cereals, herbal teas and vegetable products. However, as the samples to be tested in the survey were different to the vegetable mixes tested during initial validation some verification analyses were carried out to ensure the method was suitable. The results of these analyses are given in Table C16. A variety of sample types were analysed. For each sample type two blank (unspiked portions) and a sample spiked before extraction at 10 µg/kg were extracted and cleaned up. One of the blank portions was spiked at a level equivalent to 10 µg/kg after clean-up, this was labelled ‘tissue standard’, the sample spiked before analysis was the ‘spike’. The use of the tissue standard allows any effects (e.g. signal enhancement or suppression) due to that matrix to be observed. A calculation of the ratio of the recovery measured in the ‘spike’ sample to the recovery in the ’tissue standard’ can be made to assess true recovery. The results in Table C16 showed that there was some signal enhancement in the tissue standards, however these recovery results were generally within the acceptable range of 70-120% and none were greater than 130% which can be deemed acceptable (22). The method uses isotopically labelled internal standards for both atropine and scopolamine which corrects for any losses or enhancement. Therefore, during analysis of the survey samples only spiked samples were used for QC. 

The results of the survey samples are presented in Table C17. Four samples contained tropane alkaloids above the individual LOQs of 0.1 µg/kg. The LOQ for method FSG 827 is lower than the combined method used in Phase 2 as the final sample extract injected on the LC-MS/MS is more concentrated. 

Two samples contained only atropine (at 0.19 and 0.2 µg/kg, S24-069325 and S24-069323 respectively), while two samples contained both atropine and scopolamine. One sample (S24-069322) contained 0.4 µg/kg atropine and 0.14 µg/kg scopolamine. The other sample (S24-069883) contained 1.23 µg/kg atropine and 0.48 µg/kg scopolamine. All four samples were soy based products, the sample with the highest concentration (S24-069883) was a tofu sample. Sample S24-069322 was a soy meat free mince, sample S24-069323 was soy based vegan ‘chicken’ pieces. Sample S24-069325 was a soy & wheat product described as a vegetarian shredded duck. 

There are no MLs for tropane alkaloids in plant protein products, the only MLs in force in GB are for baby food and processed cereal based food for infants and young children, where MLs are set at 1.0 µg/kg each for atropine and scopolamine (2). 

The results of QC samples analysed with the Phase 1 survey samples are given in Table C18. The individual recovery values were all within the acceptable range of 70-120% apart from the lowest concentration spikes for the pea/soy/wheat mixed product. All other QC, including blanks, and LC-MS parameters (linearity, residuals, ion ratios and IS response) met criteria specified in FSG 002 Quality Control for LC-MS.

Table C19 presents the results of the acrylamide analysis of the Phase 1 products. Of the 26 samples analysed only two contained acrylamide above the LOQ of 30 µg/kg. Both were samples of meat replacement products. One sample (S24-069324) was a burger style product based on pea protein that contained 36.8 µg/kg. The other product was a vegan sausage product (S24-069891) made from soy and wheat, it contained 41.7 µg/kg. 

There are no MLs for acrylamide but there are Benchmark levels (BMLs), these are designed to help food business operators judge the success of any mitigation measures and are not intended to be used for enforcement. The BMLs range from 40 µg/kg for baby foods, processed cereal based for infants and young children to 4000 µg/kg for chicory based coffee substitutes (5). 

The two samples that contained residues both contained acrylamide just below and just above the BML of 40 µg/kg for baby food. The majority of samples (24 out of 26) did not contain acrylamide above the LOQ of 30 µg/kg. 

Samples were analysed for Fusarium mycotoxins using method FSG 818. The samples are extracted by a common extraction and clean-up method and analysed by LC-MS/MS. Due to the chemical differences in the analytes the samples have to be run on LC-MS/MS twice using both positive and negative ionisation mode to ensure the best analytical performance for all the analytes.

The results for plant protein powders and protein bars are presented in Table C20 and Table C21. Zearalenone was the only mycotoxin that was confirmed in any of these samples. Two protein powders (S25-037300 and S25-037454) contained ZON at a level just below the LOQ (both contained 2.3 µg/kg). Two protein bars also contained residues of ZON that were confirmed by ion ratio, one at 1.7 µg/kg (S25-037304) and the other at 4.6 µg/kg (S25-037453).

These results are very similar to the ZON results found for the same samples using the 11+ IAC method (Section 8.3). The results found using the IAC method for protein powder samples S25-037300 and S25-37454 were 2.8 and 2.56 µg/kg respectively. While for the protein bars S25-037304 and S25-037453 the ZON levels measured by the IAC method were 1.67 and 4.85 µg/kg respectively. These results show good agreement and give confidence in the levels measured. 

One residue of fusarenon X (22.5 µg/kg) was found in sample S25-037304 (a vegan protein bar) but the ion ratio did not confirm the identity of the peak so this result is indicative only. 

The results for the ready to drink shakes and meat replacements are given in Table C22 and Table C23. The only mycotoxin residue detected in these samples was DON at 12.6 µg/kg, in sample S25-037319, a sample of plant based steak strips. Again a similar result was found using the IAC method, 11.2 µg/kg DON was found in the same sample by that method. 

None of the samples contained any Fusarium mycotoxins at levels close to MLs in force for other foodstuffs. 

Results for enniatins, beauvericin and fusaric acid are presented in Table C24 to Table C26. None of these mycotoxins were detected in the ready to drink shakes. 

All four of the protein bar samples contained fusaric acid, although the concentration in 3 of the samples was below the LOQ of 10 µg/kg. The concentrations found were 4.8, 6.1 and 8.9 µg/kg. These levels were above the lowest calibration standard and confirmed by ion ratio so have been reported for information. The sample that contained the highest concentration of fusaric acid (S25-037303) at 26.9 µg/kg, did not contain any of the other compounds. The other 3 protein bar samples contained different combinations of beauvericin and the enniatins. Sample S25-037453 contained beauvericin, enniatin A1, B and B1 as well as fusaric acid. 

Table C25 contains the results of the plant protein powders. One sample contained no detectable residues of the mycotoxins, one sample contained only a low level of enniatin B (below the LOQ). One sample contained beauvericin (below LOQ) and enniatin B1, however this did not confirm by ion ratio. The other 6 samples each contained at least 3 of the 6 mycotoxins tested by this method. The majority of the residues found for enniatins and beauvericin were at or below the LOQ. The identity of most of the mycotoxins detected was confirmed by ion ratio. However, for four residues of enniatin B1 the analyte response was not confirmed and the results are reported as indicative. Spiked samples, that contained a known added amount of the analyte showed a similar effect where the peak response was not confirmed by ion ratio. It is possible there was an underlying interference for enniatin B1 for this matrix that affected the ion ratio response. 

One sample (S25-037454) contained a large residue of fusaric acid that was above the maximum of the calibration line (250 µg/kg) and therefore could not be quantified accurately. The analytical recovery for fusaric acid for protein powders was also low (average 36.6 %). When this correction factor was applied to the results this gave a value of >1000µg/kg. This is an indicative result. There are no MLs or guidance values for fusaric acid and very little information regarding the toxicological relevance of this toxin and therefore whether this concentration is significant. For interest it may be worthwhile testing this sample further, using step wise dilutions and matrix matched calibration to allow accurate quantification of the fusaric acid concentration. This would allow accurate quantification and address the issues of low recovery and matrix suppression that were observed during the initial analysis.

Table C26 contains the results of the meat replacement products. These results were similar to other products tested. There were several samples that contained low levels of beauvericin, enniatins and fusaric acid at concentrations just below the LOQ. In all cases the identity of the compounds was confirmed by ion ratio. Two samples contained enniatin B1, at levels of 2.6 and 6 µg/kg that did not confirm by ion ratio. All ten samples contained a residue of at least one mycotoxin, four samples contained fusaric acid only. One sample contained fusaric acid at 118.4 µg/kg. This result was confirmed by ion ratio. Again, in the absence of MLs in any products that could be used as a comparator it is not possible to comment on the significance of this result.

Table C27 contains the pyrrolizidine alkaloid (PA) results for the meat replacement and ready to drink shake samples. No residues of PAs were found above the LOQ of 1 µg/kg for the analytes reported. 

Table C28 contains the PA results for the protein powders and protein bars. No residues of PAs were found above the LOQ of 1 µg/kg for the analytes reported.

Some recovery values were low for some analytes, particularly for seneciphylline, and outside the normal range deemed acceptable (50 – 130%) (22). However, the lower end of the calibration range used for the analysis is much lower than the reporting limit (LOQ). This give confidence that despite the low recovery if residues were present at or above 1 µg/kg they would have been detected. The EU Regulation on method performance (18) stipulates that a method used for PAs in dry products should have an LOQ of ≤ 10 µg/kg, so the method used complied with this. 

Table C29 presents the tropane alkaloid results for the meat replacement and ready to drink shake samples. Atropine and scopolamine were not detected in the shake samples. Two meat replacement samples, S25-037318 and S25-037339, contained atropine at 0.52 and 1.15 µg/kg respectively. The sample that contained the lower concentration also contained 0.25 µg/kg scopolamine, although this was not confirmed by ion ratio. 

The tropane alkaloid results for protein powders and protein bars are presented in Table C30. One protein bar, S25-037453, contained 0.79 µg/kg atropine and 0.38 µg/kg scopolamine, the other three samples did not contain tropane alkaloids above the LOQ. 

Three protein powders contain scopolamine at 0.25, 0.45 and 0.5 µg/kg, although none of these residues were confirmed by ion ratio. 

The LOQ for tropane alkaloids was slightly higher for this method than the method used in Phase 1 (0.25 µg/kg compared to 0.1 µg/kg). This is because they were analysed in a combined method with the PAs. For this method, a smaller amount of sample is taken through clean-up, and the final volume for analysis by LC-MS/MS is larger as a compromise to improve the method performance and reduce suppression and interferences for the PAs that are also in the method. However, the method still meets the LOQ requirements for tropane alkaloids of 1 µg/kg each for atropine and scopolamine as set out in Regulation (EU) 2023/2783 (22). 

The results of the method verification for the Alternaria toxin method are summarised in Table C31. For all 5 toxins in the method, altenuene, alternariol, alternariol monomethylether, tentoxin and tenuazonic acid average recoveries were in the range 96.7 – 109.8% across three spiking levels. The only exception was for the LOQ spikes for altenuene where the average recovery was 126.2%, this is still within the range of 50 - 130% that can be deemed acceptable (21). 

The relative standard deviation values (RSDr or cv) were all in the range 1.3 to 9.1%, well below the value of 20% stipulated in the EURL-MP guidance document (19) and the recently implemented EU Regulation (EU) 2023/2782 that lays down methods for sampling and analysis for mycotoxins (21). This Regulation does not apply in GB, but it formalises method performance criteria in the EURL-MP guidance document (19) and is generally recognised as setting out good practice for laboratory performance.

The results of the analysis for Alternaria toxins are presented in Table C32. Low concentrations of tenuazonic acid (from 3.8 to 6.5 µg/kg) were found in all three protein shake samples. Two of the three results found were below the validated LOQ for the method, but the concentration was above the lowest calibration solution and the ion ratio confirmed so the results have been reported for information. 

No other Alternaria toxins were detected above the LOQ in any of the shake samples. 

Two of the four protein bars contained tenuazonic acid at levels of 11.4 µg/kg (S25-037303) and 18 µg/kg (S25-037453). No other Alternaria toxins were detected. 

Three of nine protein powders contained no detectable Alternaria mycotoxins. Of the other six samples one contained a low level of tenuazonic acid below the LOQ. One sample contained a level of alternariol just above the LOQ, at 1.3 µg/kg (S25-037455), but this was not confirmed by ion ratio. One sample contained a similar unconfirmed level of alternariol and 5.4 µg/kg tenuazonic acid (S25-037454). Two samples contained both alternariol and alternariol monomethylether at 5.6 and 3.4 µg/kg (S25-037299) and 13.5 and 6.1 µg/kg (S25-037298) respectively. The final sample contained alternariol (4.1 µg/kg), alternariol monomethyl ether (2.3 µg/kg) and tenuazonic acid (5.6 µg/kg).

Of the meat replacement products, 3 out of 10 samples did not contain Alternaria toxins. Six samples contained tenuazonic acid at levels from 5.2 to 62.7 µg/kg, five of these samples did not contain other Alternaria toxins, but one sample that contained 22.3 µg/kg tenuazonic acid also contained 1.1 µg/kg alternariol (S25-037320). Sample S25-037318 that contained the highest concentration of tenuazonic acid (62.7 µg/kg) was a pea based plant burger. One sample contained 5.0 µg/kg alternariol and 2.1 µg/kg alternariol monomethylether (S25-037341). 

There are no regulations in force for Alternaria toxins, so there are no MLs to compare these results to. However, these findings are similar to results reported in other published studies (6-10).

Table C33 presents the results of the citrinin analysis of the Phase 2 protein products. Citrinin was not detected in any of the samples above the LOQ of 2.5 µg/kg. The survey samples were extracted in two analytical batches: Batch PC25-04884 – protein powders and protein bars and Batch PC25-04988 – meat replacement and ready to drink shakes. Quality control samples matched to the products being analysed were included in each batch. The mean analytical recovery for each batch was 99.3 % and 100.9 %.

The results of the erucic acid analysis are presented in Table C34. 

Regulation EC (No) 1881/2006 sets MLs for erucic acid content of oil (2). In this case the ML of 20 g/kg in vegetable oils and fats placed on the market for the final consumer or for use as an ingredient in food would apply. The samples selected for analysis of erucic acid all contained rapeseed oil as ingredient as listed on the sample label. Ten samples were analysed, the levels of erucic acid found ranged from 0 to 6 g/kg of the fat / oil. These are all well below the ML of 20 g/kg for vegetable oils and fats placed on the market for the final consumer or for use as an ingredient in food.