Collaborative test of the fluorimetric determination of selenium in a test solution, milk powder and bovine liver

Summary Results are reported for an inter-laboratory test conducted to assess precision (repeatability, reproducibility) and accuracy of a collaborative method for the fluorimetric determination of selenium (Se). The seven participating laboratories analysed one test solution, four samples of milk powder and two samples of freeze-dried bovine liver. Each set of samples comprised three duplicates: two colorant-disguised milk powders and one code disguised freeze-dried bovine liver. Two of the milk powders were enriched with 90.7 g/kg Se as seleno-dl-methionine. One set of results had to be rejected because the laboratory involved did not adhere to the collaborative method. Results from a second laboratory contained both stragglers and outliers. The five remaining laboratories performed the method satisfactory. Results from these laboratories were statistically evaluated according to ISO 5725. The average coefficient of variation within a laboratory (repeatability) was 4.8% and between laboratories (reproducibility) 6.0% for the milk powder and bovine liver samples. Recovery for the test solution, target value 120 g Se/1, was 96% and the average recovery for the Se enriched milk powder was 88%. The mean result for the milk powder was 98.9 g/kg (n=10), coefficient of variation (CV) 6.7%, and for Se enriched milk powder 178.3 g/kg, coefficient of variation (CV) 3.6%. For freeze-dried bovine liver, these results were 238.4 g/kg and 4.1 % respectively.

---

Ringversuch zur fluorimetrischen Bestimmung von Selen in einer Testlösung, in Milchpulver und in Rinderleber

Zusammenfassung Es wurden Ergebnisse eines Ringversuches beschrieben, dessen Ziel war, die Genauigkeit (Wiederholbarkeit und Vergleichbarkeit) sowie die Richtigkeit einer fluorimetrischen Untersuchungsmethode für Selen zu erfassen. Teilnehmer aus sieben Laboratorien analysierten je eine Testlösung, je zwei Proben Milchpulver und gefriergetrocknete Rinderleber. Jeder Probensatz enthielt drei Doppelproben; zwei davon waren mit Farbstoff maskierte Milchpulver, während die beiden Rinderleberproben sich nur in der Probenkodierung unterschieden. Je zwei Milchpulverproben waren mit Selen-dl-methionin, was 90,7 g/kg Selen entspricht, angereichert. Die Ergebnisse eines Laboratorium wurden nicht in die Auswertung einbezogen, da die gemeinsame Arbeitsvorschrift nicht eingehalten wurde. Die Analysendaten eines weiteren Laboratorium enthielten Ausreisser. Die Analysenwerte nach der gemeinsamen Arbeitsvorschrift durch die fünf übrigen Laboratorien wurden nach ISO 5725 statistisch ausgewertet. Für die Milchpulver-und Rinderleberproben war der Mittelwert des Variationskoeffizienten für die Wiederholbarkeit 4,8% und für die Vergleichbarkeit 6,0%. Die mittlere Wiederfindung für die Testlosung mit einem Richtwert des Selens von 120 g/l war 96% und für das mit Selen angereicherte Milchpulver 88%. Der mittlere Selengehalt von Milchpulver war 98,9 g/kg (N=10) Wiederholvariationskoeffizient 6,7%, und für den mit Selen angereicherten Milchpulver 178,3 g/kg Wiederholvariationskoeffizient 3,6%. Die entsprechenden Werte für gefriergetrocknete Rinderleber waren 238,4 g/kg und 4,1%.

---

---

This study has been carried out under the sponsorship of the IUPAC Commission on Food Chemistry     

EC-collaborative study on the determination of aflatoxin B1 in animal feeding stuffs

A collaborative study was conducted to test a method, proposed in the European Community (EC) as a candidate-official method for the determination of aflatoxin B1 in compounded feeding stuffs. The study was undertaken on behalf of the European Commission's Community Bureau of Reference (BCR). It involved 25 laboratories from 11 EC countries. The method, based on chloroform extraction and Sep-Pak Florisil and C18 cartridge clean-up, offered either reverse-phase high performance liquid chromatography (HPLC) with iodine post-column derivatization, or two-dimensional thin-layer chromatography (TLC) as determinative steps. In the study, 22 laboratories applied HPLC, three laboratories applied TLC. The study involved six unknown samples. These consisted of blind duplicate samples of compounded feeding stuff, with target concentrations of aflatoxin B1 at less than 2, 8 and and 14 micrograms/kg. Statistical analysis of the HPLC data was carried out according to ISO 5725. For the less than 2 micrograms/kg sample, all reported aflatoxin concentrations were less than 2 micrograms/kg. At the 8 and 14 micrograms/kg level (pooled), repeatability (r) and reproducibility (R), expressed as ratios, were 1.4 and 1.7 respectively, and within-laboratory and between-laboratory coefficients of variation were 11% and 18% respectively. The study revealed that admittance of daylight in the laboratory caused losses of aflatoxin B1 and must be avoided. New glassware coming into contact with aqueous solutions containing aflatoxin B1 was found to be a potential cause of loss of aflatoxin B1. The method has been recommended to the European Commission to be considered for adoption in EC regulations.     

DNA-based qualitative and quantitative identification of bovine whey powder in goat dairy products

As one of the main ingredients in some milk powders, whey powder is sometimes added to pure goat milk products, which can cause health risks, economic fraud, and unfair competition of food industries. This study is the first to explore qualitative and quantitative methods to identify adulteration of bovine whey powder in goat dairy products based on DNA. We extracted DNA from whey powder using a modified DNA extraction method; this exhibited good quality and integrity, with purity of 1.53 to 1.75 and concentration of 122 to 179 ng/μL. Conventional PCR and real-time PCR were compared for qualitative detection of bovine whey powder; real-time PCR demonstrated sensitivity of 0.01 ng/μL, which was higher than the 0.05 ng/μL detected by the conventional PCR method. Furthermore, real-time PCR was conducted for DNA quantitative detection, with good linearity (R² = 0.9858) obtained for bovine whey powder contents from 0.1% to 30%. Relative error decreased with increase of the mixing proportion of whey powder; the coefficient of variation above 0.1% of the mixing ratio was close to or less than 5%; and the relative standard deviation of repeatability results was less than 5%. Considering the economic costs of testing, conventional PCR could be performed first, and samples with obvious intentional adulteration detected can be further accurately quantified by real-time PCR. Overall, this research provides a realistic and effective method for qualitative and quantitative identification of bovine whey powder in goat dairy products, thus laying a good foundation for verification of goat dairy product label claims and industrial control.     

Whipping test for the determination of foaming capacity of protein: a collaborative study

A standard whipping test, which involved preparing a 0.5% protein dispersion, whipping the dispersion in a Kenwood Chef food mixer and measuring the foaming properties (foam expansion, FE; foam volume stability, FVS; foam liquid stability, FLS) was collaboratively evaluated. A total of sixty determinations was made involving six laboratories, each of which analysed five samples in duplicate. The repeatability coefficient of variation ( CV ₀) in the estimation of FE and FVS properties of proteins was 6.25%. The reproducibility coefficient of variation ( CV _X) for FE and FVS was 9.8%, excluding the FE value (i.e. 20%) for spray-dried albumen. The CV ₀ and CV _X values for FLS properties of proteins ranged from 7.4 to 13.4% and 17.2 to 24.6%, respectively. Thus, FE, FVS and FLS properties can be used for intralaboratory comparisons (e.g. to determine batch-to-batch variation in commercial protein products), whereas FE and FVS values are more reliable for inter-laboratory comparisons of the foaming properties of proteins.     

Enzyme-linked immunosorbent (ELISA) determination of aflatoxin B1 in peanut butter: collaborative trial

Fourteen laboratories in the United Kingdom participated in a collaborative trial of a commercially available ELISA test kit for the determination of aflatoxin B1 in peanut butter. Each laboratory carried out six replicate analyses of each of six individual samples. Collaborators received a control, uncontaminated sample, together with samples prepared by blending naturally-contaminated and control material to give target levels of 8 micrograms/kg, 25 micrograms/kg and 75 micrograms/kg. Two of these samples (8 micrograms/kg and 25 micrograms/kg) were supplied as undisclosed duplicates. The repeatabilities of the assay ranged from 6.2 to 16.7 micrograms/kg. The reproducibilities for aflatoxin B1 concentrations in naturally-contaminated samples ranged from 3.6 to 18.7 micrograms/kg using uncontaminated peanut butter as a reference blank. Modifications to the format of the commercial kit were recommended as a result of the collaborative trial.     

Biosensor-based determination of folic acid in fortified food

A newly developed method for the quantification of folic acid in fortified food is presented. An immunoaffinity-based optical biosensor was used to determine folic acid concentration levels in milk powder, infant formula and cereal samples. Accuracy of the method (88–101%) was demonstrated with the analysis of five reference samples. A collaborative precision study, where ten participants at four different laboratories analysed a set of ten samples, resulted in repeatability relative standard deviations of 2–8% and reproducibility relative standard deviations of 4–10%.     

Joint IDF-IUPAC-IAEA(FAO) interlaboratory validation for determining aflatoxin M1 in milk by using immunoaffinity clean-up before thin-layer chromatography.

A collaborative study was conducted under the auspices of the International Dairy Federation (IDF), the International Union of Pure and Applied Chemistry (IUPAC) and the International Atomic Energy Agency (IAEA), a collaborative Food and Agricultural Organization (FAO) body fully to validate a method combining immunoaffinity clean-up to thin-layer chromatography for the determination of aflatoxin M(1) in milk. Work was done in order to afford those laboratories not equipped with high-performance liquid chromatography, mainly from developing countries, with a simplified but fully validated method as an alternative to the European validated immunoaffinity-high performance liquid chromatography method published as an EN ISO Standard 14501, February 1999. The validation study was carried out on samples of aflatoxin M(1)-contaminated milk and milk powder at levels close to the tolerable level of 0.5 microg l(-1) as recommended by the Codex Alimentarius and to the regulatory level of 0.05 microg l(-1) as laid down by the European Commission. Fourteen laboratories representing 11 countries participated in the trial. The relative standard deviations for repeatability and reproducibility based on raw data were in the range 27-48 and 35-54%, respectively. The recovery rate varied from 32 to 120%. The mishandling of two crucial steps of the protocol such as matrix sample reconstitution and extract evaporation could explain the wide variation of the recovery rate. For this reason, data were then corrected for recovery. Consequently, the relative standard deviations for repeatability and reproducibility were recalculated after correction for recovery and were in the range 26-54 and 34-53%, respectively. The method will be published as a standard ISO/DIS 14674--IDF 190.     

Comparison of dichloran 18% glycerol (DG18) agar with general purpose mycological media for enumerating food spoilage yeasts

Dichloran 18% glycerol (DG18) agar was originally developed to enumerate xerophilic foodborne moulds. However, some laboratories are using DG18 agar as a general medium to enumerate foodborne moulds and yeasts. A collaborative study, with the participation of seven laboratories, was undertaken to compare DG18 agar with dichloran rose bengal chloramphenicol (DRBC) agar, tryptone glucose yeast extract chloramphenicol (TGYC) agar, and plate count agar supplemented with chloramphenicol (PCAC) for enumerating 14 species of common food spoilage yeasts. Comparison of the mean values of populations of all yeasts recovered on each medium revealed no significant differences among DRBC agar, PCAC, and TGYC agar, while each of these media supported the development of significantly (P < or = 0.05) higher numbers of colonies than DG18 agar. However, differences were only 0.08 to 0.10 log10 cfu/ml, making the practical significance questionable. The overall coefficient of variation (CV) for within laboratory repeatability was 1.71%, while the CV for reproducibility of counts obtained among laboratories was 6.96%. Compared to DRBC agar, TGYC agar, and PCAC, yeast colonies were smaller on DG18 agar. Growth of Brettanomyces anomalus, Cryptococcus albidus, and Rhodotorula mucilaginosa was particularly retarded or inhibited on DG18 agar. Based on the performance of media in supporting colony development and ease of counting colonies, the use of DG18 agar as a general enumeration medium for foodborne yeasts cannot be recommended.     

Aluminium determination in food matrices IUPAC check sample survey of analytical performance.

Aluminum has been determined by 24 laboratories in the context of a check sample survey. Samples studied were two duplicate diets, one of which was spiked with 15.87 mg Al/kg, and two blind duplicate milk powders. Target values for the duplicate diets were 11.80 and 27.90 mg Al/kg, respectively, and 15.65 mg Al/kg for the milk powders. Participants were requested to make only single determinations per sample. A two-step approach was used to assess the raw data. In the first step, those data were excluded that were outside a +/- 50% range of the duplicate diet spike and the target value for milk powder. Likewise, only one single data set per participant was accepted and results were ruled out stemming from procedures that have a detection limit of greater than 5 mg Al/kg. The remaining data were evaluated both statistically and in the context of the method performance parameters available. Best scores for aluminium were from laboratories applying wet-pressurized digestion in combination with electrothermal atomic absorption spectrometry. Results for laboratories applying dry-ashing for sample decomposition were unreliable. The overall performance for aluminium is very disappointing given the relatively high aluminium levels of the samples studied. Out of 24 laboratories 11 have one or more major problems with their aluminium determination. They should dramatically improve or replace their methodology for this element.     

Immunochromatographic lateral-flow test strip for the rapid detection of added bovine rennet whey in milk and milk powder

An immunochromatographic lateral-flow test dipstick test was developed for the fast detection of bovine rennet whey in liquid milk and milk powder. The test is based on the binding of casein glycomacropeptide (cGMP) by two specific anti-bovine κ-casein monoclonal antibodies and has a visual detection limit of around 15 ng mL^?1 for cGMP and 1% (v/v) for rennet whey in milk using standards and spiked samples. The dipstick performance was evaluated in a collaborative trial using skim milk powder and raw, pasteurized and UHT milk. The suitability of the dipstick was demonstrated in comparison with gel permeation-high performance liquid chromotography and a colorimetric method, by analysing 60 raw milk samples collected from farms in Brazil. The dipstick results correlated well with the HPLC results and were more reliable than those obtained with the colorimetric method. The dipstick test correctly identified all raw milk samples with a rennet whey content above 4%.     

Validation of a multi-residue enzyme-linked immunosorbent assay for qualitative screening of corticosteroids in liver, urine and milk

A rapid and sensitive enzyme-linked immunosorbent assay (ELISA) was applied for the qualitative screening analysis of dexamethasone, betamethasone, flumethasone, and prednisolone in milk and urine, and dexamethasone, flumethasone and prednisolone in liver samples at levels corresponding to the European Union maximum residue limit (MRL), or at required performance levels (RPLs) for substances for which there is no established MRL. Method validation was performed according to Commission Decision 2002/657/EC criteria established for qualitative screening methods. In this regard, the following parameters were determined: detection capability (CCβ), specificity, limit of detection (LOD), limit of quantitation (LOQ), recovery, within-laboratory reproducibility, linearity and ruggedness. LODs were 0.2, 1.2 and 0.6 μg kg(-1) in milk, urine and liver samples, and LOQ values were 0.3, 1.2 and 1.4 μg kg(-1) in milk, urine and liver, respectively. Recoveries from spiked samples ranged from 68% to 131% for dexamethasone, from 57% to 120% for flumethasone, from 60% to 155% for betamethasone, and from 23% to 32% for prednisolone, with a coefficient of variation (CV) between 1.6% and 21.2%. The CCβ value was below the MRL/RPL for all examined matrices. Moderate variations of some critical factors in the sample pre-treatment for liver and milk samples were deliberately introduced for ruggedness evaluation and did not result in any negative effects on corticosteroid detection. The proposed method is suitable for qualitative screening analysis of corticosteroids in the above-mentioned food in conformity with the current European Union performance requirements.     

乳品中维生素K1的测定

在奶粉样品中加入适量的脂酶，酶解后用NaOH皂化，用正己烷萃取，蒸干后用甲醇乙酯混合注溶解；用甲醇作流动相；用液相色谱法测定维生素K1，此法成功地提取了乳粉中的维生素K1，回收率达87．8％，变异系数CV为1．147％。     

Determination of aflatoxin B1 levels in peanut butter using an immunoaffinity column clean-up procedure: inter-laboratory study

Ten United Kingdom laboratories participated in an evaluation of an immunoaffinity column sample preparation procedure used to prepare aflatoxin B1 containing extracts obtained from peanut butters contaminated by aflatoxins. Each laboratory was sent seven randomly numbered samples of roasted peanut butter which included two sets of undisclosed triplicates. These two peanut butters were naturally contaminated with aflatoxin B1 at levels of about 12 and 35 micrograms/kg. The other sample was a nominal blank peanut butter containing approximately 2 micrograms/kg of aflatoxin B1 which was also employed by participants for recovery experiments. Participating laboratories were instructed to follow a protocol regarding the use of the immunoaffinity columns for extract preparation, but were allowed a free choice of instrumental technique for quantification of aflatoxin levels. Mean recovery for spikes was 72%. Coefficients of variation for the results from the 10 participants for the two contaminated roasted peanut butters were, respectively, 45% (on a mean of 13.6 micrograms/kg) and 36% (on a mean of 37.2 micrograms/kg).     

Gas chromatographic method for the analysis of residues of seven penicillins in food of animal origin]

A capillary gas chromatographic method is described for the determination of residues of benzylpenicillin, phenoxymethylpanicillin, methicillin, oxacillin, cloxacillin, dicloxacillin and nafcillin in bovine muscle, liver, kidney, adipose tissue and milk. The samples are extracted with acetonitrile under slightly acidic conditions, the co-extracted water is separated with the addition of sodium chloride and dichloromethane and discarded. Clean-up is performed by liquid/liquid partitioning steps and anion exchange chromatography. The penicillin residues are methylated with diazomethane. After derivatization, only the extracts from liver and kidney needed further clean-up using cartridges with a polar diol sorbent. The gas chromatographic procedure is based on split/splitless injection, programmed temperature vaporization, separation on a methyl silicone fused silica column and nitrogen-specific thermionic detection. Internal standardization is used for quantification. The limits of detection for all penicillins are well below 3 microgram/kg in milk and all tissues. Recoveries of spiked samples at 3 and 10 micrograms/kg are in the range of 65-80% for milk and 50-70% for bovine tissues.     

Why semicarbazide (SEM) is not an appropriate marker for the usage of nitrofurazone on agricultural animals

A comprehensive global database on semicarbazide (SEM) in foodstuffs and food ingredients is presented, with over 4000 data collected in foods such as seafood (crustaceans, fish powders), meat (beef, chicken powders), dairy products (e.g. raw milk, milk powders, whey, sweet buttermilk powder, caseinate, yoghurt, cheese), honey and other ingredients. The results provide evidence that the presence of SEM in certain dairy ingredients (whey, milk protein concentrates) is a by-product of chemical reactions taking place during the manufacturing process. Of the dairy ingredients tested (c. 2000 samples), 5.3% showed traces of SEM > 0.5 µg/kg. The highest incidence of SEM-positive samples in the dairy category were whey (powders, liquid) and milk protein concentrates (35% positive), with up to 13 µg/kg measured in a whey powder. Sweet buttermilk powder and caseinate followed, with 27% and 9.3% positives, respectively. SEM was not detected in raw milk, or in yoghurt or cheese. Of the crustacean products (shrimp and prawn powders) tested, 44% were positive for SEM, the highest value measured at 284 µg/kg. Fish powders revealed an unexpectedly high incidence of positive samples (25%); in this case, fraudulent addition of shellfish shells or carry-over during processing cannot be excluded. Overall, the data provide new insights into the occurrence of SEM (for dairy products and fish powders), substantially strengthening the arguments that SEM in certain food categories is not a conclusive marker of the use of the illegal antibiotic nitrofurazone.     

Indoxylsulfate in milk

Indoxylsulfate in 27 individual milk samples ranged from 25.4 to 111 micrograms/l (average 52.3 micrograms/l); pooled milk samples from 12 farms contained 81.1 micrograms/l (46.4-146 micrograms/l); the variation in indoxylsulfate concentration of dried skimmed milk over a period of one year amounted to 23%. This variability is likely attributable to regional and seasonal, and hence to feeding effects. The indoxylsulfate content of milk seems also to be dependent upon the degree of fermentation during processing of milk; yoghurt contained very low amounts of this component (6.4 micrograms/kg). On the other hand, heat treatment of the milk (HTST, UHT, sterilization) apparently does not affect its indoxylsulfate content. Indoxylsulfate concentrations in milk correlated positively with blood-serum indoxylsulfate content (r = 0.752, n = 20) and with the urea content of milk (r = 0.61, n = 12 pooled milks). Further research is suggested on the use of indoxylsulfate determinations as an aid to determine sweet whey added to dried skimmed milk, also as an analytical tool to differentiate bovine and sheep milks.     

Multicenter validation study of two blockcycler- and one capillary-based real-time PCR methods for the detection of Salmonella in milk powder

A collaborative study including 13 German laboratories was conducted to evaluate the performance of two non-patented real-time PCR methods for the detection of Salmonella in milk powder targeting the ttrC/ttrA- or the invA gene. The enrichment procedure and sample DNA preparation method prior to the real-time PCR was the same for both systems and the identical DNA extraction samples were analysed. The traditional cultural method according to EN ISO 6579:2002 for the detection of Salmonella in food was performed in each laboratory as the reference. The participants received twelve coded milk powder samples each of 25 g for the analysis. Four of them were Salmonella negative (level L0), four artificially contaminated with <3 MPN/g Salmonella Typhimurium (level L1) and four artificially contaminated with 3.6 MPN/g S. Typhimurium (level L2) to the beginning of the experiment. Of the 13 laboratories 12 used various models of real-time PCR blockcyclers conducting both real-time PCR assays and three laboratories the Light Cycler 2.0 system (Roche Bioscience) conducting the ttr-based real-time PCR assay only. The relative accuracy for both real-time PCR assays performed on blockcyclers was for level L0 97.5%. For level L1 the relative accuracy was 94.1% and for level L2 it was 100% for both assays. The relative accuracy on the Light Cycler 2.0 system was 100% for all levels applied to the ttr-real-time PCR.     

Proficiency testing for the evaluation of the ability of European Union-National Reference laboratories to determine aflatoxin M1 in milk at levels corresponding to the new European Union legislation

In 1992, the European Union set up a network of National Reference Laboratories and charged the Community Reference Laboratory with the responsibility to design a proficiency testing scheme for assessing the analytical ability of laboratories involved in the official control of aflatoxin M1 in milk. Since 1996, two exercises of proficiency testing have been performed on samples of milk powder and liquid milk at various levels of aflatoxin M1 contents. The trials were conducted according to ISO Guide 43, in particular for the homogeneity testing of sample batches and for the calculation of laboratory z-scores. The National Reference Laboratories officially designated by their governments participated in this programme. Samples were naturally-contaminated milk obtained by feeding cows with aflatoxin B1-contaminated feed. The levels of aflatoxin M1 in the samples ranged from 0.2 to 0.7 microg/kg in milk powder and from 0.05 to 0.07 microg/l in liquid milk. These levels were chosen as being close to the European Union-regulated limit of 0.05 microg of aflatoxin M1 per litre. The results produced by laboratories were compiled and statistically analysed to detect any outlying results and to calculate the individual z-scores. Except for one laboratory in each exercise, all laboratories exhibited acceptable or questionable z-scores. The interlaboratory relative standard deviation for reproducibility (RSDR) obtained for both 1996 and 1998 exercises were in the range 15.7-30.3%. Compared with other published studies, this indicates a very good precision for the performance of this laboratory network in the analysis of traces of aflatoxin M1 in milk.     

气相色谱-串联质谱法测定婴幼儿配方乳粉中的3-氯丙醇酯和缩水甘油酯及其风险评估

建立检测婴幼儿配方乳粉中3-氯丙醇酯和缩水甘油酯的气相色谱-串联质谱法,测定不同市售婴幼儿配方乳粉中3-氯丙醇酯和缩水甘油酯的含量,掌握婴幼儿配方乳粉中酯类污染情况并进行安全风险评估。采用正己烷提取婴幼儿配方乳粉中的3-氯丙醇酯和缩水甘油酯,经过水解、苯基硼酸衍生、气相色谱-串联质谱法测定,内标法定量。结果表明,3-氯丙醇酯和缩水甘油酯总量在0.040 0～4.00μg/m L、3-氯丙醇酯含量在0.020 0～2.00μg/m L的范围内线性良好,相关系数R²>0.999,检出限均为10.0μg/kg,定量限均为25.0μg/kg。在25.0、100、300μg/kg添加水平下,平均回收率在95.0%～98.1%之间。该方法准确率高、回收率好,可用于婴幼儿配方乳粉中的3-氯丙醇酯和缩水甘油酯的检测。150份市售婴幼儿配方乳粉样品中,3-氯丙醇酯检出率为12.7%,含量为ND～52.4μg/kg,平均检出值为29.8μg/kg。缩水甘油酯检出率为6.67%,含量为ND～40.1μg/kg,平均检出值为31.9μg/kg。3-氯丙醇酯的平均暴露水平为0.33～...     

International interlaboratory study for the determination of the Fusarium mycotoxins zearalenone and deoxynivalenol in agricultural commodities

Twenty-eight laboratories from 12 different countries participated in an interlaboratory study for the determination of the Fusarium mycotoxin zearalenone (ZON) in maize and deoxynivalenol (DON) in maize and wheat employing their usual in-house methods. The aim of this study was to obtain information about the state-of-the-art of ZON and DON analysis in cereals and to support a knowledge and experience exchange between the participating laboratories in the field of mycotoxin analysis. Eight different sample types were distributed to the participants, 'blank' materials, spiked samples (102 mu g/kg ZON in maize and 475 mu g/kg DON in wheat) and naturally-contaminated maize and wheat. For the final separation and quantification either gas chromatography (GC), high performance liquid chromatography (HPLC), thin layer chromatography (TLC) or enzyme linked immunosorbent assays (ELISA) were employed by the participating laboratories. Coefficients of variation (CV) between laboratory mean results (outliers rejected) ranged from 28 to 41% for ZON and from 32 to 38% for DON. The results are close to the between laboratory CV criteria of 40% for DON and ZON at concentration levels of > 100 mu g/kg established by the CEN in 1999. A good trueness was obtained for the wheat samples spiked at 475 mu g/kg DON. However, a significant deviation at p = 0.01 from the respective target value was observed for the maize samples spiked at 102 mu g/kg ZON. The high CVs can be traced back to problems occurring by determination of the concentration of the participants' own calibrant solutions. Additionally, the variability of the results is strongly influenced by the use of different final separation and quantification procedures.