SIST-TP CEN/TR 15298:2006

SLOVENSKI STANDARD
SIST-TP CEN/TR 15298:2006
01-september-2006
äLYLOD±0OHWMHY]RUFHY]DDQDOL]RPLNRWRNVLQRY±3ULPHUMDYDPHGVXKLPPOHWMHPLQ
PHãDQMHPRESULVRWQRVWLWHNRþLQH
Foodstuffs - Sample comminution for mycotoxins analysis - Comparison between dry
milling and slurry mixing
Lebensmittel - Probenvorbereitung für die Mycotoxinanalytik - Vergleich zwischen
Trockenvermahlung und Aufschlämmung
Produits alimentaires - Préparation d'échantillons gros volume pour l'analyse des
mycotoxines - Comparaison entre broyage a sec et broyage par voie humide
Ta slovenski standard je istoveten z: CEN/TR 15298:2006
ICS:
67.050
SIST-TP CEN/TR 15298:2006 en
2003-01.Slovenski inštitut za standardizacijo. Razmnoževanje celote ali delov tega standarda ni dovoljeno.

---------------------- Page: 1 ----------------------

TECHNICAL REPORT
CEN/TR 15298
RAPPORT TECHNIQUE
TECHNISCHER BERICHT
May 2006
ICS 67.050

English Version
Foodstuffs - Sample comminution for mycotoxins analysis -
Comparison between dry milling and slurry mixing
Produits alimentaires - Préparation d'échantillons gros Lebensmittel - Probenvorbereitung für die
volume pour l'analyse des mycotoxines - Comparaison Mycotoxinanalytik - Vergleich zwischen Trockenvermahlung
entre broyage à sec et broyage par voie humide und Aufschlämmung
This Technical Report was approved by CEN on 30 November 2005. It has been drawn up by the Technical Committee CEN/TC 275.
CEN members are the national standards bodies of Austria, Belgium, Cyprus, Czech Republic, Denmark, Estonia, Finland, France,
Germany, Greece, Hungary, Iceland, Ireland, Italy, Latvia, Lithuania, Luxembourg, Malta, Netherlands, Norway, Poland, Portugal, Romania,
Slovakia, Slovenia, Spain, Sweden, Switzerland and United Kingdom.
EUROPEAN COMMITTEE FOR STANDARDIZATION
COMITÉ EUROPÉEN DE NORMALISATION
EUROPÄISCHES KOMITEE FÜR NORMUNG
Management Centre: rue de Stassart, 36  B-1050 Brussels
© 2006 CEN All rights of exploitation in any form and by any means reserved Ref. No. CEN/TR 15298:2006: E
worldwide for CEN national Members.

---------------------- Page: 2 ----------------------

CEN/TR 15298:2006 (E)
Contents Page
Foreword.3
Introduction.4
1 Scope .5
2 Test methods.5
3 Results and discussion.6
4 Acknowledgements . 17
Bibliography. 18

2

---------------------- Page: 3 ----------------------

CEN/TR 15298:2006 (E)
Foreword
This Technical Report (CEN/TR 15298:2006) has been prepared by Technical Committee CEN/TC 275 “Food
analysis - Horizontal method”, the secretariat of which is held by DIN.
Attention is drawn to the possibility that some of the elements of this document may be the subject of patent
rights. CEN [and/or CENELEC] shall not be held responsible for identifying any or all such patent rights.

3

---------------------- Page: 4 ----------------------

CEN/TR 15298:2006 (E)
Introduction
Since 1999-01-01, EC directives for aflatoxins entered into force, which consisted of sampling plans resulting
in sample weights of up to 30 kg. This raised questions on how these relatively big samples could fulfil the
requirement to “finely grind and mix thoroughly each laboratory sample using a process that has been
demonstrated to achieve complete homogenisation” [1]. Since the analytical sample is taken out of this big
sample, the critical step is to take a representative increment out of it. As such this topic has been subject of
several studies in the past. Dickens and Satterwhite [2] developed a mill that could handle up to 25 kg peanut
samples. They presented results of tests on 5 kg samples from which they withdrew 50 g sub-samples, but
gave no data on larger samples. Velasco and Morris [3] considered use of a water slurry to obtain finer
particles and a more uniform particle distribution. Another advantage of slurry preparation is the avoidance of
clogging of samples that have high oil content. They presented experiments on different matrices with sample
weights up to 4,5 kg, whereas they mentioned that slurry preparation is limited only by the capacity of the
equipment. Whitaker et al. [4] considered a compromise. They prepared a slurry from a sample, which was
first comminuted by another milling process. Due to the regulations of the USDA they limited themselves to an
amount of only 1 100 g. Nevertheless this restriction in their method was developed into the alternative best
foods method used for aflatoxin in peanuts [5]. Dorner and Cole [6] started all over again from the beginning:
the 218 kg sample of raw, shelled peanuts for analysis in official USDA approved laboratories. They compared
variability by grinding with four different mills, but only with sub-sample sizes up to 4 kg. So the question how
the result would be on 21,8 kg samples remained unanswered. Their statistical data, especially CV values, on
the 2 kg and 4 kg sub-samples were less favourable than the ones that can be achieved by applying the slurry
method. Scholten and Spanjer [7] published data on slurry preparation for samples up to 10 kg, whereas the
laboratory of Wiertz, Eggert and Jörissen had similar experiences, even when applying samples up to 30 kg.
Data of the latter are compiled in this report. Worldwide however, sub-sampling mills are in favour because
they are easy to apply and fast in comminuting samples into analytical portions. Calori-Domingues et al. [8]
th
demonstrated this with a poster presentation at the X International IUPAC symposium on mycotoxins and
phycotoxins in May 2000. They tested variability for aflatoxin analysis in peanuts associated with sample
preparation by dry milling with a RAS mill. Unfortunately however they only investigated samples up to 5 kg.
So the labs of the Inspectorate for Health Protection, a delivery unit of the Dutch Food and non-food Authority,
and of Wiertz, Eggert and Jörissen, a member of the Eurofins Scientific group, decided to perform new
experiments with following goals: 1. what CV values are achieved when milling 10 kg samples, and 2. are
correct aflatoxin values measured while doing so? The choice of matrices has been discussed at a
CEN/TC 275/WG 5 (Comité Européen de Normalisation, Technical Committee 275, Working Group 5,
Biotoxins) meeting, considering existing and upcoming legislation for different mycotoxins and food types.
Combining both items lead to the conclusion that a lot of matrices, existing as dried, whole or ground raw
material are to be considered. Also differences in sample weight, i.e. between nuts and spices, exist.
Suggestions for representative commodities were:
 cereals, since for this staple food directives exist on as well as aflatoxins, as ochratoxin A and as DON;
 raisins, because these are included in directives for aflatoxins and ochratoxin A;
 paprika powder as an example of a ground commodity.
In practice however it turned out that the availability of naturally contaminated lots that could be used for these
experiments was the limiting factor. The presented results show what exactly has been examined. After these
experiments the detailed work of Schatzki and Toyofuku [9], who measured particle size distributions on
nd
pistachio slurries, became available. This lead to a joint presentation at the 2 World Mycotoxin Forum,
February 2003, in The Netherlands [10]. This report is a combined outline of both investigations.
4

---------------------- Page: 5 ----------------------

CEN/TR 15298:2006 (E)
1 Scope
A comparison was made between dry milling and slurry mixing as comminution step preceding mycotoxins
analysis. Such in respect to EC legislation that consists of sample schemes up to 30 kg. Cacao, green coffee,
almonds and pistachio samples of 10 kg were milled by a RAS mill and all three sub-samples were completely
analysed for aflatoxin B or Ochratoxin A. The differences in analytical results are explained by measurements
1
of particle size distributions of both milling types. The obtained data are compared with literature data on
coefficients of variation (CV) for various milling procedures. For dry milling CV values were generally not
below 20 % for aflatoxin B levels up to 38 µg/kg in peanuts, whereas slurry mixing could achieve CV values
1
below 5 % at aflatoxin B levels down to 4 µg/kg in pistachios. Measurements also showed possible difference
1
in mycotoxin content of a sample between both milling types. This could lead to false positive or negative
results when rejecting or accepting a lot, as this is based on the sample result. It was concluded that slurries
contain smaller particles than dry milled samples and thus generate the lowest possible CV values which in
turn leads to better sample homogenisation.
2 Test methods
2.1 Apparatus
)
1
2.1.1 Slurry mixer, Slurry mixer - Silverson type EX mixer ® ;
1)
2.1.2 RAS mill, Romer Analytical Sampling mill ®
Other laboratory equipment and slurry preparation procedures as described before (see [7] and [9]). The RAS
mill was applied according to the manual (Release 2, January 1998) of the supplier. Before the dry milling
process the pistachio samples were frozen overnight at minus 20 °C.
2.2 Reagents and materials
Aflatoxin measurements were performed as described in EN 14123. Ochratoxin measurements were carried
out in cacao and in green coffee beans as described in EN 14132, including quality control. The only
difference is that fluorescence detection for ochratoxin A is carried out as published by Zimmerli and Dick [11].
2.3 Procedure
For each commodity, experiments were carried out by the following procedure:
1. sampling according to the EC directive, resulting in 10 kg sample;
2. milling the 10 kg sample by a Romer mill with a split ratio of 10 %;
3. taking a dry sample out of the 10 % part as usual for Romer mill users (sub-sample A);
4. slurry mixing of the remaining part of the 10 % part of the sample (sub-sample B);
5. slurry preparation of the 90 % part by Silverson mixing (sub-sample C);
6. analysing the three sub-samples A, B and C by HPLC methods.

1
Silverson type EX mixer is the trade name of a product supplied by Silverson Machines Ltd., Waterside, Chesham,
Bucks, England. Romer Analytical Sampling (RAS) mill is the trade name of a product supplied by Coring-System
Diagnostic GmbH, Robert-Bunsen-Straβe 4, D-64579 Gernsheim, Germany. This information is given for the convenience
of the users of this Technical Report and does not constitute an endorsement by CEN of the product named. Equivalent
products may be used if they can be shown to lead to the same results.
5

---------------------- Page: 6 ----------------------

CEN/TR 15298:2006 (E)
Doing so the complete mycotoxins content in the sample can be reconstructed afterwards by calculation.
3 Results and discussion
The results of all experimental values are given in the second, third and fourth column of Table 1. They
consist of measurements of ochratoxin A in cacao and green coffee beans and of aflatoxins, of which only
aflatoxin B is useful for this purpose, in almonds, pistachios and a sample of mixed spices. All other columns
1
in Table 1 are filled with figures that are calculated from these data. From the weight of each sub-sample and
its mycotoxins content, it is possible to calculate what the mycotoxins content would have been in the total
sample if it had been measured in one sample as a whole. This calculated value is presented in the column
“sample value” in the first row, such as to facilitate several comparisons that will be made in the clause results
and discussion. In the last three columns the mathematical mean of the A, B and C sub-sample results, the
standard deviation and the coefficient of variation of these three measurements are given, which will be
discussed later as well.

Considering the results have to be done from the starting point of the experiments: milling the 10 kg sample by
a Romer mill, which creates a division of the original sample in two sub-samples of different weight. When
RAS milling is used in daily routine analysis this step is followed by taking an incremental sample out of the
smallest sub-sample for further clean up and chemical analysis. This situation is comparable with the results
for sub-sample A in this experiment with the crucial difference that data as presented for sub-samples B and C
are never measured in daily practice. In case of sample preparation by means of slurry, the whole sample is
dealt with. A portion of the slurry is taken for further analysis. Regarding the methods in detail reveals that it
will never be possible to do an experiment by applying both preparations towards one sample. Therefore the
best estimate of a measurement of these samples, as if they were handled by preparing a slurry, can only be
made by calculating the amount of mycotoxins from the individual A, B and C sub-sample values. This
calculated value is presented as “sample value” in Table 1.
6

---------------------- Page: 7 ----------------------

CEN/TR 15298:2006 (E)

Table 1 — Results of sampling, milling and mixing experiments as described in 2.3
Ochratoxin A Sample Sub Sub Sub A,B,C
Value A B C Mean STD CV
a
Matrix (µg/kg) (µg/kg) (µg/kg) (µg/kg) (µg/kg) (µg/kg) (%)
Cacao 0,4 0,5 0,4 0,4 0,4 0,1 13,3
Cacao 0,6 0,4 0,5 0,6 0,5 0,1 20,0
Cacao 1,0 0,9 1,7 0,9 1,2 0,5 39,6
Cacao 1,1 0,8 0,4 1,2 0,8 0,4 50,0
Cacao 1,2 1,5 0,7 1,2 1,1 0,4 35,7
Cacao 1,2 2,6 1,5 1,2 1,8 0,7 41,7
Cacao 1,7 1,1 3 1,6 1,9 1,0 51,8
Cacao 1,7 1,5 1,5 1,7 1,6 0,1 7,4
Cacao 2,2 0,8 2,1 2,2 1,7 0,8 45,9
Cacao 3,5 5,2 1,5 3,7 3,5 1,9 53,7
Cacao 11,9 1,3 1,8 13 5,4 6,6 123,3
Green coffee 1,5 8,1 0,4 1,6 3,4 4,2 123,2
Green coffee 1,9 1,8 2,3 1,8 2,0 0,3 13,4
Green coffee 2,0 2,7 2,6 2,0 2,4 0,4 16,1
Green coffee 2,0 1,5 2,0 2,0 1,8 0,3 14,4

Aflatoxin B Sample Sub Sub Sub A,B,C
1
Value A B C Mean STD CV
Matrix (µg/kg) (µg/kg) (µg/kg) (µg/kg) (µg/kg) (µg/kg) (%)
Almonds 2,0 1,0 0,2 2,2 1,1 1,0 88,8
Almonds 2,4 1,0 4,2 2,2 2,5 1,6 65,5
Almonds 3,1 0 0 3,4 1,1 2,0 173,2
Almonds 4,1 0,5 6,7 3,8 3,7 3,1 84,6
Mixed spices 7,8 4,2 8,1 7,75 6,7 2,2 32,6
Pistachio in shell 33,8 88,2 38 33 53,1 30,5 57,5
Pistachio in shell 44,1 51,4 42,4 44,2 46,0 4,8 10,4
Pistachio kernels 114,1 250 108 114 157,3 80,3 51,0
Pistachio kernels 126,0 204 122 126 150,7 46,2 30,7
a
Measurements carried out according to the schedule as mentioned above, by Kastrup, WEJ Hamburg,
Germany and Scholten, Inspectorate for Health Protection, Amsterdam, the Netherlands.

For the enforcement of a directive the analytical results are important at the point of accepting or rejecting a lot.
Aflatoxin B is regulated in EC directives: 2 µg/kg for nuts and 5 µg/kg for spices. For ochratoxin A only values
1
from a working document [12] can be used: 2 µg/kg for cacao and 3 µg/kg for coffee beans. The latter values
are under discussion and are only used in this report to evaluate the presented measurements. With these
figures, without adding measurement uncertainties, the differences between judgements of a lot based on dry
milling (sub-sample A data) are compared with the data that would have been obtained after slurry preparation
7

---------------------- Page: 8 ----------------------

CEN/TR 15298:2006 (E)
of the sample as a whole (sample value data). Doing so for cacao 2 out of 11 lots would be rejected after a dry
milling procedure and 3 out of these 11 lots after slurry preparation. Only in 1 of these cases the lot would be
rejected by both procedures. In 2 out of 3 cases the dry milling procedure would accept the lot that is rejected
according to the sample value. In 1 case the dry milling procedure would reject a lot, which is accepted by the
slurry preparation method. The latter happens also with 1 of the 4 coffee samples. The other 3 commodities
are judged likewise for both methods. The results for the measurements on aflatoxin B are worse. In 5 out of
1
9 cases the dry milling procedure would lead to acceptance of the lot, whereas the slurry preparation would
reject 8 out of 9. A striking detail in this respect is the fact that in all 5 cases the dry milling leads to
acceptance of a lot, this happens at low levels, i.e. around the limit of the directive. The aflatoxin levels in the
pistachios are so high that the measurements lead to rejection in any case. If the overall results of Table 1
were considered the dry milling procedure would reject 7 out of 24 lots, whereas the preparation of slurry
would reject 11 out of the same 24 lots. Table 2 gives an overview. It also reveals that dry milling lead to 2
false positive results, one with cacao and one with coffee beans. This is an interesting detail, since both
commodities are rather expensive, so from this point of view even false positive results are not desirable.
Table 2 — Overview on rejection of lots and false positive and negative decisions
Mycotoxin Ochratoxin A Aflatoxin Both
Sample n = 15 9 24
Rejected by Dry 3 4 7
Slurry 3 8 11
Both 1 4 5
False negative 2 4 6
False positive 2 0 2

Apart from compliance of samples to a directive, another point can be learned from Table 1. The data of
sub-samples A, B and C reveal that the dry milling process resulted in different mycotoxins content in both
samples in which the sample is divided by dry milling. Due to applying slurry preparation to sub-samples B
and C, all 3 sub-samples have been analysed exactly. From Table 1 the differences can be seen easily by
comparing the columns of A versus B and A versus C. The data on B and C are not available when dry milling
is applied in daily routine analysis. For this investigation these values were measured to be able to
reconstitute the “sample value”. But since these data are available it is also possible to calculate the standard
deviation and the coefficient of variation of the dry milling process on these 10 kg samples. The results are
given in the last 3 columns of Table 1. They show that CV values are only once less than 10 % and can be
more than 50 %, whether the mycotoxins level is low (< 10 µg/kg) or high (> 100 µg/kg). In this particular case
it has to be kept in mind that on a mathematical-statistical basis these values are not of very much significant
value. They are based on only three data per calculation and these values are on their turn originating from
three different types of processing. So the CV data that are added to Table 1 are just given because they are
the only way to express some CV level for the experiments that are performed.
These CV values can be compared with several sets from literature on milling experiments. To be accurate,
like in the cited references, these CV values are composed of sub-sampling and analytical variance [13]. If we
assume the analytical error to be far less than any other error in any mycotoxins study, we can focus on the
published CV’s as being caused by sub-sampling. In all cases we neglect the variance of sampling error. We
only focus on the sub-sampling error in all studies. In chronological order we start with the results obtained
with the sub-sampling mill of Dickens and Satterwhite for peanut kernels [2]. Their data are given in Table 3.
They show that CV varies from 9 % to 43 % in 5 kg samples with aflatoxin B content of 15 µg/kg to 233 µg/kg.
1
It is remarkable that CV is not decreasing when the aflatoxins content increases, as should be expected. Their
data on 500 g samples show that 11 out of 18 have CV values below 10 %, also without correlation between
CV and aflatoxin B content.
1
8

---------------------- Page: 9 ----------------------

CEN/TR 15298:2006 (E)
Table 3 — Dickens mill testing 50 g out of 5 kg (n=4) and 25 g out of 500 g (n=2) peanuts
Sub 1Sub 2Sub 3 Sub 4Mean STD CV Sub 1Sub 2 Mean STD CV
(µg/kg) (µg/kg) (µg/kg) (µg/kg) (µg/kg) (µg/kg) (%) (µg/kg) (µg/kg) (µg/kg) (µg/kg) (%)
14 14 17 14 14,81,510,27,27,47,3 0,1 1,9
17 17 17 14 16,31,59,27,47,37,4 0,1 1,0
51 40 51 40 45,56,414,015,924,320,1 5,9 29,6
57 40 40 51 47,08,418,018,218,218,2 0,0 0,0
57 69 127 90 85,830,735,821,729,625,7 5,6 21,8
70 63 63 113 77,324,131,122,617 19,8 4,0 20,0
257 114 171 129 167,864,238,322,824,823,8 1,4 5,9
257 257 171 103 197,074,637,923,715,719,7 5,7 28,7
257 257 228 343 271,349,718,333,835,434,6 1,1 3,3
343 257 228 103 232,899,342,737,340,138,7 2,0 5,1
   39,239,239,2 0,0 0,0
   44,949,247,1 3,0 6,5
   57,944,551,2 9,5 18,5
   58,189 73,6 21,8 29,7
   64,865,965,4 0,8 1,2
   75,331,953,6 30,7 57,3
   89 90 89,5 0,7 0,8

...