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. Jrgensen; 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 610% and RSDR values 1125%) 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; 47) have
all been successfully used for patulin analysis. For TLC with
3-methyl-2-benzothiazolinone hydrazoneHCl 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 1218% and RSDR values of 1417%. 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

1388 MACDONALD ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 83, NO. 6, 2000

clear apple juice. The study reported here, however, was extended to cloudy apple juice and apple puree, which have presented problems in the past in terms of recoveries and are
more prone to interference. The method has been modified
and includes pectinase enzyme treatment to improve the clarity of juices and puree before analyte isolation. The collaborative trial results for clear juices in the present study may be
compared directly with those of Brause et al. (11), but there
are no comparable studies for cloudy juices or apple puree.
Regulations control patulin contamination in about
11 countries worldwide (13) with limits in most cases set at
50 ng/g for fruit (apple) juices, although in the Czech Republic
a limit of 30 ng/g is applied to infant foods. At present, there
are no European Union (EU) limits for patulin, although
5 member states of the EU have limits in their national regulations and others (e.g., the UK) have voluntary guideline limits
and there are some moves to harmonize limits in the EU. The
European Standardization Organization CEN has indicated a
need for a validated method for patulin as a European standard, which would cover not only clear juice but also the increasingly popular organic cloudy juices and apple puree,
which may be produced from damaged and molded apples.
Test Materials for the Collaborative Trial

Contaminated Cloudy Apple Juice

A blend of juices from Cox, Bramley, and Crispin apple
varieties was obtained from a local apple juice producer, analyzed by the method reported here, and found to contain
patulin at a concentration of >180 ng/g. Commercially available blank apple juice (a blend of English apple juices) contained patulin at a concentration of <5 ng/g. 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. The juice was analyzed periodically to assess the
patulin concentration and homogeneity. Further volumes of
contaminated or blank juice were added as necessary to reach
the required concentration of patulin.

Clear Juice Samples

The clear juice test samples were prepared in-house from a
naturally contaminated cloudy juice. A subsample of the naturally contaminated cloudy apple juice was clarified. 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. After incubation at 40C for 2 h,
the juice was centrifuged at 7200 g for 10 min. Analysis of
the resulting clear juice showed that it contained patulin at a
concentration of >180 ng/g. This clear juice was diluted with
blank clear apple juice, which was obtained from a commercial source and contained patulin at <5 ng/g. As with the
cloudy juice, the clear apple juices were mixed together in a
25 L container for 2 h on an automatic shaker, and aliquots
were analyzed periodically to assess the exact patulin concentration and homogeneity. Further volumes of contaminated or

blank juice were added as necessary to reach the required concentration of patulin.

Apple Puree Samples

The puree test samples were produced within the laboratory. A supply of windfall apples, naturally contaminated with
patulin, was obtained from a local source, peeled, cored, and
chopped, and then pureed in a food processor and pasteurized
at 60C for 30 min. The resulting puree contained patulin at
>9000 ng/g.
Blank test samples were prepared from fresh apples purchased from a commercial source and made into a puree. The
apples were peeled, cored, chopped into 8 segments, and pureed in a food processor. To prevent discoloration, 0.5 g ascorbic acid was added to every 1 kg of apple immediately after
chopping. This is done in commercial practice and does not
interfere with patulin analysis. The puree was then
microwaved on full power for 6 min, and was found to contain
patulin at <5 ng/g.
The test materials were produced by mixing together
known quantities of contaminated and blank purees. The puree was initially mixed for 3 h in a food processor and then for
an additional 1 h in a Silverson blender. As with the apple
juice, the puree was analyzed periodically to assess the patulin
concentration and homogeneity. Further quantities of contaminated or blank puree were added as necessary to reach the required content of patulin.
Organization of Collaborative Trial
Fourteen collaborators from 12 different European countries representing a cross-section of government, food control,
and food industry affiliations took part in the collaborative
trial. Before the collaborative trial, each laboratory received a
set of familiarization test samples, comprising a blank and naturally contaminated test sample of both cloudy apple juice and
apple puree, a calibrant solution for spiking, a vial of pectinase
enzyme, and a solution of 5-hydroxymethyl furfural (HMF).
A precollaborative trial workshop was also organized, where
any problems encountered with the familiarization test samples were discussed, and details of the collaborative trial were
outlined.
For the collaborative trial, each participant received:
(a) Eight units each of clear juice, cloudy juice, and apple
puree labelled a, b, c, d.
(b) Two known blank materials of clear juice, cloudy apple juice, and apple puree for spiking experiments.
(c) An ampule of patulin calibrant solution (10 g/g
patulin in absolute alcohol).
(d) Two ampules marked Spike solution A and Spike
solution B for spiking experiments.
(e) An amber vial marked HMF standard containing ca
5 mg 5-hydroxymethyl furfural.
(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.
(g) A copy of the method.

MACDONALD ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 83, NO. 6, 2000 1389

(h) A set of additional instructions for the spiking protocol.

(i) A report form for analytical data, criticisms, and
suggestions.
(j) A collaborative study materials receipt form.
Each participant was required to prepare one extract from
each material, perform the cleanup, and analyze the extracts
by LC. Additionally, each participant was required to spike
each of the indicated blank materials for each of the 3 matrixes with syringe, using the provided Spike solutions A and
B. After adding the spike solution, 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.
Participants were advised to analyze the 3 different matrixes on separate days. This would mean analyzing a batch of
10 samples (8 coded plus 2 spike samples per matrix) on separate days, thus completing the experimental work for the trial
in 3 days, assuming LC analysis was performed overnight between days.

See Table 2000.02 for results of the interlaboratory study

supporting the acceptance of the method.
A. Principle

Apple juice or puree is extracted with ethyl acetate and

then cleaned up by extraction with sodium carbonate solution. (Cloudy apple juice and apple purees are pretreated
with pectinase enzyme.) The ethyl acetate extract is dried
with anhydrous sodium sulfate. After evaporation of the
solvent, patulin is quantitatively determined by LC with
UV detection.
B. Apparatus

(a) LC apparatus.(1) Pump.LC pump(s) and eluent

reservoir. (2) Injection system.Test sample applicator.
(3) Separation columns.A 4.3 mm id octadecylsilane
(ODS) precolumn with 5 m particle stationary phase. An analytical reversed-phase LC column such as ODS, fully end
capped with 5 m particle stationary phase, 25 nm pore size,
and 12% carbon loading, or ODS super end capped with 5 m
particle stationary phase, 12 nm pore size, and carbon loading
of 17%. (4) Detector.UV detector at 276 nm and data integration system.
(b) Spectrophotometer.350250 nm. Calibrate as follows: Determine absorbance (A) of the 3 solutions of K2Cr2O7
in H2SO4, C(b), C(c), and C(d), 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, cloudy apple juice, and apple puree.)

Table 2000.02. Interlaboratory study results for patulin in clear and cloudy apple juices and apple puree
0, ng/g

ID

No. of labsa(b)

Sr

RSDr, %

SR

RSDR, %

HORRAT

Rec., %

Clear Apple Juice

c

67

12 (0)

8.4

13

15.3

23

0.95

89

26

12 (0)

3.7

14

8.4

33

1.18

nc (c)

54

12 (0)

11

13.6

25

1.02

nc (d)

128

10 (2)

9.9

14

11

0.50

12.5

21

0.85

80

75 ng/g (a)
d

nc (b)

8
Cloudy Apple Juice

75 ng/g (a)

60

11 (1)

7.8

13

nc (b)

26

12 (0)

8.9

35

nc (c)

69

9 (2)

4.3

nc (d)

106

10 (2)

10.2

10

35

1.25

10

8.9

14

0.61

12.9

12

0.54

Apple Puree
75 ng/g (a)

69

9 (1)

7.5

11

9.2

13

0.56

92

nc (b)

23

8 (1)

6.4

27

8.5

36

1.23

nc (c)

38

9 (2)

3.8

10

12.6

33

1.27

nc (d)

121

10 (0)

23.6

19

34.8

29

1.31

a(b)
c
d

a = number of labs retained after eliminating outliers; (b) = number of labs removed as outliers.
Laboratories received 4 sets of duplicate test samples labeled a, b, c, and d.
nc = naturally contaminated.

1390 MACDONALD ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 83, NO. 6, 2000

350 nm, against 0.009M H2SO4 as solvent blank. Calculate

molar absorptivity () at each concentration as follows:
=

A 1000
C

where A = absorbance at maximum near 350 nm, C = mM

concentration of K2Cr2O7 solution.
If the 3 values vary by more than guaranteed accuracy of A
scale, check either technique or instrument. Average 3 values to obtain . Determine correction factor (CF) for particular instrument and cells by substituting in equation:
CF =

3160

where 3160 = value for of K2Cr2O7 solutions, = average of

the 3 values calculated above. If CF is <0.95 or >1.05, check
either technique or instrument to determine and eliminate
cause. (Use same set of cells in calibration and determination
of purity of patulin, C[r].)
(c) Quartz cells.Optical path length 1 cm.
(d) Centrifuge.4500 g.
(e) Centrifuge tubes.50 mL with screw cap.
(f) Rotary evaporator.
(g) Round bottomed flasks.
(h) Hand-held pipets.25, 50, 1000 L.
(i) Syringe filters.0.45 m pore size, 13 mm, PTFE.
C. Reagents

(a) Ethanol.99.7% (v/v), LC grade.

(b) Potassium dichromate, ca 0.25mM.Accurately
weigh ca 78 mg K2Cr2O7 (primary standard) and dissolve in
1.0 L 0.009M H2SO4 (ca 1 mL H2SO4 diluted to 2 L); calculate mM to 3 significant figures (MW K2Cr2O7 = 294.2).
(c) Potassium dichromate, ca 0.125mM.Dilute 25 mL
0.25mM K2Cr2O7 (b) to 50 mL with 0.009M H2SO4 in volumetric flask.
(d) Potassium dichromate, ca 0.0625mM.Dilute 25 mL
0.125mM K2Cr2O7 (c) to 50 mL with 0.009M H2SO4 in volumetric flask.
(e) Glacial acetic acid.99.5% (v/v).
(f) Acetonitrile.99.5% (v/v), LC grade.
(g) Ethyl acetate.
(h) Pectinase enzyme solution, endogalacturonase.Typical activity 1400 U/g. Unit definition: the amount of enzyme
which catalyzes the decrease in viscosity of 1% pectin solution by 20% in 5 min at pH 3.4 and 25C. Macer8 FJ supplied
by Biocatalysts Ltd., (Main Ave, Treforest Industrial Estate,
Pontypridd CF37 5UT, Wales, UK) is suitable.
(i) Perchloric acid.60% (v/v).
(j) 5-Hydroxymethyl furfural (HMF).
(k) Sodium carbonate.Anhydrous.
(l) Sodium sulfate.Anhydrous.
(m) Sodium carbonate solution.1.5%. Dissolve 1.5 g
sodium carbonate (k) in 100 mL H2O.
(n) pH 4 water.Adjust water with acetic acid (e) to
pH 4.

(o) Elution solution for LC.Add 310% acetonitrile (f)

to water containing 0.095 parts per volume perchloric acid
60% (i). Exact amount of acetonitrile used will depend on test
portion extract and LC column chosen for analysis. Degas this
solution before use.
(p) Patulin.(4-Hydroxy-4H-furo(3,2-c)pyran-2(6H)-one),
99%.
(q) Patulin stock solution.Dissolve 5 mg patulin (p) in
5 mL ethyl acetate (g). Transfer to 25 mL volumetric flask and
dilute to volume with ethyl acetate (g). Stock solution stored
in freezer at 20C, is stable for several months.
(r) Patulin calibrant solution, ca 10 mg/mL patulin.Evaporate 1000 L stock solution (q) to dryness under N, and immediately dissolve residue in 20 mL ethanol (a). Solution
stored at 4C is stable for several months.
To determine exact mass concentration in calibrant solution, 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. Calculate patulin mass concentration (g/g) using the following equation:
g/g Patulin =

A MW 1000 CF

where A = absorbance of patulin solution at 276 nm, MW =

molecular mass of patulin (154 Dalton), CF = correction factor for quartz cells and spectrophotometer obtained by following procedure in B(b), = molecular absorbance coefficient of
patulin solution at the wavelength maximum (276 nm) of absorption spectrum (14 600 L mol1 cm1 in ethanol).
(s) Patulin working calibrant solution, 1 mg/g
patulin.Evaporate 500 L calibrant solution (r) or aliquot,
which is equivalent to absolute amount of 5 g patulin to dryness, dissolve in 5 mL pH 4 water (n), stopper, and shake vigorously. Use the same day to make patulin LC calibration
standard solutions.
(t) Patulin LC calibration calibrant solutions.Into a series of 2 mL volumetric flasks transfer by pipet B(h) 1000,
800, 500, 200, and 100 L patulin calibrant standard solution
(s). Dilute to mark with pH 4 water (n), stopper, and shake
vigorously to mix. These solutions contain 0.5, 0.4, 0.25, 0.1,
and 0.05 g/mL patulin, respectively. Transfer standards to
vials for LC analysis and use on same day as preparation.
(u) HMF solution.Dissolve 5 mg HMF (j) in 25 mL
ethyl acetate (g).
(v) HMF-patulin solution.Transfer by pipet B(h)
100 L patulin calibrant solution (r) and 100 L HMF solution (u) to 10 mL volumetric flask and evaporate to dryness
under stream of N. Dissolve residue and dilute to volume with
pH 4 water (n).
D. Procedure

Preparation of test portion.For clear apple juice, no

preparation is required. For cloudy juices, measure 20 mL test
portion into centrifuge tube B(e) and add 150 L pectinase enzyme solution C(h). Leave overnight at room temperature or
for 2 h at 40C; then centrifuge at 4500 g for 5 min. For apple puree, weigh 10 g test portion into centrifuge tube B(e),

MACDONALD ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 83, NO. 6, 2000 1391

add 150 L pectinase enzyme solution C(h) followed by

10 mL H2O, and mix thoroughly. Leave solution at room temperature overnight or for 2 h at 40C; then centrifuge at
4500 g for 5 min.
Extraction of patulin from the test solution.Pipet 10 mL
clear juice (or cloudy juice or puree as prepared above) into
100 mL separating funnel. Add 20 mL ethyl acetate C(g) and
shake 1 min. Let layers separate and drain them into 2 separate
conical flasks. Transfer aqueous layer back into same separating funnel and re-extract with second 20 mL portion of ethyl
acetate. 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. Repeat this extraction procedure for a third time. After layers separate, drain
lower aqueous layer to waste. Combine the 3 ethyl acetate
phases in separating funnel. Rinse conical flask used to collect
ethyl acetate phases with additional 5 mL ethyl acetate; add
this to ethyl acetate extract in separating funnel.
Add 4 mL Na2CO3 solution C(m) to separating funnel and
shake 0.5 min. Let layers separate; then drain the lower aqueous layer into conical flask. Pour top layer into
round-bottomed flask B(g) through a funnel and filter paper
containing 15 g anhydrous Na2SO4 C(l). Transfer aqueous
layer back into separating funnel, rinse conical flask with
10 mL ethyl acetate C(g), add this to separating funnel, and
shake 0.5 min. Let layers separate, drain the lower layer to
waste, and pour top layer through the Na2SO4 into the
round-bottomed flask, wash with 2 10 mL ethyl acetate
C(g), and collect in round-bottomed flask.
Note: Patulin is not stable in alkaline solutions; therefore,
perform this stage as quickly as possible to avoid losses.
Preparation of extract for LC analysis.Evaporate extract
to dryness and redissolve in final volume of 1 mL (500 L for
puree) pH 4 water C(n). Transfer to LC vial. If necessary, filter solution through a syringe filter B(i) before analysis by LC.
Check filter with standard solution to assess any loss of
patulin before filtering test extracts.
LC operating conditions.See below for typical LC operating conditions. 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. After the wash,
re-equilibrate system with mobile phase before next injection.
Column evaluation.Using chosen LC conditions, inject
50 L HMF-patulin solution C(v). HMF and patulin should
elute as 2 separate peaks with baseline separation. It may be
necessary to raise acetonitrile content of LC eluent C(o)
(10%) and decrease flow rate to 0.75 mL/min if HMF and
patulin do not separate. On some columns, reduction of
acetonitrile at fixed flow rate will also improve separation.
E. Calculation

Inject 50 L each patulin working standard solution C(t).

By using measured peak areas (or peak heights) from recorder, prepare standard curve by plotting peak areas vs concentrations of patulin working standard solutions.
Inject 50 L extract. Read patulin concentration in extract
directly from plotted graph. If peak area of extract is outside

range of standard curve, dilute extract with pH 4 water, reinject, and re-analyze diluted extract 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), 10 = ratio of test portion in test solution (5 g apple juice or apple puree is represented by 0.5 mL test solution); d = dilution factor,
which = 1 for undiluted test portion.
Ref.: J. AOAC Int. 83, 13891391(2000)

Results and Discussion

Collaborative Trial Results

Of the 14 participants who received the test materials, only
12 completed the study. All data submitted for the study for
the 3 commodities are presented in Tables 13. The data are
given as individual pairs of results for each laboratory (identified as 114). Blanks (identified as a) were spiked with
patulin at 75 ng/g in each case. Samples b, c, and d are
blind duplicates for naturally contaminated materials.

Statistical Analysis of Results

Precision estimates were obtained by using one-way analysis of variance approach according to IUPAC Harmonized
Protocol (14). Details of food matrixes, average analyte concentration, standard deviations for repeatability (Sr) and
reproducibility (SR), relative standard deviations for repeatability (RSDr) and reproducibility (RSDR), number of statistical outlier laboratories, number of noncompliant data sets,
HORRAT values, and percent recovery are presented in Table 2000.02. Where no numerical values were reported (i.e.,
not detectable), pairs of results were removed as
noncompliant before statistical analysis. A total of 5 pairs of
results were removed as being noncompliant. In addition, results for Laboratory 8 for analysis of apple puree were removed from statistical consideration on technical grounds.
The collaborative trial results were examined for evidence of
individual systematic error (p < 0.025) using Cochrans, and
Grubbs tests progressively (14). Pairs of results identified as
outliers are indicated by footnotes in Tables 13.

Comments from Collaborative Trial Participants

All participants found the method clear and easy to understand except for Laboratory 11, who found the section describing dilution of standard solutions difficult to follow. Laboratory 4 indicated problems with apple puree in terms of
matrix interferences and believed that pectinase treatment had
not worked well. Laboratory 8 followed the method but could
not detect patulin in apple puree. Several laboratories suggested extending the range of LC calibration standards to
avoid dilution of extracts before LC.

1392 MACDONALD ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 83, NO. 6, 2000
Table 1. Collaborative trial results of liquid chromatographic determination of patulin in clear apple juice
Patulin concentration, ng/g
Lab ID

75

aa

75

68.8

68

67.4

68.6

<5

<5

80.2

78.1

11.8

50.4

48.2

76

76.4

20.1

71.9

86.8

5.5b

27.2

26.7

52.6

53.9

109.6

113.9

28

27.6

59.2

58.6

120

115.6

8.3

40.8

34.8

64.7

68.8

136.7

124.7

1.9b

18

26.8

38.2

44.7

91b

149b

17.5

29.3

28.8

52.4

53.7

111.2

145

11.6

21.1

62.8

44.7

140.3

133.6

50.5

59.1

106.7b

78

78.6

147.8

141.6

49.8

47

3.3

6.8

11.7

3.9

82.4

73.2

7.8

8.6

32.4

29.7

10

68.5

72.5

<5

<5

24.4

25.1

40.1

50.8

116.1

120.6

11

51.6

28.2

23.5b

<4.2b

28.1

21

38.9

37.1

129.2

151.3

13

69.4

97.8

18.7

15.8

26.6

24.1

27.2

43.8

116.9

109.8

14

61.8

65.4

6.5

8.6

34.2

36.2

70

69.1

139.3

138.9

a
b

a, b, c, d = blind duplicate pairs of naturally contaminated samples.

Results identified as outliers and not included in statistical analysis.

Table 2. Collaborative trial results of liquid chromatographic determination of patulin in cloudy apple juice
Patulin concentration, ng/g
Lab ID
1

aa

75

75

60.6

54.7

<3

a
<3

21.6

26.4

69.3

65.8

95

96
109.4

62.2

60.8

nd

nd

24.3

24.2

64

64.4

107

73.4

72.9

8.9

8.1

29.9

23.7

70.2

82.1

103.6

8.4

6.8

nd

nd

26.5

73.6

72.4

21.5

7
8

65.6

43.2

43.4

49.6

6.5

4.7
5.4

108.7
2.8c

6.4

nd

12.8

29.6

73.1

84

116.1

128

51.8

23.6

57.5

60

113.4

117.8

32.2

8.6

24.7c

14.4c

30.6

29.3

84.3

82

116.6

123.1

29

31

79.6

75.6

78.6

113.6

0c

2.8c

73.1

66.9

10

66.8

81

<5

<5

11

51.1

44.5

<4.2

<4.2

20.8

21.8

59.6

54.3

87.1

96.2

13

47.8

67

64.8c

0c

28.8

25.3

0c

64.6c

93.3

116.6

14

50.2

39

2.7

3.4

25.6

28.7

62.4

62.1

99.9

95.7

a
b
c
d

5.4

a, b, c, d = blind duplicate pairs of naturally contaminated samples.

nd = not detectable.
Results identified as outliers and not included in statistical analysis.
Results identified as noncompliant.

MACDONALD ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 83, NO. 6, 2000 1393
Table 3. Collaborative trial results of liquid chromatographic determination of patulin in apple puree
Patulin concentration, ng/g
75

75

aa

62.2

61.9

<3

<3

19.9

23.7

37.4

39.4

127.4

67.4

65.6

ndb

nd

22.6

25.6

42.2

38.3

135

120.2

75.1

85.6

79.3c

25.9c

89.6c

19.2c

161.2

151.6

3.7

nd

Lab ID

d
96.9

17.5d

6.8

21.4

nd

nd

9.3

nd

10.4

69.4

77.9

24.5

39.1

50.5

46.2

140

78.4

79.5

0c

29.5c

103.7

0e

0e

0e

0e

0e

0e

0e

0e

81.6

62.2

4.2

2.4

29.7

23.7

44.5

35.1

128.9

127.3

19

0e

59.8
104
0e

10

57.1

78.5

<5

19

14.3

<5

42.8

48.3

153.7

152.2

11

67.7d

xd,f

<2.3

<2.3

18.5

19.3

34

41.5

132.7

78.6

13

61.9

64.2

21.5

24.4

35.2

37.9

129.3

151.9

14

57.8

59.6

17.4

2.9

30.4

32.8

46.4

47.4

137.2

129.5

a
b
c
d
e
f

a, b, c, d = blind duplicate pairs of naturally contaminated samples.

nd = not detectable.
Results identified as outliers and not included in statistical analysis.
Results identified as noncompliant.
Results not included on technical grounds.
x = result not reported.

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,
meaning not detectable in the blank materials. No statistical
analysis was carried out for blank samples, which were intended as controls in the study. The results, however, 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. All but Laboratory 13 correctly identified blank pairs of cloudy juice, and only Laboratory 14 experienced some difficulty with blank apple puree.
The precision data for all samples are summarized in Table 2000.02. No technical reasons were obvious for Laboratories 5 and 8, whose results were removed as outliers in 4 instances for Laboratory 5, and in 3 instances for Laboratory 8.
In all cases, outlier results were removed by Cochrans test,
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.
Based on results for spiked samples (blind pairs at one level)
and naturally contaminated samples (blind pairs at 3 levels),
the RSDr ranged from 6 to 35%. The corresponding RSDR
ranged from 11 to 36%. Not surprisingly, the higher RSDr and
RSDR values were found at the lower patulin concentrations:
at 26 ng/g, RSDR values of 33 and 35% were obtained for
clear and cloudy apple juice, respectively; and RSDR values of
36 and 33% for apple puree at patulin contents of 23 and
38 ng/g, respectively. Thus, at patulin contents >50 ng/g, all

RSDr values were <20% and all RSDR values <25%, except
apple puree containing 121 ng/g patulin for which an RSDR of
29% was obtained. Although LC chromatograms showed no
evidence of interferences, any co-extractive compounds from
juices that interfered with the patulin determination would have
been more in evidence at lower patulin levels. There was no obvious correlation between laboratories that found higher than
average patulin levels in the control juices and their results for
positive samples. Thus, Laboratories 6 and 13 found mean levels of 18.8 and 17.2 ng/g, respectively, in the control clear apple
juice, but mean levels of 29.0 and 25.3 ng/g, respectively, in the
lowest naturally contaminated juice as compared with 26 ng/g
patulin, which was the mean result from14 laboratories.
Where it was possible to make direct comparisons e.g., for
clear apple juice, the values for precision characteristics for
this trial compared favorably with those for First Action
AOAC Method (12). Thus, for First Action AOAC Method,
RSDr values of 11 and 13%, and RSDR values of 22 and 21%
were obtained for spiked apple juice containing 50 and
100 ng/g patulin, respectively, 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. Recoveries of 91108% were reported for
the AOAC First Action Method compared with 89% for this
study; the AOAC Method covered a wider range
(20200 ng/g) of patulin spike concentrations.
The principal difference between this study and others was
the application of the collaborative trial to cloudy apple juice
and apple puree, which were previously more difficult matrixes to analyze. The values for recoveries of patulin derived

1394 MACDONALD ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 83, NO. 6, 2000

from the spiked samples ranged from 80 to 92% (Table 2000.02). There were negligible differences in recoveries
for all 3 matrixes, which were acceptable for determining
patulin in all 3 product types.

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. When
outliers were excluded, the HORRAT values for patulin
ranged from 0.5 to 1.3. Because all HORRAT values were
<2.0, which indicates an acceptable precision, and were better
than or comparable to values reported in the AOAC-IUPAC
Official First Action Method (12), this method is clearly acceptable as an AOAC Official Method.
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. The method is a
minor modification of the existing First Action Method for
clear juice, but has an additional step required for matrix extension to cloudy juice and apple puree. This is the first time
validation of a method has been carried out for these matrixes.
Acknowledgments
This project was financially supported by the European
Commission, Standards Measurement and Testing Program
(Brussels, Belgium) SMT Project CT96-2045. There have
been many who contributed to the success of this study,
including A. Boenke (EC SMT-Program Scientific Officer for
the project); K. Mathieson (CSL Food Science Laboratory,
York, UK), who performed the statistical analysis; A. Williams (formerly of Leatherhead Food RA, Leatherhead, UK),
who prepared calibrant standards; H.P. van Egmond (RIVM,
Bilthoven, Netherlands), who took responsibility for the pretrial workshop for participants; and A.E. Buckle (CSL, York,
UK), who assisted at the workshop. The authors also express
their appreciation to the following collaborators for their participation in the study:
Carlo Brera, Istituto Superiore di Sanita, Rome, Italy
Kevin Jrgensen, Danish Veterinary and Food Administration, Srborg, Denmark
Maria Lgia Martins, Laboatorio Nacional De Veterinria,
Lisbon, Portugal
Ma Luz Macho, Laboratrio de Salud Pblica, Bilbao,
Spain
Paul Majerus, Chemiches Untersuchungsamt, Trier, Germany

Lutz Mevissen, Gesellschaft fur Lebensmittel-Forschung

MbH, Berlin, Germany
Jean-Yves Michelet, Scientific Institute of Public
HealthLouis Pasteur, Brussels, Belgium
Kirsti Nuotio, Finnish Customs Laboratory, Espoo, Finland
Alain Pittet, Nestl Research Center, Lausanne, Switzerland
Louis Szymanski, Laboratoire Interrgional de la Repression des Frauds de Paris-Massy, Massy, France
Nicola Tucker, Reading Scientific Services Ltd., Reading,
UK
Remei Viladrich, IRTA, Lleida, Spain
J. Voogt, I.G.B. Keuringsdienst van waren, Rotterdam,
The Netherlands
Anne Wennemar, Chemisches Landes- und Staatliches
Veterinruntersuchungsamt, Mhnster, Germany
References
(1)
(2)
(3)
(4)
(5)

(6)
(7)
(8)

(9)
(10)

(11)
(12)
(13)

(14)
(15)

Scott, P.M. (1974) J. Assoc. Off. Anal. Chem. 57, 621625

Tarter, E., & Scott, P.M. (1991) J. Chromatogr. 538, 441446
Price, K.R. (1979) Biomed. Mass Spectrom. 6, 573574
Moller, T., & Joefsson, E. (1980) J. Assoc. Off. Anal. Chem.
63, 10551056
Bartolome, B., Bengoechea, M.L., Perez-Ilzarbe, F.J.,
Hernandez, T., Estrella, I., & Gomez-Cordoves, C. (1994)
J. Chromatogr. A. 664, 3943
Rovira, R., Ribera, F., Sanchis, V., & Canela, R. (1993)
J. Agric. Food Chem. 41, 214216
Gokmen, V., & Acar, J. (1996) J. Chromatogr. A. 730, 5358
International Standard ISO 8128-2:1993(E) (1993) Apple
Juice, Apple Juice Concentrates and Drinks Containing Apple
JuiceDetermination of Patulin ContentPart 2: Method
using TLC. International Organization for Standardization,
Geneva, Switzerland; and also available from national standardization organizations
Kubacki, S., & Goszcz, H. (1988) Pure Appl. Chem. 60,
871876
International Standard ISO/DIS 8128-1(E) (1993) Apple
Juice, Apple Juice Concentrates and Drinks Containing Apple
JuiceDetermination of Patulin ContentPart 1: Method by
High Performance Liquid Chromatography. International Organization for Standardization, Geneva, Switzerland; and also
available from national standardization organizations
Brause, A.R., Trucksess, M.W., Thomas, F.S., & Page, S.W.,
(1996) J. AOAC Int. 79, 451455
Official Methods of Analysis (1995) 16th Ed., AOAC INTERNATIONAL, Gaithersburg, MD, Method 995.10
FAO (1997) Worldwide Regulations for Mycotoxins 1995A
Compendium, FAO Food and Nutrition Paper No. 64, Rome,
Italy
IUPAC (1995) Pure Appl. Chem. 67, 331343
Horwitz, W., & Albert, R. (1991) J. AOAC Int. 74, 718744