Methods of analysis for ochratoxin A have greatly improved

Transcription

1 MACDONALD, ET. AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 86, NO. 6, FOOD CHEMICAL CONTAMINANTS Determination of Ochratoxin A in Currants, Raisins, Sultanas, Mixed Dried Fruit, and Dried Figs by Immunoaffinity Column Cleanup with Liquid Chromatography: Interlaboratory Study SUSAN J. MACDONALD, 1 SHARRON ANDERSON, and PAUL BRERETON Central Science Laboratory, Sand Hutton, York, YO41 1LZ, UK ROGER WOOD Food Standards Agency, Aviation House, 125 Kingsway, London, WC2B 6NH, UK Collaborators: G. Barrett; C. Brodie; P.A. Burdaspa; D. Conley; J. Cooper; J. Darroch; C. Donnelly; N. Embrey; R.A. Ennion; I. Felguerias; J. Griffin; M. Kitching; S. Knight; J. Lanham; T.M. Legarda; P. Lenartowicz; E. Luis; J.C. Lundie; T. Möller; D. Norwood; R. Novo; M. Nyberg; C. O Donnell; g. Panzarini; M. Pascale; S. Patel; W. Paulsch; N. Payne; P. Rawcliffe; K. Reid; A. Rizzo; A. Rothin; L. Saari; S.G. Stangroom; W. Swanson; P. Sweet; T. Thomas; R. Trani; E. Turpin; H.P. von Egmond; M. Walker; J.D. Watkins; C. Williams An interlaboratory study was performed on behalf of the UK Food Standards Agency to evaluate the effectiveness of an affinity column cleanup liquid chromatographic (LC) method for the determination of ochratoxin A in a variety of dried fruit at European regulatory limits. To ensure homogeneity before analysis, laboratory samples are normally slurried with water in the ratio of 5 parts fruit to 4 parts water, and test materials in this form were used in the study. The test portion was extracted with acidified methanol. The extract was filtered, diluted with phosphate-buffered saline, and applied to an affinity column. The column was washed and ochratoxin A was eluted with methanol. Ochratoxin A was quantified by reversed-phase LC. The use of post-column ph shift to enhance the fluorescence of ochratoxin A by the addition of 1.1M ammonia solution to the column eluant is optional. Determination was by fluorescence. Currants, sultanas, raisins, figs, and mixed fruit (comprising dried pineapple, papaya, sultanas, prunes, dates, and banana chips) both naturally contaminated and blank (very low level) were sent to 24 collaborators in 7 European countries. Participants were asked to spike test portions of all test samples at a level equivalent to 5 ng/g ochratoxin A. Average recoveries ranged from 69 to 74%. Based on results for 5 naturally contaminated test samples (blind duplicates) the relative standard deviation for repeatability (RSD r ) ranged from 4.9 to 8.7%, and the relative standard deviation for reproducibility (RSD R ) ranged from 14 to 28%. The method showed acceptable within- and Received March 26, Accepted by AP on August 23, Corresponding author s s.macdonald@csl.gov.uk between-laboratory precision for all 5 matrixes, as evidenced by HORRAT values <1.3. Methods of analysis for ochratoxin A have greatly improved in recent years (in particular giving improved limits of detection), and with better performance characteristics, as demonstrated by recent interlaboratory studies (1). One of the main contributing factors to this improvement in methodology has been the introduction of monoclonal antibodies used in affinity columns (2). The use of affinity columns for cleanup, following extraction of matrixes and prior to LC, provides a robust methodology which can be automated and offers good specificity. By removing a large number of coextractives, much improved limits of detection can be readily achieved. There have been several interlaboratory studies of methods for the determination of ochratoxin A in a variety of food matrixes (1). An affinity column method for barley flour at ochratoxin A levels >1.3 ng/g (3), and an affinity column combined with phenyl silane column for roasted coffee at levels of ochratoxin A >1.2 ng/g (4) were validated and accepted as AOAC INTERNATIONAL First Action methods. Subsequently a further affinity column method for ochratoxin A in baby food (milk powder-based infant formula) was validated (5), in this case at the very low limit of >0.1 ng/g, although not put forward as a First Action method. It has been recognized for a number of years that in addition to well established sources of occurrence such as cereals, meat, and coffee, ochratoxin A can also occur in wine (6, 7) and grape juice (6) at low ng/g and sub-ng/g levels, presumably through fungal contamination associated with the grapes probably during harvest prior to pressing the juice. It was therefore no surprise when ochratoxin A was reported to be present in dried vine fruit such as currants, raisins, and sultanas (8) originating from a variety of European countries at levels ranging from 0.5 to 53.6 ng/g. Ochratoxin A has also

2 2 MACDONALD, ET. AL. : JOURNAL OF AOAC INTERNATIONAL VOL. 86, NO. 6, 2003 been found in dried figs (9) and presumably can occur in any dried fruits which may be at risk of fungal growth and mycotoxin production during drying, particularly if sun-drying is used. Several countries have regulatory or guideline limits for ochratoxin A in foods. Levels range from 2 to 50 ng/g ochratoxin A in a variety of commodities, although mostly these limits cover cereals (10). In 2002, the European Commission amended Regulation No. 466/201 (11), setting maximum levels for ochratoxin A in certain foodstuffs. This Regulation sets a limit of 5 ng/g for raw cereal grains, 3 ng/g for all products derived from cereals, and 10 ng/g for dried vine fruits (currants, raisins, and sultanas). This Regulation has to be viewed alongside Commission Directive 98/53/EC (12), which laid down methods of sampling and performance criteria that must be met for methods of analysis. To support the Regulation, a validated method suitable for use in the determination of ochratoxin A in dried vine fruit covering the range of interest is required. Although at present the Regulation is only applicable to dried vine fruit such as raisins, currants, and sultanas it was decided that the method should also be applicable to dried figs and mixed fruit in the event that the scope of the Regulations should be extended at a future date. Interlaboratory Study Test Materials Currants, sultanas, raisins, dried figs, and mixed fruit (comprising dried pineapple, papaya, sultanas, prunes, dates, and banana chips) both naturally contaminated and blank (very low level) were procured from commercial sources. In all cases the naturally contaminated material was initially minced before being blended 5 parts dried fruit with 4 parts water (w/w). After thorough homogenization approximately 110 g of the slurried currant material and 50 g of the other slurried fruit materials were packed into polypropylene screw cap bottles. The samples were labeled with 2 series of numbers (not sequential) for each naturally contaminated material to provide blind duplicates, and labeled as blank currants, raisins, sultanas, mixed fruit, and dried figs to indicate the sample intended for spiking. The test materials were stored at -20 C until dispatched. For distribution the frozen samples were packed into polystyrene boxes containing an ice pack. The samples were sent to participants by carrier to arrive within 24 h of dispatch. Participants acknowledged receipt of the samples upon their arrival. Homogeneity Testing of Packaged Material Between 120 and 200 bottles of each dried fruit material containing naturally occurring levels of ochratoxin A and approximately 150 bottles of the blank materials were prepared. All samples were kept frozen at -20 o C prior to homogeneity testing. Ten samples randomly selected from each batch of packaged material were removed for homogeneity testing. The contents of each bottle (50 bottles in total) was analyzed in duplicate at the authors laboratory using the proposed collaborative trial method, but taking only half the sample size. Data produced from the homogeneity testing were evaluated by means of analysis of variance (ANOVA) and for any trend (t 0.05 test) of contamination levels due to the filling sequence or drift of the applied method (13). From the results of the statistical evaluation, all the test materials were demonstrated to be homogenous, as in each case the calculated F-value was less than the F-critical value. Thus, the results of the ANOVA without any outlier exclusion showed that the difference of the Between-Group-Variance and the Within-Group-Variance regarding all materials was not significant. Therefore, the Between-Sample-Standard Deviation was negligible. A possible trend or drift of the results of the analysis was investigated by using the t-test at the 95% confidence level for the slopes. No drift due to the filling sequence or the analysis order was found for any test material. It was concluded that the 5 analyzed test materials could be considered homogenous at a minimum sample intake of 22.5 g. Organization of Interlaboratory Study Twenty-four collaborators from 7 different European countries representing a cross-section of government, food control, and food industry affiliations took part in the interlaboratory study. The study was performed in 2 stages, initially only with currant material being distributed. After these results were processed, a second stage was performed by distributing the sultanas, raisins, dried mixed fruits, and dried figs. For the first stage, each collaborator received 1 test material of currants which they were requested to analyze in duplicate by the prescribed method. Participants were also sent two blank currant test materials to be used for spiking and blank correction purposes. Each participant was also supplied with a set of instructions, a copy of the method to be followed, and a results proforma. For the second stage of the study, which was carried out approximately 11 months after the first, the same set of participants received (a) 8 coded samples of dried fruit (blind duplicates at four content levels) plus 8 labeled Blank samples for spiking and blank correction purposes; (b) a copy of the method; (c) a set of detailed instructions; (d) a report form for analytical data, criticisms and suggestions; (e) a collaborative study materials receipt form. Each participant was required to prepare 1 extract from each material, perform the cleanup, and analyze the extracts by liquid chromatography (LC). Additionally, each participant was required to spike each of the indicated blank materials for each of the matrixes using the spiking solution (2.5 g/ml ochratoxin A) prepared from as described in Reagents (m). Participants were asked to weigh g blank fruit slurry (as supplied) into a beaker or blender jar then pipet 50 L 2.5 g/ml ochratoxin A spiking solution onto the blank fruit. After adding the spike solution, participants were instructed to let the samples stand for at least 30 min to allow the solvent to evaporate before extraction. The recoveries for each matrix spiked at 5 ng/g were to be reported. Participants were asked to perform the analyses as one batch, but cleanup and analysis could be performed on different days.

3 MACDONALD, ET. AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 86, NO. 6, Statistical Analysis of Results The results of the 2 stages of the study were examined for evidence of individual systematic error (p 0.025) using Cochran s and Grubbs tests progressively, by procedures described in the internationally agreed Protocol for the Design, Conduct and Interpretation of Method-Performance Studies (14). Calculations for repeatability (r) and reproducibility (R) as defined by that protocol (14) were performed on those results remaining after removal of outliers. When assessing a new method there is often no validated reference or statutory method with which to compare precision criteria; hence it is useful to compare the precision data obtained from a collaborative trial with predicted levels of precision. These predicted levels are calculated from the Horwitz equation. Comparison of the trial results and the predicted levels give an indication as to whether the method is sufficiently precise for the level of analyte being measured (15). Historically, the Horwitz predicted value has been calculated from the Horwitz equation (15). (1-0.5 logc) RSD R =2 where C = measured concentration of analyte expressed as a decimal (e.g., 1 g/100g = 0.01). Thompson has recently described the use of a modified Horwitz function to predict levels of precision at ng/g and sub-ng/g levels (16). The use of this function is shown to give an improved statistical representation at these levels. Therefore for the purposes of this trial the Horwitz predicted value was calculated from the modified Horwitz function RSD R = R = 0.22c. The HORRAT value (17) gives a comparison of the actual precision measured with the precision predicted by the Horwitz equation for a method measuring at that particular level of analyte. It is calculated as follows: Ho R = RSD R (measured)/rsd R (Horwitz) In the case of this trial Ho R = RSD R (measured)/ R A HORRAT value (Ho R ) of 1 usually indicates satisfactory interlaboratory precision, whereas a value of >2 usually indicates unsatisfactory precision, i.e., one that is too variable for most analytical purposes or where the variation obtained is greater than that expected for the type of method used. Ho r is also calculated and used to assess intralaboratory precision, using the following approximation: RSD r (Horwitz) = 0.66 RSD R (Horwitz) (this assumes the approximation r = 0.66 R). For this trial RSD r (Horwitz) = 0.66 R Method (Applicable to determination of ochratoxin A at 5 ng/g in currants and raisins, 10 ng/g in sultanas, 1 ng/g in dried mixed fruit, and 2 ng/g in dried figs). See Table 1 for the results of the interlaboratory study for ochratoxin A in currants, raisins, sultanas, mixed dried fruit and dried figs. Tab 1 Caution: Ochratoxin A is a potent nephrotoxin with immunotoxic, teratogenic, and potential genotoxic properties. The International Agency for Research on Cancer (IARC) has classified ochratoxin A as a possible human carcinogen (group 2B). Protective clothing, gloves, and safety glasses should be worn at all times, and all standard and sample preparation stages should be performed in a fume cupboard. Toluene is highly flammable and harmful. Standard preparation involving this solvent must be performed in a fume cupboard. Operations outside the fume cupboard, such as measurement of standards by UV spectrophotometry, must be performed with the standards in closed containers. Principle Dried fruit is initially slurried with water to form a homogenous paste for subsampling. Test portion is extracted with acidified methanol. The extract is filtered, diluted with phosphate-buffered saline, and applied to an immunoaffinity column containing antibodies specific to ochratoxin A. Ochratoxin A is removed from the affinity column with methanol. Ochratoxin A is quantified by reversed-phase LC with optional post-column ph shift by ammonia solution. Determination is by fluorescence detection. Performance Standard for Affinity Column The immunoaffinity column should contain antibodies raised against ochratoxin A. The column should have a maximum capacity of not less than 100 ng of ochratoxin A. The performance of the column should be checked by applying a solution of 100 ng ochratoxin A in a solvent mixture of the same composition as the sample extract to be applied. This should give a recovery of not less than 70%. Apparatus (a) Silanized glass vials (optional). To prepare vials, fill them with silanizing reagent (e.g., Surfasil or equivalent, 5% - Fluka, Gillingham, UK). Mix silanizing fluid + toluene (1 19) and leave this in the vial for 1 min. Rinse vial first with solvent of low polarity, for example toluene, and then with methanol. Finally, wash vials twice with distilled water and dry before use. The use of silanized glassware may prevent ochratoxin A binding to glass during evaporation. (b) Blender or homogenizer. Explosion proof. (c) Displacement pipets. 5 ml, 1 ml, and 200 L capacity with appropriate pipet tips. (d) Vacuum manifold. To accommodate immunoaffinity columns.

4 4 MACDONALD, ET. AL. : JOURNAL OF AOAC INTERNATIONAL VOL. 86, NO. 6, 2003 (e) Reservoirs and attachments. To fit to immunoaffinity columns. (f) Vacuum pump. Capable of pulling a vacuum of 10 mbar and pumping 18 L/min.(g) Filter paper. Pore size 20/25 m (Whatman No. 4 or similar). (h) LC apparatus. Variable injection system, valve injection system with, e.g., 100 L injection loop. Pump, isocratic, pulse free, capable of maintaining volume flow rate of 1 ml/min. Column oven, (optional) capable of maintaining constant temperature ( 0.5 C). (i) LC column. C 18 reversed-phase ODS, which ensures resolution of ochratoxin A from all other peaks. The maximum overlapping of peaks must be 10% (it might be necessary to adjust the mobile phase for a sufficient baseline resolution). A pre-column should be used. (j) Fluorescence detector. Fitted with flow cell and set at 333 nm (excitation wavelength) and 477 nm (emission wavelength), or set at 390 nm (excitation wavelength) and 440 nm (emission wavelength), if optional post column system is used. (k) Post column system. (Optional), comprising pump, isocratic, pulse free, capable of maintaining a volume flow rate of 0.3 ml/min, zero dead volume t-piece, and reaction coil 1.5 m id tubing (stainless steel or PEEK). (l) UV spectrophotometer. (m) Analytical balance. Accurate to 10 mg. Reagents All reagents are analytical grade unless otherwise stated. (a) Acetonitrile. (b) Glacial acetic acid. (c) Methanol. (d) Ortho-phosphoric acid, 0.1M. (e) Injection solvent. Mix water with acetonitrile (4.4) and acetic acid ( , v/v). (f) Phosphate-buffered saline (PBS). Dissolve 8 g sodium chloride, 1.2 g anhydrous disodium hydrogen orthophosphate, 0.2 g anhydrous potassium dihydrogen phosphate, and 0.2 g potassium chloride in 900 ml distilled water. Adjust ph to 7.4 with sodium hydroxide (or hydrochloric acid), and then dilute to 1 L with water. (Note: Commercially available PBS tablets with equivalent properties may be used). (g) Helium purified compressed gas. (h) Mobile phase. Mix water acetonitrile acetic acid ( ) and de-gas with, for example, helium. (i) Ammonia solution for post column ph shift. To be made fresh when required. Solution of 1.1M ammonia in distilled water. (j) Toluene. (k) Analytical reagent mixture of toluene and glacial acetic acid. Mix toluene with acetic acid (99 1, v/v). (l) Surface silanizing fluid, 5%(v + v) optional. Mix surface silanizing fluid with toluene (1 + 19, v/v). (m) Ochratoxin A spiking solution. Transfer volume of stock solution containing 12.5 g ochratoxin A to 5 ml volumetric flask. Evaporate to dryness under nitrogen at no more than 50 o C. Re-dissolve immediately in methanol and dilute to volume. This solution contains 2.5 g/ml ochratoxin A. Store this solution in freezer or refrigerator when not in use. Allow to reach room temperature before opening. This solution is stable for at least 4 weeks. (n) Ochratoxin A working standard solution. Transfer 500 L ochratoxin A spiking solution to 5 ml volumetric flask, dilute to volume with methanol. This solution contains 0.25 g/ml ochratoxin A. Store this solution in freezer or refrigerator when not in use. Allow to reach room temperature before opening. This solution is stable for at least 4 weeks. (o) Immunoaffinity columns. See Performance Standard for Affinity Column. For example, columns from Vicam (Watertown, MA) and R-Biopharm Rhone (Glasgow, UK) meet these criteria. Extraction Weigh g fruit slurry (as supplied) into beaker. Add 50 ml methanol and 5 ml 0.1M ortho-phosphoric acid. Blend to a smooth suspension using the homogeniser for 3 to 4 min. Filter the slurry through filter paper by gravity. Affinity Column Cleanup The cleanup may be performed by using a vacuum, by positive pressure, or by allowing the specified volumes to pass through the column under gravity. Do not exceed maximum specified flow rates. Prepare affinity column according to manufacturer s instructions. Accurately measure 12 ml filtrate and dilute to 100 ml with PBS. Transfer diluted extract to a flask and shake well to mix. Add 50 ml diluted sample extract to reservoir and pass through affinity column. Flow rate should not exceed 5 ml/min. The affinity column shall not be allowed to run dry. Wash the affinity column with 10 ml water (or PBS depending on column manufacturer s instructions). Place a vial under the affinity column. Elute the ochratoxin A into a vial with suitable solvent as recommended in the affinity column manufacturer s instructions. Evaporate affinity column eluate to dryness under nitrogen. Re-dissolve in 1.0 ml injection solvent. Transfer to LC vial (V 3 ). Alternatively dilute column eluate with LC water (or water with 2% acetic acid depending on elution solvent) to prepare test solution for analysis. To 1 ml eluate (accurately measured) add 2 ml water. Mix well. Note: The cleanup, preparation, and LC steps of this method may be performed by an automated system such as an ASPEC, provided that the conditions described in this method, e.g., volumes and flow rates are adhered to. LC Determination with Fluorescence Detection (a) LC operating conditions. When column specified in apparatus (i), ( 4.6 x 250 mm with 5 m particle size) and mobile phase specified in reagents (h) were used, the following settings were appropriate: flow rate mobile phase (column), 1.0 ml/min; fluorescence detection, emission wavelength, 477 nm; fluorescence detection, excitation wavelengthm 333 nm; injection volume, L; sample extracts diluted with water, L.

5 MACDONALD, ET. AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 86, NO. 6, (b) Post column reaction conditions (optional). Using post-column system specified above, and column specified in apparatus (i), the following LC conditions have been enhanced the response of ochratoxin A by a factor of 3 to 4: flow rate mobile phase (column), 1.0 ml/min; flow rate post column reaction solution, 0.3 ml/min; fluorescence detection, emission wavelength, 440 nm; fluorescence detection, excitation wavelength, 390 nm; post column reaction loop, 1.5 m id; injection volume, L; sample extracts diluted with water, L. (c) Calibration curve. Prepare 4 LC standard solutions in separate 5 ml volumetric flasks according to Table 2. Dilute each standard to volume (5 ml) with injection solvent. Prepare calibration graph at beginning of every day of analysis. Plot mass of ochratoxin A in the aliquot injected against peak area (or height) response. Tab 2 Calculations Determine from the calibration graph, the masses in ng of ochratoxin A in the aliquot of test solution injected onto the LC column. Calculate the mass fraction of ochratoxin A, wota, in g/kg using the equation: W m V OTA a V V V ms where m a is the mass of ochratoxin A in the aliquot of test solution injected onto the column, in nanogram; V4 is the volume of the aliquot of test solution injected onto the column, in ml; V3 is the volume of the test solution, in ml (V3 = 1,0 ml); V2 is the volume of sample filtrate used in cleanup, in ml (V2 = 6 ml as 50mL filtrate after cleanup); V1 is the volume of the extraction solvent, in ml (V1 =75 ml); ms is the mass of the sample extracted, in g (25 g); Express the final result in g/kg as this is equivalent to ng/g. Results and Discussion Collaborative Trial Results Of the 24 laboratories that received the test samples, all successfully completed the study. All data submitted for the study are presented in Table 3. The data are given as individual pairs of results for each laboratory (identified as 1 24). Blank samples for each matrix were spiked with 5 ng/g ochratoxin A, but because the spiking was performed by participants (rather than by the study organizer) and spikes were not analyzed as blind duplicates, no precision data could be generated from these samples. The coordinating laboratory calculated recovery values for each participating laboratory, and the recovery values (as means) are reported in Table Tab 1. 3 Statistical Analysis of Results Precision estimates were obtained using a one-way analysis of variance approach according to the IUPAC Harmonized Protocol (13). Details of the dried fruit matrix, average analyte concentration, standard deviations for repeatability (RSD r ) and reproducibility (RSD R ), number of statistical outlier and noncompliant laboratories, the HORRAT ratio, and percentage recovery are presented in Table 1. The collaborative trial results were examined for evidence of individual systematic error (p < 0.025) using Cochran s, and Grubbs tests progressively (13). Pairs of results that were identified as outliers are indicated in bold and are individually identified by code in Table 3. Noncompliant results were identified as those where no statistics were possible, such as single results instead of pairs of results, or less than values instead of numerical results being reported. For the results given in Table 3, the maximum number of outliers identified was 4 laboratories, giving acceptable data ranging from 20 to 24 laboratories. Comments from Collaborative Trial Participants The study was carried out in 2 stages. Initially only currants (naturally contaminated and spiked blanks) were analyzed, and comments and results were received. Subsequently, the second part of the study involved the 4 further dried fruits (sultanas, raisins, mixed dried fruits, and dried figs) and again comments were invited. Some minor modifications were made to the method after phase 1 in response to the comments received from the participants. In particular, a detailed description of ochratoxin A standard preparation was included. In general, the participants found the method clear and easy to perform, and where comments were received there was no evident pattern in the observations. Information supplied on the results proforma showed that participants used 2 different types of commercial affinity columns and a range of LC columns. However, most participants used the same chromatography conditions, although some modified conditions slightly to achieve the desired performance. All participants were asked to include their chromatograms with their trial results. All chromatograms provided were assessed as satisfactory by the coordinating laboratory. Precision Characteristics of the Method Although the study was carried out in 2 parts, the data were collated as single set in Table 3. Four participating laboratories were found to be noncompliant by the coordinating laboratory in stage 1. Failure to return results, unsatisfactory calibration, and a procedural error constituted noncompliant data. These results, if reported, were excluded from the calculation of performance criteria. All these laboratories continued to participate in stage 2, for which there were no further noncompliant results. The precision data for all dried fruit samples is summarized in Table 1. Based on results for naturally contaminated samples (blind pairs) the RSD r ranged from 4.9 to 8.7%. RSD R ranged from 14 to 28%, and average recovery of ochratoxin A derived from the spiked dried fruit samples ranged from 69 to 74%.

6 6 MACDONALD, ET. AL. : JOURNAL OF AOAC INTERNATIONAL VOL. 86, NO. 6, 2003 Interpretation of Results The acceptability of the precision characteristics of the method were assessed on the basis of HORRAT values (15) which compare the RSD R at the various levels with those values predicted from collaborative trial studies taken from the published literature. When outliers were excluded, the HORRAT values for ochratoxin A were 1.3 for currants and ranged from 0.6 to 0.8 for the other dried fruits. These precision values for sultanas, raisins, dried mixed fruits, and dried figs were well within the statistically predicted range (<1). A HORRAT value (H OR ) of 1.3 for currants was within the satisfactory range (<2). Conclusions This work reports the first interlaboratory study of a method for ochratoxin A in dried fruit. The method involving affinity column cleanup and determination by reversed-phase LC was successfully validated at 1 11 ng/g for a variety of dried fruit matrixes. The method is suitable for enforcement purposes to test compliance with European Directives (11) and was shown to have performance characteristics which fulfill European requirements (18) and justify putting the method forward for consideration as a CEN Standard. This study forms part of the Food standards Agency Collaborative Trial Programme. In addition to producing validated methods which can be used in the UK and by the European Commission for enforcement purposes, the Programme also addresses wider measurement issues. Further data from the present trial (not included here), examining the issue of recovery, will be published elsewhere. Acknowledgments This study was funded by the UK Food Standards Agency. We thank the following analysts who participated in the collaborative trial. G. Barrett and M. Walker, ADAS, (CITY?) UK P.A Burdaspal and T.M Legarda, Centro Nacional de Alimentation, (CITY?) Spain J. Darroch, C. Brodie and S. Knight, Aberdeen City Council Public Analysts Laboratory, UK C. Donnelly, C. O Donnell and E. Luis, R-Biopharm Rhône, Glasgow, UK N. Embrey, Staffordshire County Laboratory and Scientific Services, UK R.A. Ennion and J.D. Watkins, Ruddock and Sherratt, (CITY?) UK I. Felguerias and R. Novo, INETI/DTIA, Lisbon, Portugal J. Griffin and J. Cooper, Kent Scientific Services, UK J. Lanham, Weston Research Laboratories, Maidenhead, Birkshire, UK M. Pascale and G. Panzarini, CNR, Pisa, Italy P. Lenartowicz and T. Thomas, Analytical Services, (CITY?) UK J.C. Lundie, Nestlé Regional Laboratory, (CITY?) UK T. Möller and M. Nyberg, National Food Administration, Uppsala, Sweden D. Norwood and M. Kitching, Sundora Foods, Pocklington, York, UK S. Patel, RHM Technology Ltd, Buckinghamshire, UK N. Payne, K. Reid and D. Conley, Pattinson Scientific Services, Newcastle upon Tyne, UK A. Rizzo and L. Saari, EELA, Helsinki, Finland A. Rothin, Worcestershire Scientific Services, UK S.G. Stangroom and R. Trani, Lincolne Sutton and Wood Ltd, Norwich, UK W. Swanson, Glasgow Scientific Services, UK P. Sweet, Cardiff Scientific Services, UK E. Turpin and P. Rawcliffe, Lancashire County Council Laboratory, Preston, UK H.P. van Egmond and W. Paulsch, RIVM, (CITY?) The Netherlands C. Williams, LGC Ltd, (CITY?) UK (PLEASE VERIFY AND/OR CORRECT CITY AND COUNTRY FOR EACH PARTICIPANT LISTED.) References (1) Gilbert, J., & Anklam, E., (2002) Trends Anal. Chem. 21, (2) Scott, P.M., & Trucksess, M.W. (1997) J. AOAC Int. 80, (3) Entwisle, A.C., Williams, A.C., Mann, P.J., Slack, P.J., & Gilbert, J. (2000) J. AOAC Int. 83, (4) Entwisle, A.C., Williams, A.C., Mann, P.J., Russell, J., Slack, P.T., & Gilbert, J. (2001) J. AOAC Int. 84, (5) Burdaspal, P., Legarda, T.M., & Gilbert, J. (2001) J. AOAC Int. 84, (6) Zimmerli, B., & Dick, R. (1996) Food Addit. Contam. 13, (7) Visconti, A., Solfrizzo, A., & De Girolamo (2001) J. AOAC Int. 84, (8) MacDonald, S., Wilson, P., Barnes, K., Damant, A., Massey, R., Mortby, E., & Shepherd, M.J. (1999) Food Addit. Contam. 16, (9) MAFF Joint Food Standards and Safety Group (1999) 1998 Survey of Retail Products for Ochratoxin A, Food Surveillance Information Sheet Number 185, London, UK (10) FAO (1997) Food and Nutrition Paper 64: Worldwide regulations for mycotoxins 1995 A compendium (FAO, Rome, 1997) pp 43 (PLEASE GIVE ENDING PG NO.) (11) Commission Regulation (EC) No. 472/2002 of (12 March 2002) Amending Regulation (EC) No. 466/2001 Setting Maximum Levels for Certain Contaminants in Foodstuffs, Official Journal of the European Communities L 75/18, Brussels, Belgium (12) Commission Directive 98/53/EC of 16. July 1998, Laying Down the Sampling Methods and the Methods of Analysis for the Official Control of the Levels for Certain Contaminants in Foodstuffs, Official Journal of the European Communities L 201/93, Brussels, Belgium (13) Thompson, M., & Wood, R. (1993) Pure Appl. Chem. 65, (14) IUPAC (1995) Pure Appl. Chem. 67,

7 MACDONALD, ET. AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 86, NO. 6, (15) Horwitz, W. (1982) Anal. Chem. 54, 67A 76A (16) Thompson, M. (2000) Analyst 125, (17) Peeler, J.T., Horwitz, W., & Albert, R. (1989) J. Assoc. Off. Anal. Chem. 72, Table 1. Statistical analysis of interlaboratory trial results for ochratoxin A in currants, Avg.,,ng/g NC a No. sets of results r S r ng/g RSD r,,% R S R,,ng/g RSD R,,% No. outlier labs Horrat value Currants Sultanas (2 22) Raisins