MACDONALD ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 83, NO.6, 2000 1387 FOOD CHEMICAL CONTAMINANTS Liquid Chromatographic Method for Determination of Patulin in Clear and Cloudy Apple Juices and Apple Puree: Collaborative Study SUSAN MACDONALD, MELINDA LONG, and JOHN GILBERT Ministry of Agriculture, Fisheries and Food, Central Science Laboratory, Sand Hutton, York YO41 1LZ, UK ILIDIA FELGUEIRAS Instituto Nacional de Engenharia e Tecnologia Industrial (INETI), Estrada do Paco do Lumiar, 1699 Lisboa Codex, Portugal Collaborators: C. Brera; K. Jørgensen; M.L. Macho; P. Majerus; M.L. Martins; L. Mevissen; J.-Y. Michelet; K. Nuotio; A. Pittet; L. Szymanski; N. Tucker; R. Viladrich; J. Voogt; A. Wennemar A collaborative trial was conducted to validate the effectiveness of a liquid chromatographic (LC) procedure for determination of patulin in both clear and cloudy apple juices and apple puree. The test portion of clear apple juice was directly extracted with ethyl acetate; cloudy apple juice and apple puree were treated with pectinase enzyme before extraction. After back-extraction into sodium carbonate to remove interfering acidic compounds, the extract was dried and concentrated, and patulin was determined by LC with UV detection. Clear and cloudy apple juices, apple puree test samples naturally contaminated with patulin, and blank test samples for spiking with patulin were sent to 14 collaborators in 12 different European countries. Test portions of each of the 3 test sample types were spiked with patulin at 75 ng/g. Recoveries of patulin ranged from 80 to 92%. Based on the results for spiked test samples (blind pairs) and naturally contaminated test samples (blind pairs at 3 levels), the relative standard deviations for repeatability (RSDr) and reproducibility (RSDR) ranged from 8 to 35% and 11 to 36%, respectively. Although HORRAT values of <1.4 were obtained for all 3 matrixes at patulin levels ranging from 26 to 121 ng/g, better performance values (RSDr values 6–10% and RSDR values 11–25%) were obtained for clear and cloudy apple juice spiked above 50 ng/g, which is either the statutory limit or the advisory level for patulin contamination in apple juices in many countries. Submitted for publication July 2000. The recommendation was approved by the Methods Committee on Natural Toxins, and was adopted by the Official Methods Board of AOAC INTERNATIONAL. See “Official Methods Board Actions,” (2000) Inside Laboratory Management, April issue. arious analytical approaches have been used for analysis of patulin in apple juice with essentially similar (fairly simplistic) extraction and cleanup but with different determinative steps. Thin-layer chromatography (TLC; 1), gas chromatography (GC; 2) with derivatization, gas chromatography/mass spectrometry (GC/MS) with derivatization (3), and liquid chromatography (LC; 4–7) have all been successfully used for patulin analysis. For TLC with 3-methyl-2-benzothiazolinone hydrazone⋅HCl as spray reagent, a full collaborative study was undertaken in 1974 (1), resulting in a First Action AOAC Method. For that study at patulin levels of 50, 120, and 340 ng/g, the relative standard deviations for repeatability (RSDr) and reproducibility (RSDR) ranged from 31 to 38%, and 21 to 30%, respectively. An ISO standard for determination of patulin in apple juice, apple juice concentrates, and drinks containing apple juice was also based on a TLC approach (8). More recent studies have used LC rather than TLC or GC. A collaborative study by IUPAC in 1988 (9) compared 2 LC methods, one extracting into ethyl acetate and back-extracting into sodium carbonate, and the other extracting into ethyl acetate with cleanup on a silica gel column before LC. Both approaches were acceptable. The first gave RSDr values of about 7% and RSDR values of about 8%; the second gave RSDr values of 12–18% and RSDR values of 14–17%. Apple juice test samples with patulin contents ranging from 40 to 200 ng/g were used in the 2 trials, and led to an ISO standard for patulin (10) based on the first of the 2 approaches (without silica gel cleanup). The most recent collaborative study was undertaken by Brause et al. (11) and led to the AOAC First Action LC Method (12). That study involved 22 participants from 10 countries who analyzed blind duplicates of apple juice containing patulin at 4 spiked levels (20, 50, 100, and 200 ng/g) and one naturally contaminated apple juice containing an average level of 35 ng/g. The method used by Brause et al. (11) was essentially identical to that used in the present study for V NO. which was obtained from a commercial source and contained patulin at <5 ng/g. the juice was centrifuged at 7200 × g for 10 min. and aliquots were analyzed periodically to assess the exact patulin concentration and homogeneity. (c) An ampule of patulin calibrant solution (10 µg/g patulin in absolute alcohol). (f) A vial marked ‘Enzyme’ containing pectinase enzyme to be used as described in the method for pretreatment of cloudy apple juice and apple puree. Clear Juice Samples The clear juice test samples were prepared in-house from a naturally contaminated cloudy juice.: JOURNAL OF AOAC INTERNATIONAL VOL. naturally contaminated with patulin. cored. The resulting puree contained patulin at >9000 ng/g. Organization of Collaborative Trial Fourteen collaborators from 12 different European countries representing a cross-section of government. cloudy apple juice. The European Standardization Organization CEN has indicated a need for a validated method for patulin as a European standard. A supply of windfall apples. peeled. and details of the collaborative trial were outlined. there are no European Union (EU) limits for patulin. Commercially available blank apple juice (a blend of English apple juices) contained patulin at a concentration of <5 ng/g. was obtained from a local source. b. chopped into 8 segments. and pureed in a food processor. and found to contain patulin at a concentration of >180 ng/g. Bramley. The puree was then microwaved on full power for 6 min. 6.g. The method has been modified and includes pectinase enzyme treatment to improve the clarity of juices and puree before analyte isolation.. (b) Two known blank materials of clear juice. analyzed by the method reported here. a calibrant solution for spiking. and food industry affiliations took part in the collaborative trial. The test materials were produced by mixing together known quantities of contaminated and blank purees. Before the collaborative trial. each laboratory received a set of familiarization test samples. This is done in commercial practice and does not interfere with patulin analysis. Analysis of the resulting clear juice showed that it contained patulin at a concentration of >180 ng/g. but there are no comparable studies for cloudy juices or apple puree. although in the Czech Republic a limit of 30 ng/g is applied to infant foods. food control. . Blank test samples were prepared from fresh apples purchased from a commercial source and made into a puree. cloudy juice. Test materials of the required concentrations were produced by mixing known volumes of naturally contaminated and blank cloudy apple juices together in a 25 L container for 2 h on an automatic shaker. where any problems encountered with the familiarization test samples were discussed. 2000 clear apple juice. This clear juice was diluted with blank clear apple juice. As with the apple juice. As with the cloudy juice. Test Materials for the Collaborative Trial Contaminated Cloudy Apple Juice A blend of juices from Cox. Further volumes of contaminated or blank juice were added as necessary to reach the required concentration of patulin.5 g ascorbic acid was added to every 1 kg of apple immediately after chopping. and apple puree labelled a. (11). 83. and chopped. To prevent discoloration. which may be produced from damaged and molded apples. At present. each participant received: (a) Eight units each of clear juice. the UK) have voluntary guideline limits and there are some moves to harmonize limits in the EU. Further quantities of contaminated or blank puree were added as necessary to reach the required content of patulin. cored. and Crispin apple varieties was obtained from a local apple juice producer. Regulations control patulin contamination in about 11 countries worldwide (13) with limits in most cases set at 50 ng/g for fruit (apple) juices. c. Apple Puree Samples The puree test samples were produced within the laboratory. (d) Two ampules marked ‘Spike solution A’ and ‘Spike solution B’ for spiking experiments. (g) A copy of the method. For the collaborative trial. d. was extended to cloudy apple juice and apple puree. and then pureed in a food processor and pasteurized at 60°C for 30 min. The puree was initially mixed for 3 h in a food processor and then for an additional 1 h in a Silverson blender. The study reported here. After incubation at 40°C for 2 h. A precollaborative trial workshop was also organized.1388 MACDONALD ET AL. (e) An amber vial marked ‘HMF standard’ containing ca 5 mg 5-hydroxymethyl furfural. and a solution of 5-hydroxymethyl furfural (HMF). A 75 mL aliquot of the pectinase enzyme (typical activity 1400 U/g expresses as endogalacturonase) was added to 3 L naturally contaminated cloudy apple juice. however. which have presented problems in the past in terms of recoveries and are more prone to interference. The apples were peeled. and was found to contain patulin at <5 ng/g. although 5 member states of the EU have limits in their national regulations and others (e. The collaborative trial results for clear juices in the present study may be compared directly with those of Brause et al. comprising a blank and naturally contaminated test sample of both cloudy apple juice and apple puree. which would cover not only clear juice but also the increasingly popular organic cloudy juices and apple puree. the puree was analyzed periodically to assess the patulin concentration and homogeneity. a vial of pectinase enzyme. the clear apple juices were mixed together in a 25 L container for 2 h on an automatic shaker. A subsample of the naturally contaminated cloudy apple juice was clarified. The juice was analyzed periodically to assess the patulin concentration and homogeneity. and apple puree for spiking experiments. Further volumes of contaminated or blank juice were added as necessary to reach the required concentration of patulin. 0. —LC pump(s) and eluent reservoir. 83.50 — 12.9 12 0. and 12% carbon loading. b.MACDONALD ET AL. using the provided ‘Spike solutions A and B. each participant was required to spike each of the indicated ‘blank’ materials for each of the 3 matrixes with syringe. C(c). (3) Separation columns. thus completing the experimental work for the trial in 3 days.5 11 9.6 25 1.—Test sample applicator.31 — a(b) c d a = number of labs retained after eliminating outliers.3 23 0.25 — 10 8.4 13 15. and carbon loading of 17%.’ After adding the spike solution.3 6 nc (d) 106 10 (2) 10.85 80 75 ng/g (a) d nc (b) 8 Cloudy Apple Juice 75 ng/g (a) 60 11 (1) 7. (b) Spectrophotometer. criticisms. and d.8 29 1. B. Calibrate as follows: Determine absorbance (A) of the 3 solutions of K2Cr2O7 in H2SO4. 12 nm pore size. and analyze the extracts by LC. at maximum absorption near AOAC Official Method 2000.02 Patulin in Clear and Cloudy Apple Juices and Apple Puree Liquid Chromatographic Method First Action 2000 (Applicable to determination of patulin at >25 ng/g in clear apple juice.95 89 26 12 (0) 3. % HORRAT Rec. Principle Apple juice or puree is extracted with ethyl acetate and then cleaned up by extraction with sodium carbonate solution. NO. This would mean analyzing a batch of 10 samples (8 coded plus 2 spike samples per matrix) on separate days. patulin is quantitatively determined by LC with UV detection.8 10 12.) Table 2000. Additionally. fully end capped with 5 µm particle stationary phase. An analytical reversed-phase LC column such as ODS. % Clear Apple Juice c 67 12 (0) 8. 25 nm pore size. perform the cleanup. % SR RSDR. and apple puree. of labsa(b) Sr RSDr. cloudy apple juice. (2) Injection system.23 — nc (c) 38 9 (2) 3.—(1) Pump. Interlaboratory study results for patulin in clear and cloudy apple juices and apple puree 0.02 — nc (d) 128 10 (2) 9.02.56 92 nc (b) 23 8 (1) 6. and C(d). (i) A report form for analytical data. Each participant was required to prepare one extract from each material. and suggestions.2 10 35 1.: JOURNAL OF AOAC INTERNATIONAL VOL.54 — Apple Puree 75 ng/g (a) 69 9 (1) 7.—350–250 nm. or ODS super end capped with 5 µm particle stationary phase. Laboratories received 4 sets of duplicate test samples labeled a.27 — nc (d) 121 10 (0) 23.9 14 0.5 21 0. 2000 1389 (h) A set of additional instructions for the spiking protocol. See Table 2000. c. After evaporation of the solvent. Participants were advised to analyze the 3 different matrixes on separate days.. (j) A collaborative study materials receipt form. participants were instructed to shake the juices or mix the apple puree with a spatula and let stand for at least 2 h before extraction.9 14 11 0. Apparatus (a) LC apparatus.6 33 1.—A 4. nc = naturally contaminated.3 mm id octadecylsilane (ODS) precolumn with 5 µm particle stationary phase.7 14 8. (b) = number of labs removed as outliers. 6. ng/g ID No.02 for results of the interlaboratory study supporting the acceptance of the method.9 35 nc (c) 69 9 (2) 4.6 19 34. assuming LC analysis was performed overnight between days.18 — nc (c) 54 12 (0) 6 11 13.61 — 12.) The ethyl acetate extract is dried with anhydrous sodium sulfate.8 13 nc (b) 26 12 (0) 8.2 13 0. (Cloudy apple juice and apple purees are pretreated with pectinase enzyme. . C(b).4 33 1. (4) Detector.—UV detector at 276 nm and data integration system. A.4 27 8.5 36 1. (f) Rotary evaporator.—≥99.0625mM.—≥99. measure 20 mL test portion into centrifuge tube B(e) and add 150 µL pectinase enzyme solution C(h). (m) Sodium carbonate solution. (g) Ethyl acetate. Calculate molar absorptivity (ε) at each concentration as follows: ε= A × 1000 C where A = absorbance at maximum near 350 nm.5%.—Add 3–10% acetonitrile (f) to water containing 0. (p) Patulin. (h) Hand-held pipets. ≥99%. NO. Degas this solution before use. then centrifuge at 4500 × g for 5 min. and shake vigorously. (h) Pectinase enzyme solution.—25. PTFE. 1000 µL. and shake vigorously to mix. ε = molecular absorbance coefficient of patulin solution at the wavelength maximum (276 nm) of absorption spectrum (14 600 L mol–1 cm–1 in ethanol). To determine exact mass concentration in calibrant solution.2-c)pyran-2(6H)-one). Wales. endogalacturonase.—4500 × g. check either technique or instrument.2). Transfer standards to vials for LC analysis and use on same day as preparation. (r) Patulin calibrant solution.05 µg/mL patulin. 50.—For clear apple juice.125mM.—Anhydrous. 0. which is equivalent to absolute amount of 5 µg patulin to dryness. If CF is <0. If the 3 values vary by more than guaranteed accuracy of A scale. C. Determine correction factor (CF) for particular instrument and cells by substituting in equation: CF = 3160 εε where 3160 = value for ε of K2Cr2O7 solutions. (b) Potassium dichromate. 0.1390 MACDONALD ET AL. (Use same set of cells in calibration and determination of purity of patulin.—Anhydrous.—Evaporate 1000 µL stock solution (q) to dryness under N. Dilute to mark with pH 4 water (n). UK) is suitable.095 parts per volume perchloric acid 60% (i). (n) pH 4 water. 6. Procedure Preparation of test portion. (v) HMF-patulin solution.5. (o) Elution solution for LC. (j) 5-Hydroxymethyl furfural (HMF).—≥99. respectively.—50 mL with screw cap.009M H2SO4 in volumetric flask.—Optical path length 1 cm.7% (v/v). (i) Perchloric acid. (Main Ave. ca 0. (e) Centrifuge tubes. Average 3 ε values to obtain εε. (e) Glacial acetic acid.45 µm pore size.4 and 25°C.0 L 0. (s) Patulin working calibrant solution.25. (k) Sodium carbonate. 200.—Dilute 25 mL 0. ca 0. ca 10 mg/mL patulin. D. and 100 µL patulin calibrant standard solution (s). Macer8 FJ supplied by Biocatalysts Ltd.5% (v/v). (c) Potassium dichromate. Pontypridd CF37 5UT. 1 mg/g patulin. (i) Syringe filters.—Dissolve 5 mg patulin (p) in 5 mL ethyl acetate (g). 0.—Accurately weigh ca 78 mg K2Cr2O7 (primary standard) and dissolve in 1. Unit definition: the amount of enzyme which catalyzes the decrease in viscosity of 1% pectin solution by 20% in 5 min at pH 3.: JOURNAL OF AOAC INTERNATIONAL VOL.009M H2SO4 as solvent blank. Exact amount of acetonitrile used will depend on test portion extract and LC column chosen for analysis. Solution stored at 4°C is stable for several months. weigh 10 g test portion into centrifuge tube B(e). CF = correction factor for quartz cells and spectrophotometer obtained by following procedure in B(b).—Into a series of 2 mL volumetric flasks transfer by pipet B(h) 1000. Calculate patulin mass concentration (µg/g) using the following equation: µg/g Patulin = A × MW × 1000× CF ε where A = absorbance of patulin solution at 276 nm. (g) Round bottomed flasks. 2000 350 nm.—60% (v/v).—(4-Hydroxy-4H-furo(3. LC grade. calculate mM to 3 significant figures (MW K2Cr2O7 = 294.05.—1.25mM K2Cr2O7 (b) to 50 mL with 0. (q) Patulin stock solution.125mM K2Cr2O7 (c) to 50 mL with 0. 500. These solutions contain 0. (f) Acetonitrile.—Adjust water with acetic acid (e) to pH 4. Dissolve 1.—Typical activity 1400 U/g.1. Stock solution stored in freezer at –20°C. MW = molecular mass of patulin (154 Dalton). is stable for several months. stopper. (u) HMF solution. C[r].) (c) Quartz cells. Use the same day to make patulin LC calibration standard solutions. Transfer to 25 mL volumetric flask and dilute to volume with ethyl acetate (g). Reagents (a) Ethanol. and immediately dissolve residue in 20 mL ethanol (a). (d) Centrifuge. dissolve in 5 mL pH 4 water (n)..5% (v/v). LC grade. and 0.009M H2SO4 (ca 1 mL H2SO4 diluted to 2 L). (d) Potassium dichromate.—Transfer by pipet B(h) 100 µL patulin calibrant solution (r) and 100 µL HMF solution (u) to 10 mL volumetric flask and evaporate to dryness under stream of N. εε = average of the 3 ε values calculated above.—Dilute 25 mL 0. no preparation is required. Treforest Industrial Estate. Dissolve residue and dilute to volume with pH 4 water (n).009M H2SO4 in volumetric flask. For cloudy juices. 13 mm. record absorption spectrum between 350 and 250 nm in 1 cm quartz glass cell B(c) in spectrophotometer B(b) with ethanol (a) in reference path.95 or >1.—Dissolve 5 mg HMF (j) in 25 mL ethyl acetate (g). (t) Patulin LC calibration calibrant solutions.4. 800. For apple puree.—0. stopper. C = mM concentration of K2Cr2O7 solution.—Evaporate 500 µL calibrant solution (r) or aliquot. check either technique or instrument to determine and eliminate cause. . 83. against 0. Leave overnight at room temperature or for 2 h at 40°C.25mM.5 g sodium carbonate (k) in 100 mL H2O. (l) Sodium sulfate. ca 0. then centrifuge at 4500 × g for 5 min. add this to separating funnel. Ref. Let layers separate. NO. reduction of acetonitrile at fixed flow rate will also improve separation.MACDONALD ET AL. Pour top layer into round-bottomed flask B(g) through a funnel and filter paper containing 15 g anhydrous Na2SO4 C(l). drain the lower layer to waste. only 12 completed the study.025) using Cochran’s.’ ‘c. and Grubbs tests progressively (14). If necessary. HMF and patulin should elute as 2 separate peaks with baseline separation. If peak area of extract is outside range of standard curve. After the wash. number of statistical outlier laboratories.—Using chosen LC conditions.e. Rinse conical flask used to collect ethyl acetate phases with additional 5 mL ethyl acetate. and mix thoroughly.75 mL/min if HMF and patulin do not separate. and collect in round-bottomed flask. wash with 2 × 10 mL ethyl acetate C(g). results for Laboratory 8 for analysis of apple puree were removed from statistical consideration on technical grounds.’ and ‘d’ are blind duplicates for naturally contaminated materials. 83. Let the layers separate and drain lower aqueous layer into empty conical flask and top layer into conical flask containing ethyl acetate layer from first extraction. then drain the lower aqueous layer into conical flask.02.. After layers separate. Details of food matrixes. Leave solution at room temperature overnight or for 2 h at 40°C. and percent recovery are presented in Table 2000. drain lower aqueous layer to waste. Preparation of extract for LC analysis. Transfer aqueous layer back into separating funnel. not detectable). which = 1 for undiluted test portion. On some columns. Check filter with standard solution to assess any loss of patulin before filtering test extracts. By using measured peak areas (or peak heights) from recorder. Transfer to LC vial. Calculation Inject 50 µL each patulin working standard solution C(t). . 10 = ratio of test portion in test solution (5 g apple juice or apple puree is represented by 0. inject 50 µL HMF-patulin solution C(v).5 min. In addition. standard deviations for repeatability (Sr) and reproducibility (SR). filter solution through a syringe filter B(i) before analysis by LC. Inject 50 µL extract. therefore. The collaborative trial results were examined for evidence of individual systematic error (p < 0. AOAC Int. add this to ethyl acetate extract in separating funnel. perform this stage as quickly as possible to avoid losses. All data submitted for the study for the 3 commodities are presented in Tables 1–3. dilute extract with pH 4 water. Read patulin concentration in extract directly from plotted graph. who found the section describing dilution of standard solutions difficult to follow. Blanks (identified as ‘a’) were spiked with patulin at 75 ng/g in each case. Combine the 3 ethyl acetate phases in separating funnel. E. Note: It may be necessary to wash LC system with 100% acetonitrile after each test extract injection to ensure that no materials are retained on column.5 mL test solution). pairs of results were removed as noncompliant before statistical analysis. average analyte concentration. Samples ‘b. Transfer aqueous layer back into same separating funnel and re-extract with second 20 mL portion of ethyl acetate. d = dilution factor. A total of 5 pairs of results were removed as being noncompliant.—See below for typical LC operating conditions. 6. Add 4 mL Na2CO3 solution C(m) to separating funnel and shake 0.—Evaporate extract to dryness and redissolve in final volume of 1 mL (500 µL for puree) pH 4 water C(n). 83. Repeat this extraction procedure for a third time. rinse conical flask with 10 mL ethyl acetate C(g). Laboratory 4 indicated problems with apple puree in terms of matrix interferences and believed that pectinase treatment had not worked well. reinject. relative standard deviations for repeatability (RSDr) and reproducibility (RSDR).: J. LC operating conditions. and pour top layer through the Na2SO4 into the round-bottomed flask. HORRAT values.—Pipet 10 mL clear juice (or cloudy juice or puree as prepared above) into 100 mL separating funnel. 1389–1391(2000) Results and Discussion Collaborative Trial Results Of the 14 participants who received the test materials. prepare standard curve by plotting peak areas vs concentrations of patulin working standard solutions. Column evaluation. Note: Patulin is not stable in alkaline solutions.: JOURNAL OF AOAC INTERNATIONAL VOL. Add 20 mL ethyl acetate C(g) and shake 1 min. Statistical Analysis of Results Precision estimates were obtained by using one-way analysis of variance approach according to IUPAC Harmonized Protocol (14). Pairs of results identified as outliers are indicated by footnotes in Tables 1–3. and shake 0. Extraction of patulin from the test solution. Calculate concentration of patulin in test sample (ng/g) as follows: ng/g Patulin = C T × 1000 ×d 10 where CT = concentration of patulin in extract (ng/g). and re-analyze diluted extract solution. re-equilibrate system with mobile phase before next injection. Let layers separate and drain them into 2 separate conical flasks. number of noncompliant data sets. The data are given as individual pairs of results for each laboratory (identified as 1–14). 2000 1391 add 150 µL pectinase enzyme solution C(h) followed by 10 mL H2O. Let layers separate. Several laboratories suggested extending the range of LC calibration standards to avoid dilution of extracts before LC. Where no numerical values were reported (i. It may be necessary to raise acetonitrile content of LC eluent C(o) (≤10%) and decrease flow rate to 0. Comments from Collaborative Trial Participants All participants found the method clear and easy to understand except for Laboratory 11.5 min. Laboratory 8 followed the method but could not detect patulin in apple puree. 3 133.6c 93.3 6.2 70 69.7 3.6 24.6 147.6 <5 <5 4 80.1 120.8 25.8c 9 73.1 84 116.7 70.9 8.2 39 2.3 40.7 62.2 151.1 139. Table 2.5 60 113.5 72.8 95 96 109.8 11.7 2.4 nd 12.1 99.6 0c 2.8 21.8 8.8 52. Collaborative trial results of liquid chromatographic determination of patulin in cloudy apple juice Patulin concentration.6 28.7 5.1 129.9 109.8 34.4 97.2 13 47. .3 87.7 140.1 103.8 18.5 4.9b 18 26.9 86.4 48.8 65.4 0 0 21.6 29.1 29 31 79.4 20.8 141.6 43.3 65.8 44.1 40.6 73. ng/g Lab ID 75 aa 75 a 1 68.8 116.2 145 0 11.4 6.3 28.5 29.1 62.9 28 27.2 20. 6.9 23.7 111.3 13 69.7 15.4 53.8c 6.2 <4.3 84.6 28.6 11 51.3 82 116.6 59.4 29.7 <3 a <3 b b b c c d d 21.8 81 <5 <5 11 51.9 37.7 3. 83.8c 0c 28.9 95.8 38.6 26.2 64 64.6 72. d = blind duplicate pairs of naturally contaminated samples.4 d c 108.5 <5 <5 24.3 138.8 64.1 21 38.4 49.7 1.9 8.2 8.2 44.2b 28.9 9 82.8 5 50.2 78. Results identified as outliers and not included in statistical analysis.5 7 8 65.2 60.6 8.4 3 62.1 44.4 62.8 0 5.5 6 6 73.7 124. c.4 25.8 23.6 14 50.2 58.2 82.4 72.5 59.1 27.6 120 115.2 43. Results identified as outliers and not included in statistical analysis.4 d a.3 116.6 113.1 66. NO. nd = not detectable.2 0 0 43.5b b b c c d d 27.3 24.1 50.9 109.6 21.6 24. Results identified as noncompliant.4 107 4 73. d = blind duplicate pairs of naturally contaminated samples.9 10 66.6 75. b.2 36.6 32.5 8.6 57.8 14 61.7 91b 149b 17.8 47 3.8 29.7 68.6 b 50.4c 30.1 9 106.6 54.7 a b c d 5.3 0c 64.6 78.4 117.2 7.7b 78 78.6 34.4 69.6 123.: JOURNAL OF AOAC INTERNATIONAL VOL.8 116.8 136.4 25.6 8 49.6 6.5b <4.1 96.1 29.6 c c 5 8.4 6.7 52.6 53.1 11.1 7 71.5 <4.6 113.8 nd nd 26.8 26. Collaborative trial results of liquid chromatographic determination of patulin in clear apple juice Patulin concentration.8 59.2 26. ng/g Lab ID 1 aa 75 75 60.4 68.8 nd nd 24.2 6 76 76.8 32.4 73.7 10 68. b.7c 14. c.9 a b a.1 128 51.8 68 4 4 3 67.8 67 64. 2000 Table 1.1392 MACDONALD ET AL.6 54.2 23. whose results were removed as outliers in 4 instances for Laboratory 5. the higher RSDr and RSDR values were found at the lower patulin concentrations: at 26 ng/g. for First Action AOAC Method.4 2.9c 89. however.f <2.6 25.7 78.7 nd Lab ID c c b b d c c d d 96.3 18. Collaborative trial results of liquid chromatographic determination of patulin in apple puree Patulin concentration.2 38. In all cases. and in 3 instances for Laboratory 8.: JOURNAL OF AOAC INTERNATIONAL VOL.7 8 0e 0e 0e 0e 0e 0e 0e 0e 9 81.3 ng/g.8 46.9 0 0 24.2 2. in the control clear apple juice.4 79. 2000 1393 Table 3. Thus.1 78.2 4. meaning not detectable in the blank materials. and RSDR values of 36 and 33% for apple puree at patulin contents of 23 and 38 ng/g.8 104 0e 10 57.1 128.5 <5 19 14. respectively. All but Laboratory 13 correctly identified blank pairs of cloudy juice.5d 5 6. in the lowest naturally contaminated juice as compared with 26 ng/g patulin.1 50.7 37.5 24. nd = not detectable.9 127. The values for recoveries of patulin derived .5 19.6 42.9 23.4 47. except apple puree containing 121 ng/g patulin for which an RSDR of 29% was obtained.3 19 d d 0e 59. ng/g 75 75 aa a 1 62.5 132.4 6 69. and only Laboratory 14 experienced some difficulty with blank apple puree. Results identified as outliers and not included in statistical analysis. compared with RSDr values of 11 and 8% and RSDR values of 25 and 11% for naturally contaminated apple juice containing average levels of 54 and 128 ng/g patulin.9 <3 <3 19. for clear apple juice.2 4 75.8 59. d = blind duplicate pairs of naturally contaminated samples.9 129.8 21. b. Not surprisingly. at patulin contents >50 ng/g.2 11 67. No statistical analysis was carried out for blank samples. the AOAC Method covered a wider range (20–200 ng/g) of patulin spike concentrations. RSDr values of 11 and 13%.3 nd 10. all RSDr values were <20% and all RSDR values <25%. the values for precision characteristics for this trial compared favorably with those for First Action AOAC Method (12). indicated clearly that all but 2 participants (Laboratories 6 and 13) identified the blank pairs as not containing detectable patulin in clear juice or containing levels that were detectable but close to limits of determination.5 a b c d e f a. Thus.3 <2.1 85. c.4 77.g. which were intended as controls in the study. The principal difference between this study and others was the application of the collaborative trial to cloudy apple juice and apple puree.6 17.4 137. and RSDR values of 22 and 21% were obtained for spiked apple juice containing 50 and 100 ng/g patulin.5 46.4 127.6 ndb nd 22. Where it was possible to make direct comparisons e.02. There was no obvious correlation between laboratories that found higher than average patulin levels in the control juices and their results for positive samples. Laboratories 6 and 13 found mean levels of 18. indicating poor replicate pairs which more likely indicated that a particular laboratory experienced difficulty than a specific technical problem with the samples or the method itself.5 35.0 and 25. The corresponding RSDR ranged from 11 to 36%.2 140 7 78. which were previously more difficult matrixes to analyze. Results not included on technical grounds. RSDR values of 33 and 35% were obtained for clear and cloudy apple juice. The results.4 39.5 39.2 0 0 21.7d xd.6c 19.3 34 41.2 37. outlier results were removed by Cochran’s test.6 13 61. respectively. 83. the RSDr ranged from 6 to 35%. Precision Characteristics of the Method There were differences in reporting limits (given in some cases as ‘0’ and in others as ‘less than’) but in both cases.6 62..2 129. NO.9 14 57.3 <5 42.MACDONALD ET AL.6 0 0 79. No technical reasons were obvious for Laboratories 5 and 8. respectively.2 ng/g. which was the mean result from14 laboratories.2c 161.8 48. Recoveries of 91–108% were reported for the AOAC First Action Method compared with 89% for this study. respectively. respectively.2 151.9 17.4 32. Based on results for spiked samples (blind pairs at one level) and naturally contaminated samples (blind pairs at 3 levels).7 44. any co-extractive compounds from juices that interfered with the patulin determination would have been more in evidence at lower patulin levels. but mean levels of 29.9 30.3 135 120.4 29.3c 25. Thus.7 23. x = result not reported.7 152.8 and 17.3 151. Although LC chromatograms showed no evidence of interferences.4 35.4 3 67.3 153. Results identified as noncompliant.2 61.4 nd nd 9.4 65.9 64.6 3.5 0 0 0 0c 29. The precision data for all samples are summarized in Table 2000. 6.5c 103. There were negligible differences in recoveries for all 3 matrixes.W.. Switzerland. Lausanne. K. Rome. V. Bilbao. UK). A. Gaithersburg. & Albert.5 to 1. Because all HORRAT values were <2. This is the first time validation of a method has been carried out for these matrixes. 621–625 Tarter. J.. 538. S. which indicates an acceptable precision. & Gomez-Cordoves. Denmark Maria Lígia Martins. Laboratório de Salud Pública. Massy. International Organization for Standardization. T. the HORRAT values for patulin ranged from 0. A.G.: JOURNAL OF AOAC INTERNATIONAL VOL. which were acceptable for determining patulin in all 3 product types.B. 6.P. 664.R.. Scientific Institute of Public Health–Louis Pasteur. Belgium Kirsti Nuotio.E. and also available from national standardization organizations Kubacki. E. Anal. The authors also express their appreciation to the following collaborators for their participation in the study: Carlo Brera.. (1974) J. V. 1055–1056 Bartolome. Italy IUPAC (1995) Pure Appl. & Scott. Spain J. & Canela. (1996) J. Geneva. (1979) Biomed. B..0.. (1991) J. Mathieson (CSL Food Science Laboratory. 60. 83. 74. AOAC Int. Acknowledgments This project was financially supported by the European Commission. Trucksess. (1994) J. 441–446 Price. 871–876 International Standard ISO/DIS 8128-1(E) (1993) Apple Juice.. International Organization for Standardization. Ribera. H. Danish Veterinary and Food Administration. R. T. Italy Kevin Jørgensen. Espoo. IRTA. A. 331–343 Horwitz. Istituto Superiore di Sanita. R. Chem. Laboatorio Nacional De Veterinária. Sanchis. 6.S.R.. Portugal Ma Luz Macho. UK). Leatherhead. Germany References (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) Scott. France Nicola Tucker. Off. F. Assoc. Lisbon. 730. UK). NO. Geneva. Apple Juice Concentrates and Drinks Containing Apple Juice—Determination of Patulin Content—Part 2: Method using TLC. A. R.. York. Estrella. Germany Jean-Yves Michelet. Gesellschaft fur Lebensmittel-Forschung MbH. AOAC INTERNATIONAL. M. Berlin. Off. The Netherlands Anne Wennemar. Switzerland. MD. Voogt. 573–574 Moller. I. (1991) J. Finnish Customs Laboratory.. but has an additional step required for matrix extension to cloudy juice and apple puree. When outliers were excluded.. Mass Spectrom. this method is clearly acceptable as an AOAC Official Method. who took responsibility for the pretrial workshop for participants. and were better than or comparable to values reported in the AOAC-IUPAC Official First Action Method (12). (1988) Pure Appl.J. Perez-Ilzarbe. 64.und Staatliches Veterinäruntersuchungsamt.M.1394 MACDONALD ET AL. 63. Chem. AOAC Int.. 2000 from the spiked samples ranged from 80 to 92% (Table 2000. Agric. Nestlé Research Center. Netherlands). Apple Juice Concentrates and Drinks Containing Apple Juice—Determination of Patulin Content—Part 1: Method by High Performance Liquid Chromatography. Method 995. van Egmond (RIVM... 53–58 International Standard ISO 8128-2:1993(E) (1993) Apple Juice. 39–43 Rovira. Chem.. Spain Paul Majerus. Switzerland Louis Szymanski. S. Srborg. The method is a minor modification of the existing First Action Method for clear juice.W. C..M.02). W. H. Chemisches Landes. 79. 451–455 Official Methods of Analysis (1995) 16th Ed. & Acar. (1993) J. 67. 57. who prepared calibrant standards. F. Williams (formerly of Leatherhead Food RA.. & Page. Keuringsdienst van waren.3. Germany Lutz Mevissen. Food Chem. & Joefsson. Bilthoven. Hernandez. Anal. Finland Alain Pittet.. Recommendation It is recommended that this LC method for patulin at >50 ng/g be adopted Official First Action for determination of patulin in cloudy apple juice and apple puree. Boenke (EC SMT-Program Scientific Officer for the project). UK Remei Viladrich. Standards Measurement and Testing Program (Brussels. who performed the statistical analysis. Chromatogr. and also available from national standardization organizations Brause. Trier. Reading Scientific Services Ltd. including A. Assoc.L.. Chem. Reading. FAO Food and Nutrition Paper No. I. York. (1980) J. who assisted at the workshop. 214–216 Gokmen. Rotterdam. 718–744 . Chromatogr. There have been many who contributed to the success of this study. Chromatogr. Brussels. E. Buckle (CSL. Thomas. & Goszcz. Lleida. Chemiches Untersuchungsamt.10 FAO (1997) Worldwide Regulations for Mycotoxins 1995—A Compendium. F. K. M. 41. (1996) J. Interpretation of Results Acceptability of the precision characteristics of the method were assessed on the basis of the HORRAT values (15) which compare the RSDR obtained for a particular level and matrix with the value statistically predicted on the basis of collaborative trial studies taken from the published literature. P. Laboratoire Interrégional de la Repression des Frauds de Paris-Massy. Mhnster. Bengoechea. Rome. and A. Belgium) SMT Project CT96-2045. P.