Pyrrolizidine alkaloids in herbal tea and honey Report on the 2017 Proficiency testing scheme Breidbach, A. Tamošiūnas, V. 2017 EUR 28960 EN This publication is a Technical report by the Joint Research Centre (JRC), the European Commission’s science and knowledge service. It aims to provide evidence-based scientific support to the European policymaking process. The scientific output expressed does not imply a policy position of the European Commission. Neither the European Commission nor any person acting on behalf of the Commission is responsible for the use that might be made of this publication. Contact information Name: Andreas Breidbach Address: Retieseweg 111, 2440 Geel, Belgium Email: [email protected] Tel.: +32 (0)14 571205 JRC Science Hub https://ec.europa.eu/jrc JRC109959 EUR 28960 EN PDF ISBN 978-92-79-77167-5 ISSN 1831-9424 doi: 10.2760/219405 Luxembourg: Publications Office of the European Union, 2017 © European Union, 2017 Reuse is authorised provided the source is acknowledged. The reuse policy of European Commission documents is regulated by Decision 2011/833/EU (OJ L 330, 14.12.2011, p. 39). For any use or reproduction of photos or other material that is not under the EU copyright, permission must be sought directly from the copyright holders. How to cite this report: Breidbach A, Tamošiūnas V, Pyrrolizidine alkaloids in herbal tea and honey—Report on the 2017 Proficiency testing scheme, EUR 28960 EN, Publications Office of the European Union, Luxembourg, 2017, ISBN 978-92-79-77167-5, doi 10.2760/219405, JRC109959. All images © European Union 2017, except: Title By I, Tony Wills, CC BY 2.5, https://commons.wikimedia.org/w/index.php?curid=3323872 268-PT Accredited by the Belgian Accreditation Body (BELAC) Executive Summary Pyrrolizidine alkaloids (PAs) and their N-oxides (PANOs) are plant toxins which can enter the food chain through different paths. Two affected foods are herbal infusions and honey. This proficiency testing scheme was executed to assess the capabilities of laboratories to determine PAs. 29 laboratories from nine EU Member States plus Singapore registered. On 04. and 06.09.2017 test items and documentation were dispatched to all of those laboratories. By the dead line of 24.10.2017 26 laboratories had reported back results and filled in a questionnaire. Test item HO (acacia honey) was fortified with six PAs/PANOs (Echimidine, Integerrimine, Intermedine, Senecionine, Seneciphylline-NO, and Senkirkine) and 23 laboratories reported results for this item. The same number of laboratories reported for test item HT (herbal infusion) which was naturally contaminated with four PAs after extraction under reductive conditions (Integerrimine, Retrorsine, Senecionine, and Senecivernine). Laboratories had to report the sums of PA and its respective PANO. Satisfying outcomes could only be registered for Senecionine in test item HT and for Echimidine, Intermedine, and Senkirkine in test item HO with 74 %, 85 %, 85 %, and 91 %, respectively, of reported results having a z'-score smaller or equal to |2|. Only four laboratories reported for Integerrimine in both test items. Contrary to test item HT, Senecionine analysis in test item HO showed very unsatisfactory results. Of the 22 z'-scores calculated for Senecionine nine (41 %) were larger than 3. Senecivernine measurements in test item HT showed a similarly unsatisfying outcome with 47 % of reported results having z'-scores larger than 3. Only three laboratories out of the 26 were able to test for all 10 measurands and only one reported all 10 values with z'-scores smaller or equal to |2|. Overall only five laboratories obtained satisfactory z'-scores (≤ |2|) for all their reported results. There are two groups of three isomeric PAs/PANOs each which apparently caused, for a number of laboratories, problems with quantification. This is an issue which deserves heightened attention. The questionnaire contained queries regarding accreditation and experience, preparation conditions for the two test items, chromatographic separation conditions, detection conditions, calibration approach, and a comments section. The answers were evaluated and for selected questions their correlation to the z'-score of Senecionine in test item HO or Senecivernine in test item HT was analysed. For none of the tested questions a significant influence could be shown. 2 Table of Contents Executive Summary ................................................................................................................................. 2 1. Introduction .................................................................................................................................... 4 2. Scope ............................................................................................................................................... 4 3. Set-up of PT scheme ....................................................................................................................... 5 3.1. Time frame .............................................................................................................................. 5 3.2. Confidentiality ......................................................................................................................... 5 3.3. Distribution ............................................................................................................................. 5 3.4. Instructions to participants ..................................................................................................... 5 3.5. Standard deviation for proficiency assessment ...................................................................... 5 4. Test items ........................................................................................................................................ 6 4.1. Preparation ............................................................................................................................. 6 4.2. Homogeneity ........................................................................................................................... 6 4.3. Stability ................................................................................................................................... 6 5. Assigned values and their uncertainties ......................................................................................... 6 6. Evaluation of the results ................................................................................................................. 7 6.1. Test item HT ............................................................................................................................ 7 6.2. Test item HO ........................................................................................................................... 8 6.3. "Smaller than" / "Larger than" values and uncertainties ..................................................... 10 6.4. The questionnaire ................................................................................................................. 11 7. Conclusions ................................................................................................................................... 13 8. Acknowledgements ....................................................................................................................... 13 References: ........................................................................................................................................... 13 Annex 1: Instructions for PT scheme .................................................................................................... 14 Annex 2: Proficiency testing materials receipt form ............................................................................ 17 Annex 3: Homogeneity testing ............................................................................................................. 19 Annex 4: Stability testing ...................................................................................................................... 20 Annex 5: Determination of the assigned values and their uncertainties ............................................. 22 Annex 6: Graphs .................................................................................................................................... 26 Annex 7: Questionnaire data ................................................................................................................ 32 Annex 8: Participating laboratories ...................................................................................................... 44 3 1. Introduction Pyrrolizidine alkaloids (PAs) are secondary metabolites of plants used as defence against plant eating animals. It is estimated that over 6000 plant species worldwide, mainly from the botanical families Boraginaceae (e.g. Heliotropium spp.), Asteraceae (e.g. Senecio spp.) and Fabaceae (e.g. Crotalaria spp.), produce at least 600 different PAs. Those with an unsaturated bond in position 1,2 of the pyrrolizidine ring system (Figure 1) are hepatotoxic. Additional diversity is created due to oxidation of the pyrrolizidine nitrogen to the N-oxide (PANO) which doubles the possible number of PAs. Figure 1: Structural features of PAs. (A) core structural motif pyrrolizidine (1,2,3,6,7,8-hexahydro-5Hpyrrolizine);(B) general description of the main necine base parts of naturally occurring PAs including the common necine base numbering; (C) necine base otonecine; a core structural motif of otonecine-type PAs; (D) general pyrrolizine structure motif and (E) structural example of 1,2-unsaturated ester PA senecionine (Figure taken from [1]) PAs may enter the food chain via different routes. Two possibly affected end products are teas / herbal infusions, if PA containing weeds had contaminated the leafs / herbs, and honey, if PA containing pollen was collected by the bees. In 2016 the European Food Safety Authority (EFSA) published a new scientific report about dietary exposure to PAs [1]. In that report teas / herbal infusions were identified as the largest contributors to total PA exposure. It has been recommended that for relevant food commodities, as honey and teas / herbal infusions, the effort to collect PAs occurrence data should be continued. In this PT scheme a naturally contaminated rooibos herbal blend for infusion and a spiked acacia honey were sent out to 29 laboratories for PAs analysis. The two test materials were prepared, tested for homogeneity, stability, and characterized with respect to contamination level in-house. This report presents the results of these tests and characterizations as well as the reported results from the participating laboratories. 2. Scope This PT scheme was executed to assess the capabilities of the participating laboratories to determine one or more of the following PAs and/or their corresponding PANOs: Echimidine, Erucifoline, Europine, Indicine, Integerrimine, Intermedine, Jacobine, Lasiocarpine, Lycopsamine, Monocrotaline, Retrorsine, Senecionine, Seneciphylline, Senecivernine, Senkirkine, Trichodesmine, in an herbal blend for infusion and acacia honey. 4 3. Set-up of PT scheme 3.1. Time frame The PT registration was open from 19.05.2017 to 15.08.2017. Registration could be done through a dedicated web site. The test materials were dispatched to the participants on 04.09. and 06.09.2017. All shipments reached their respective recipients within 2 days. The initial deadline for reporting the results was 20.10.2017. This deadline was eventually extended to 24.10.2017 because two laboratories reported problems with entering the results. 3.2. Confidentiality The procedures used for the organisation of PTs are accredited according to ISO 17043:2010 [2] and guarantee that the identity of the participants and the information provided by them is treated as confidential. 3.3. Distribution In all, 29 laboratories from nine EU Member States plus Singapore registered for participation. The following was sent to each of them: • one container with 40 – 50 g of a herbal blend for infusion naturally contaminated with PAs; • one container with 50 – 70 g of acacia honey spiked with PAs • the "Instructions for PT scheme" letter with the individual reporting password (Annex 1: Instructions for PT scheme); • the "Proficiency testing materials receipt form" (Annex 2: Proficiency testing materials receipt form); • a "Submission of results" document explaining the use of the reporting interface. 3.4. Instructions to participants The "Instructions for PT scheme" letter included with the test item shipment gave detailed instructions to the participants. In brief, the participants were informed that one or more of the following PAs: Echimidine, Erucifoline, Europine, Indicine, Integerrimine, Intermedine, Jacobine, Lasiocarpine, Lycopsamine, Monocrotaline, Retrorsine, Senecionine, Seneciphylline, Senecivernine, Senkirkine, Trichodesmine, could be present in either one of the two test items. The participants had to check the test items for integrity and store them upon receipt at temperatures between -15 °C and -25 °C. Instructions on how to access the reporting web site and how to enter results were also included. 3.5. Standard deviation for proficiency assessment Scoring of the reported results was to be done acc. to ISO 13528:2015 and was to be based on an independently determined reference value (x ) for each PA and a target standard deviation PT σ of 0.22 x acc. to the following equation: PT PT (cid:3)(cid:4)(cid:5) −(cid:4)(cid:7)(cid:8)(cid:9) (cid:1) = (cid:10)(cid:7)(cid:8) 5 with x the reported result of an individual PA from a laboratory. No scoring was to be provided i for reported sums of total PAs. For those analyte contents for which the associated uncertainty of x (u(x )) was larger than PT PT 0.3 σ the following equation was used: PT (cid:11) (cid:3)(cid:4)(cid:5) −(cid:4)(cid:7)(cid:8)(cid:9) (cid:1) = (cid:11) (cid:10)(cid:7)(cid:8) with the adjusted target standard deviation . (cid:11) (cid:13) (cid:13) (cid:10)(cid:7)(cid:8) = (cid:12)(cid:10)(cid:7)(cid:8) +(cid:15) (cid:3)(cid:4)(cid:7)(cid:8)(cid:9) 4. Test items 4.1. Preparation The herbal infusion blend (test material HT, mostly rooibos) was purchased from a provider within the EU. It was milled with a ZM200 centrifugal mill with a 0.5 mm sieve (Retsch, Hahn, Germany). After thorough homogenization the material was filled into 96 plastic screw-capped containers. The acacia honey (test material HO) was courtesy of Breitsamer + Ulrich GmbH & Co KG (München, Germany). The honey was practically free of PAs (none detected above 3 µg/kg) and spiked with Echimidine, Integerrimine, Intermedine, Senecionine, Seneciphylline-NO, and Senkirkine. It was then thoroughly mixed with a ploughshare mixer (Lödige, Paderborn, Germany) and filled into 73 plastic screw-capped containers. 4.2. Homogeneity For testing the homogeneity ten containers of HT and eight containers of HO were selected. Test portions were extracted/cleaned-up with SPE acc. to an internal protocol and measured in randomized order with a LC-HRMS. The obtained signals, without conversion to concentration units, were evaluated for sufficient homogeneity acc. to ISO 13528 Annex B [3]. More details can be found in Annex 3: Homogeneity testing. For all analytes in both test materials the results indicated sufficient homogeneity. 4.3. Stability The stability of the analytes in the two test materials was determined through an isochronous study. For this purpose randomly selected test units were stored frozen (-15 to –25 °C) and at 4 °C, and 40 °C for up to eight weeks. At the day of measurement all test units were prepared and measured as described for the homogeneity testing. More details can be found in Annex 4: Stability testing. All analytes in both test materials were sufficiently stable during the PT period. 5. Assigned values and their uncertainties For the scoring of this PT scheme individual mass fractions for all the PAs in the two test materials were determined. This was done with gravimetric standard addition using an internal standard 6 (ISTD) acc. to Hauswaldt et al [4]. These measurements were executed in early October for test item HT and early November for test item HO. This was due to time constraints and happened without prior knowledge of reported results. More details can be found in Annex 5: Determination of the assigned values and their uncertainties. The assigned values x and their associated standard uncertainties u(x ) were determined acc. to PT PT ISO 13528 clause 7 [3] and are listed in Table 1 for test item HT and Table 2 for test item HO. 6. Evaluation of the results 6.1. Test item HT Test item HT was naturally contaminated with four PAs/PANOs at detectable levels: Integerrimine, Retrorsine, Senecionine, and Senecivernine. Table 1 lists relevant parameters. The reference values (x ) are for the sum of PA/PANO after reductive extraction. Since the PT associated uncertainties of the reference values were larger than 0.3 σ in all cases the PT adjusted target standard deviation σ' was used for scoring. The robust mean (x̅ ) was PT rob computed from the reported results using Algorithm A [3, Annex C] following the Recommendation 1 in [5]. A robust mean could not be calculated for the reported results of Integerrimine because of too few results and for Senecivernine because of too high a contribution of minor modes to the multi-modal distribution of reported results. For the other two robust means, the one of Senecionine falls within the expanded uncertainty range (k=2, ~ 95 % confidence) around x , the one of Retrorsine does not. PT For the four PAs present in test item HT a satisfying outcome can only be attested for the analysis of Senecionine. It was present at the highest level and 17 out of 23 (74 %) reported results showed z'-scores ≤ |2|. No reported result had a z'-score > |3|. At the opposite end is the outcome for the measurements of Senecivernine with 47 % of reported results with z'-scores larger than 3. For Integerrimine only four laboratories reported a result, two of them as "smaller than", even though it was present at the second highest level in this naturally contaminated material. Table 3 is listing all reported results and the corresponding z'-scores. The figures in Annex 6: Graphs for Test item HT visualise these results. Integerrimine, Senecionine, and Senecivernine are three isomeric PAs with the elemental formula (C H NO ). This makes these three compounds practically indistinguishable for mass 18 25 5 spectrometry which was the detection mode used by all participating laboratories. Only proper chromatographic separation, which is challenging, can provide reliable results for the individual compounds. Insufficient separation between Integerrimine and Senecivernine, and false identification may have led to the large number of too high results for Senecivernine. A compounding factor may be that the majority of participating laboratories were apparently not testing for Integerrimine and, therefore, might not have been aware of any coelution issues. Integerrimine is not included in the list of PAs provisionally selected by the European Commission [1]. 7 Laboratory 14 reported to have also detected Senkirkine at 10 µg/kg. No other laboratory reported any other PA above its respective LOQ. Laboratory 3 reported the sum of all PAs as "retronecine equivalents" at 78 µg/kg. Table 1 Parameters of the four PAs and summary of the outcome in test item HT Integerrimine Retrorsine Senecionine Senecivernine x [µg/kg] 30.6 19.8 80.4 5.7 PT u(x ) [µg/kg] 2.8 1.5 7.8 1.2 PT x̅ [µg/kg] 15.4 65.4 rob σ [µg/kg] 6.7 4.3 17.7 1.2 PT u(x )>0.3 σ YES YES YES YES PT PT σ' [µg/kg] 7.3 4.6 19.3 1.7 PT Number reported 4 20 23 15 "<" or ">" 2 7 1 2 z' ≤ |2| 2 9 17 6 |2| < z' ≤ |3| 0 2 5 0 z' > |3| 0 2 0 7 6.2. Test item HO Test item HO was an acacia honey free of PAs. It was spiked with Echimidine, Integerrimine, Intermedine, Senecionine, Seneciphylline-NO, and Senkirkine. Table 2 lists relevant parameters. These parameters were determined as described for test item HT above. Again, for Integerrimine too few results were reported to calculate a robust mean. Of the other robust means, the ones for the content of Echimidine and Senkirkine lay within the expanded uncertainty ranges (k=2, ~ 95 % confidence) around the respective x , the ones for PT Intermedine, Senecionine, and Seneciphylline were below the lower limit of the expanded uncertainty ranges. Also, as for test item HT, the associated uncertainties of the reference values were larger than 0.3 σ in all cases but for Senecionine. Yet for reasons of consistency the PT adjusted target standard deviation σ' was used for scoring of all PAs including Senecionine. PT Table 2 Parameters of the six PAs and summary of the outcome in test item HO Echimidine Integer- Intermedine Senecionine Seneci- Senkirkine rimine phylline x [µg/kg] 70.8 54.4 43.9 29.1 12.9 62.1 PT u(x ) [µg/kg] 6.3 3.8 4.4 1.8 1.0 9.6 PT x̅ [µg/kg] 68.4 30.9 23.4 9.7 52.3 rob σ [µg/kg] 15.6 12.0 9.7 6.4 2.8 13.7 PT u(x )>0.3 σ YES YES YES NO YES YES PT PT σ' [µg/kg] 16.8 12.6 10.6 6.6 3.0 16.7 PT Number reported 20 4 20 22 22 23 "<" or ">" 1 2 1 1 3 1 z' ≤ |2| 17 2 17 9 13 21 |2| < z' ≤ |3| 1 0 2 3 3 0 z' > |3| 1 0 0 9 3 1 8