The certification of the aflatoxin mass fractions in peanut butter

World Mycotoxin Journal, August 2008; 1(3): 283-289 Wageningen g Academic P u b l i s h e r s The certification of the aflatoxin mass fractions in peanut butter G. Buttinger, S. Harbeck and R.D. Josephs* European Commission, Joint Research Centre, Institute for Reference Materials and Measurements, 2440 Geel, Belgium; *current address: Bureau International des Poids et Mesures, Section de Chimie, Pavillon de Breteuil, 92312 Sèvres, France; Gerhard.Buttinger@ec.europa.eu Received: 12 June 2007 / Accepted: 17 December 2007 2008 Wageningen Academic Publishers Abstract In the context of control activities contamination of food and feed with aflatoxins is a frequently observed non compliance. Pistachios, peanuts and products thereof are particularly affected. The Institute for Reference Materials and Measurements has therefore produced a peanut butter material certified for its aflatoxin mass fractions. This certified reference material (CRM) allows for the evaluation of analytical method performance and the assessment of the comparability of results from different laboratories. The CRM was produced using naturally contaminated raw materials to ensure equivalent behaviour compared to samples routinely encountered. The homogeneity and stability of the CRM were thoroughly tested and certified values were determined in an inter-laboratory study. Furthermore, uncertainties of the certified values were assessed including contributions of the homogeneity, stability and certification studies to the combined uncertainty. This newly prepared CRM allows an assessment of trueness of the analytical method at a concentration level corresponding to the legal limits enforced in the European Union. The material has the following certified properties: aflatoxin B 1 1.77±0.29 µg/kg, aflatoxin B 2 8±0.07 µg/kg, aflatoxin G 1 0.9± µg/kg, aflatoxin G 2 0.31±0.12 µg/kg and total aflatoxins, as sum of aflatoxins B 1, 3.5±0.5 µg/kg. Keywords: certified reference material, aflatoxin, peanut butter, groundnut 1. Introduction After mysterious turkey deaths on UK farms in the 1960s, the research for the suspect agent began. An aflatoxincontaminated feed was found to be the cause. This is where modern mycotoxin research began. Aflatoxins are the best studied group of mycotoxins. They are produced by Aspergillus strains under warm and humid conditions. Aspergilli are predominantly storage fungi, but pre-harvest infestations also occur in a wide range of agricultural commodities. Therefore, a contamination with aflatoxins can occur in nearly all plant-based food and feed leading to a health risk for humans and animals if present at sufficiently high concentrations. Aflatoxin B 1 are the most prevalent aflatoxin metabolites. Among these aflatoxin B 1 ( ) is considered to be the most harmful one. All four aflatoxins are categorised as group one carcinogens (carcinogenic to humans) by the International Agency for Research on Cancer (IARC, 2002). The principal tumours induced are hepatocellular carcinoma. The possible synergistic effects of a hepatitis B infection are still being investigated. Aflatoxins are not only carcinogenic but also genotoxic and immunosuppressant. To ensure public health and animal welfare most countries have set legal limits for and/or the sum of, aflatoxin B 2 (AFB 2 ), aflatoxin G 1 ( ) and aflatoxin G 2 ( ). 61 countries have set a limit for in food with the majority at a level of 2 or 5 µg/kg. 76 countries have established limits for the sum of and in food between 0 and 35 µg/kg with excess frequency at 4 µg/kg and 20 µg/kg. Regulations for aflatoxins in feed are existent in many countries whereas the legal limit depends very much on the intended use of the feed. The ISSN 1875-0710 print, ISSN 1875-0796 online, DOI 10.3920/WMJ2008.x038 283

G. Buttinger et al. lowest levels are set for and feed intended for dairy cattle, as this is where the greatest risk for humans lies, due to metabolism of to aflatoxin M 1 by the cattle and resulting contamination of the milk (FAO, 2004). The 2005 annual report by the Rapid Alert System for Food and Feed (RASFF) of the European Union (EC, 2006a) shows that 40% of the information notifications were due to mycotoxin contaminations. Around 95% of these notifications concerned aflatoxins. This underpinned the need for reliable and comparable measurement results to guarantee food safety and to facilitate international trade. As a result the Institute for Reference Materials and Measurements (IRMM) produced a peanut butter reference material (), certified for its aflatoxin mass fractions. The certified values are close to the regulatory limits set in Commission Regulation (EC) No 1881/2006, that is 2 µg/kg aflatoxin B 1 for groundnuts used as aliment (EC, 2006b). The importance of such a certified reference material (CRM) was indicated by the high demand for a previous material developed under the BCR programme (Gilbert et al., 1991). This article describes the production of, the certification of the aflatoxin mass fractions and the assessment of the individual contributions to the combined uncertainties of the certified values. The evaluation of the homogeneity and stability of the material, during storage as well as shipment, are discussed below. The design and evaluation of the certification study is outlined in detail. 2. Materials and methods Processing of the material was produced by blending uncontaminated peanut paste with contaminated peanut paste to a target mass fraction of about 4 µg/kg for the sum of, and. Therefore, 400 kg of Argentinean peanut kernels (mass fraction < 0.1 µg/kg) were processed into a paste and mixed with 2 kg Lecithin EMULFLUID E from Lucas Meyer (Hamburg, DE) to stabilise the emulsion. 35 kg of contaminated Chinese peanut kernels (mass fraction ~ 44 µg/kg) were roasted at 120 C in a drying cabinet and subsequently processed into a paste. 170 kg of the non-contaminated paste were mixed with 35 kg of the contaminated paste and homogenised by agitation. After assessment of the homogeneity of the bulk material, portions of ~100 g were filled in aluminium cans. The aluminium cans were flushed with nitrogen and the lid was heat sealed. Design of the homogeneity study 31 samples of the batch (~2% of total batch) were selected using a random stratified sample selection scheme to ensure on the one hand a random selection of the samples and on the other hand an equal distribution of selected samples over the whole batch. The samples were analysed in triplicate employing a method based on liquid chromatography with fluorescence detection (LC-FLD) and immunoaffinity column (IAC) clean up. Contributions to the uncertainty due to heterogeneity of the material were assessed using an ANOVA approach published by Linsinger et al. (2001a). Design of the stability studies Three stability studies were performed. A short-term stability study was performed to assess stability during transport and two long-term stability studies to assess stability during storage and to assess the contribution related to the stability assessment to the combined uncertainty of the certified value. All three studies were set up as isochronous studies (Lamberty et al., 1998). The short-term study lasted 4 weeks and was performed at 18 C, 40 C with -70 C as reference temperature. One long-term stability study was set up for 18 months at -20 C and 4 C and the second one for 36 months at -20 C, both with reference temperature -70 C. Data were evaluated as previously described in Van der Veen et al. (2001) and Linsinger et al. (2001b, 2004). Design of the characterisation study The characterisation study was performed as laboratory intercomparison between eight carefully selected laboratories. The laboratories had to prove their measurement capabilities and experience in this analytical field before being allowed to participate in the characterisation study. Each laboratory was provided with a common calibrant for and, three samples of BCR- 385R and six samples of peanut butter blank material. The laboratories had to perform the following measurements on three different days: calibration with the common calibrant, triplicate measurement of the blank material, triplicate measurement of the blank material spiked with common calibrant to a level of 0.5 µg/kg for each toxin and a duplicate measurement of. The results for were corrected for the daily recovery rate estimated with the measurements of the blank material and spiked blank material. 284 World Mycotoxin Journal 1 (3)

The certification of the aflatoxin mass fractions in peanut butter 3. Results and discussion Homogeneity study The measurements were performed on five different days. In order to exclude the influence of day-to-day variability the results of one day were normalised to the mean of results of this day. The normalised results on the five days were combined and analysed by ANOVA and linear regression. The normalised results are depicted in Figure 1. The slope of the regression was tested for its significance at a 95% level to evaluate trends eventually related to the analysis or filling sequence. No trends regarding filling or analysis sequence were found. Furthermore, the results were tested for their distribution and they all followed a unimodal if not a normal distribution allowing the ANOVA approach mentioned above. The results of the ANOVA evaluation are summarised in Table 1. Stability study The statistical evaluation of the results of all three stability studies showed that there is no significant change in concentration upon storage under the different conditions applied. Therefore, it was concluded that no special precautions regarding temperature control during shipment are necessary. The resulting contribution of possible transport instability to the total uncertainty budget was found to be negligible. Table 1. Evaluation of the homogeneity study (results normalised). BCR- 385R mean RSD [%] s wb 1 [%] s bb 2 [%] u* bb [%] u bb 3 [%] 1.00 4.2 7.3-4 1.8 1.8 AFB 2 1.00 4.1 7.7-4 1.9 1.9 1.00 16.6 17.5 13.2 4.3 13.2 1.00 17.9 21.5 12.8 5.3 12.8 1 Standard deviation within one bottle. 2 Standard deviation between bottles. 3 Higher value u * bb or s bb taken as contribution of heterogeneity. 4 Cannot be calculated as MS within > MS between. Although the material was stable at both temperatures in the 18 months study, -20 C was chosen as storage temperature as an additional precaution. The results of the 18 month and 36 month study were combined by normalising the results of each study by the means of the values for the samples stored at the reference temperature in each study. These combined results were used to estimate the contribution due to possible instability to the combined uncertainty of the certified value for a storage duration of 36 months. The contribution to the overall uncertainty due to possible instability (u lts ) is 4.8% for, 4.5% for AFB 2, 7.9% for and 11.5% for. The combined results of the two long-term stability studies with the uncertainty 2.0 Aflatoxin B 1 0.7 Aflatoxin B 2 Result [µg/kg] 1.8 1.6 Result [µg/kg] 0.6 0.5 1.4 0 400 800 1200 1600 Sample 0 400 800 1200 1600 Sample Result [µg/kg] Aflatoxin G 1 1.4 1.2 1.0 0.8 0.6 0 400 800 1200 1600 Aflatoxin G 2 0.5 0.3 0.2 0.1 0 400 800 1200 1600 Sample Sample Figure 1. Homogeneity data for aflatoxin B 1. Result [µg/kg] World Mycotoxin Journal 1 (3) 285

G. Buttinger et al. contribution due to possible instability (dashed lines) are shown in Figure 2. Certification study All laboratories used an IAC cleanup and reversed phase liquid chromatography with post-column derivatisation and fluorescence detection. Methods varied in extraction solvent and technique. For the post-column bromination either an electrochemical bromination with a Kobra cell was used or pyridinium hydrobromide perbromide was added (Table 2). All received data sets were subject to a technical evaluation. In the first step the recovery rate and precision were compared to the criteria laid down in Commission Regulation (EC) No. 401/2006 (EC, 2006c). Data not fulfilling the criteria on one day were omitted if the criteria could not be met on two days by the lab the whole data set for three days was omitted. In a second step the data were statistically evaluated using the Cochran test to determine outlying variances. As all laboratories used a similar method an outlying variance indicates poor repeatability of the laboratories and therefore a suboptimal control over Relative content [%] Relative content [%] 130 Aflatoxin B 1 120 110 100 90 80 70 0 12 24 36 48 Time [month] 140 Aflatoxin G 1 130 120 110 100 90 80 70 0 12 24 36 48 Time [month] Relative content [%] Relative content [%] 130 Aflatoxin B 2 120 110 100 90 80 70 0 12 24 36 48 Time [month] Aflatoxin G 2 160 150 140 130 120 110 100 90 80 70 0 12 24 36 48 Time [month] Figure 2. Data of combined long term stability study at -20 C for aflatoxin B 1. Table 2. Overview of analytical methods used for certification. Lab code Extraction solvent Extraction technique Defatting IAC producer Chromatography Derivatisation 1 methanol+water blender hexane R-biopharm rhône isocratic PBPB 1 2 methanol+water shaker hexane Vicam isocratic PBPB 3 methanol+water blender hexane Vicam isocratic Kobra cell 2 4 chloroform+water shaker hexane R-biopharm rhône isocratic Kobra cell 5 acetonitrile+water blender R-biopharm rhône isocratic PBPB 7 acetonitrile+water blender R-biopharm rhône isocratic Kobra cell 8 acetonitrile+water ultrasonic bath Vicam gradient Kobra cell 9 methanol+water blender hexane R-biopharm rhône isocratic PBPB 1 Bromination with pyridinium hydrobromide perbromide. 2 Electrochemical bromination with potassium bromide. 286 World Mycotoxin Journal 1 (3)

The certification of the aflatoxin mass fractions in peanut butter method performance by the laboratories. Datasets with outlying variances have subsequently been rejected. The remaining datasets have undergone a further statistical evaluation. The following tests have been performed to assess the significance between means of two laboratories, outlying means, homogeneity of the variances, the significance between and within laboratories variances and the distribution of the means: Scheffe s multiple t- test, Dixon test, Namilov t-test, Grubb s test, Bartlett test, SNEDECOR F-test, skewness and kurtosis test. The outcome of these tests is summarised in Table 3. The outlying mean of (with a level of significance of 0.05) lies within two standard deviations of the mean of the means and is therefore not significantly different from the mean of the means. The accepted individual lab means and their mean of means as well as the corresponding standard deviations are shown in Figure 3. Table 3. Summary of the statistical evaluation of accepted data. AFB 2 Number of data sets 7 6 6 4 Number of replicate measurements 38 36 36 24 Mean of means [µg/kg] 1.77 8 0.92 0.31 Standard deviation [%] 16.7 14.1 16.1 19.5 Relative standard error [%] 6.3 5.8 6.6 9.8 All data sets compatible two by two? (Scheffe s test) no no no no Outlying means? (Dixon test, Namilov t-test, Grubbs test) Lab2 Namilov no no no (P=0.05) Outlying lab variances? (Cochran test) no no no no Lab variances homogeneous? (Bartlett test) yes (P=0.01) yes yes (P=0.01) yes Distribution of means normal? (skewness & kurtosis, normal probability plot) yes yes yes yes Variances between labs significantly different? (SNEDECOR) yes yes yes yes P: Level of significance. Mass fraction [µg/kg] 2.8 2.4 2.0 1.6 1.2 Aflatoxin B 1 L1 L2 L3 L4 L5 L7 L9 Mean of means Laboratory Mass fraction [µg/kg] 0.7 0.6 0.5 0.3 Aflatoxin B 2 L1 L3 L4 L5 L7 L9 Mean of means Laboratory Mass fraction [µg/kg] 1.4 1.2 1.0 0.8 0.6 Aflatoxin G 1 L1 L2 L3 L5 L7 L9 Mean of means Laboratory Mass fraction [µg/kg] 0.5 0.3 0.2 Aflatoxin G 2 L2 L3 L7 L9 Mean of means Laboratory Figure 3. Laboratory means of accepted data sets for aflatoxin B 1. World Mycotoxin Journal 1 (3) 287

G. Buttinger et al. The mean of the means of the accepted data sets is the certified value. The standard error of the mean of the means (u char ) and the uncertainty of the mass fraction of the common calibrant (u cal ) were taken as the contribution of the certification exercise towards the combined uncertainty of the certified values. The certified values with their uncertainties and the individual contributions to these uncertainties are summarised in Table 4. Table 5. Proportionate contributions to the combined uncertainty of the mass fractions. u cal [%] u lts [%] u bb [%] u char [%] 10 34 13 43 AFB 2 8 34 14 44 6 27 45 22 3 33 36 28 Table 4. Certified values with their relative and standard uncertainty respectively. 4. Conclusions AFB 2 Certified value 1.77 8 0.9 0.31 [µg/kg] u lts [%] 4.8 4.5 7.9 11.5 u bb [%] 1.8 1.9 13.2 12.8 u char [%] 6.3 5.8 6.6 9.8 u cal [%] 1.4 1.0 1.7 1.0 u CRM [%] 8.2 7.6 16.8 19.8 U CRM (k=2) [%] 16.5 15.3 33.7 39.7 U CRM (k=2) [µg/kg] 0.29 0.07 0.34 0.12 The production of shows that, after nearly 50 years of aflatoxin analysis, it is still no easy task to measure aflatoxin contamination at low mass fraction levels. Therefore the availability of a CRM is of the utmost importance to allow method performance tests and enable comparability of results between laboratories. The certified values for are close to the threshold given in relevant European legislation. The legal limit for is 2 µg/kg and the certified value is 1.77±0.29 µg/kg. The legal limit for the sum of and is 4 µg/kg and the certified value for total aflatoxins (sum of and ) is 3.5±0.5 µg/kg. This allows verifying method performance at contamination levels of common interest. was planned, processed and certified in compliance with the ISO Guide 31, 34 and 35 (ISO, 2000a,b and 2006). The uncertainties of the certified values include, as required by ISO Guide 31 and 34, a contribution associated with a possible instability of the material at certain conditions but also a contribution coming from a possible heterogeneity of the material. Table 5 clearly indicate that the percentage uncertainty contributions associated with possible heterogeneity and instability are major contributions to the overall uncertainties of the mass fractions of and. Individual data on the production of this material will be published in a certification report for this material and made public via the IRMM s webpage. References European Commission (EC), 2006a. The Rapid Alert System for Food and Feed (RASFF) Annual Report 2005. Office for Official Publications of the European Communities, Luxembourg, Luxembourg. European Commission (EC), 2006b. Commission Regulation (EC) No 1881/2006 of 19 December 2006 setting maximum levels for certain contaminants in foodstuffs. Official Journal of the European Union L364: 524. European Commission (EC), 2006c. Commission Regulation (EC) No 401/2006 of 23 February 2006 laying down the methods of sampling and analysis for official control of the levels of mycotoxins in foodstuffs. Official Journal of the European Union L70: 12. Food and Agricultural Organisation of the United Nations (FAO), 2004. Worldwide regulations for mycotoxins in food and feed in 2003. FAO Food and Nutrition Paper 81. FAO, Rome, Italy. Gilbert, J., Wagstaffe, P.J. and Boenke, A., 1991. The certification of aflatoxins B 1, G 2 and total aflatoxins content of two peanut-butter reference materials CRM 385, 401. Office for Official Publications of the European Communities, Luxembourg, Luxembourg. International Agency for Research on Cancer (IARC), 2002. IARC monographs on the evaluation of carcinogenic risks to humans, volume 82 some traditional herbal medicines, some mycotoxins, naphthalene and styrene. IARCPress, Lyon, France. International Organisation for Standardisation (ISO), 2000a. Guide 31 Reference Materials Contents of certificates and labels. ISO, Geneva, Switzerland. International Organisation for Standardisation (ISO), 2000b. Guide 34 General requirements for the competence of reference material producers. ISO, Geneva, Switzerland. International Organisation for Standardisation (ISO), 2006. Guide 35 Reference Materials General and statistical principles for certification. ISO, Geneva, Switzerland. Lamberty, A., Schimmel, H. and Pauwels, J., 1998. The study of the stability of reference materials by isochronous measurements. Fresenius Journal of Analytical Chemistry 360: 359-361. 288 World Mycotoxin Journal 1 (3)

The certification of the aflatoxin mass fractions in peanut butter Linsinger, T.P.J., Pauwels, J., Lambery, A., Schimmel, H.G., Van der Veen, A.M.H. and Siekmann, L. 2001b. Estimating the uncertainty of stability for matrix CRMs. Fresenius Journal of Analytical Chemistry 370: 183-188. Linsinger, T.P.J., Pauwels, J., Van der Veen, A.M.H., Schimmel, H. and Lambery, A., 2001a. Uncertainty calculations in the certification of reference materials. 2. Homogeneity study. Accreditation and Quality Assurance 6: 26-30. Linsinger, T.P.J., Van der Veen, A.M.H., Gawlik, B.M., Pauwels, J. and Lambery, A., 2004. Planning and combining of isochronous stability studies of CRMs. Accreditation and Quality Assurance 9: 464-472. Van der Veen, A.M.H., Linsinger, T.P.J, Lamberty, A. and Pauwels, J., 2001. Uncertainty calculations in the certification of reference materials. 3. Stability study. Accreditation and Quality Assurance 6: 257-263. World Mycotoxin Journal 1 (3) 289