Fresenius leitschri[t [fir
Fresenius Z Anal Chem (1988) 332: 598-601
9 Springer-Verlag1988
Development of mycotoxin reference materials H. P. van Egmond
National Institute of Public Health and Environmental Protection, Laboratory for Residue Analysis, P.O. Box 1, NL-3720 BA Bilthoven, The Netherlands
Entwicklung van Mycotoxin-Referenzmaterialien Summary. The analytical difficulties in mycotoxin determination have given rise to the development of mycotoxin reference materials by the European Community Bureau of Reference (BCR) in cooperation with several European Laboratories, among them the National Institute of Public Health and Environmental Protection. The first project undertaken concerned four milk powders certified for their aflatoxin M1 content (concentrations 0, 0.1, 0.3 and 0.8 ~tg/ kg). Projects that are well-advanced include two peanut meals (aflatoxin B1 contents 0 and ca. 200 ~tg/kg) and two compounded animal feedingstuffs (aflatoxin B1 contents 0 and 10 ~tg/kg). Mycotoxin reference materials are also under development for zearalenone in wheat and maize, and for ochratoxin A in wheat. The general approach followed in the development and certification of mycotoxin reference materials (preparation, testing for homogeneity and stability, intercomparisons of methods and certification exercise) will be discussed and exemplified with the help of the completed project on aflatoxin M1 in milk powder.
contaminated with aflatoxins. Participants in the programme are requested to send in their analysis results to IARC. Reports of the Check Sample Programme continuously show the same unfavourable phenomenon: a large scatter of analysis results (Figs. 2 and 3). The obvious analytical difficulties and the general requirements for Good Laboratory Practice prompted the Community Bureau of Reference to explore the possibilities for the development of
aftatoxin B 1
0
0 afLatox[n B 2
IHO
aflatoxin G 1
aflatoxin G 2
1 Introduction
Mycotoxins are secondary metabolites of fungi which may produce various toxic effects in man and animals. At present the structures of a few hundred mycotoxins are known. The most well known are the aflatoxins, a group of closely related compounds, some of which are very potent carcinogens. Many countries have official regulations for mycotoxins in food and in animal feed, in particular for the naturally occurring aflatoxins Bt, B2, G1, 62, and for aflatoxin M1, the hydroxy derivative of aflatoxin B 1, which occurs in milk and milk products as a consequence of the metabolism of aflatoxin B1 by the cow (Fig. 1). The enforcement of these regulations requires monitoring of suspected commodities. Many laboratories perform large numbers of determinations of mycotoxins, in particular aflatoxins, and consider their results to be reliable. Nevertheless, it often happens that laboratories find quite different analysis results on samples that have been especially homogenised for interlaboratory studies. This is exemplified by the Check Sample Surveys for aflatoxins as organised by the International Agency for Research on Cancer (IARC). This Agency yearly distributes worldwide samples of peanut meal, maize meal and milk powder,
O
O afiatoxin M 1
"--OAO~'OC
H3
Fig. 1. Chemical structures of aflatoxins B1, Bz, G1, G2 and M1 participants 20 Peanut meal-mycatoxin check sample: 198/,
- I ARC
15
s ~~ I--I I'-I O.
5.
10.
nn rq n
15. 20. 25, 30.
F"-I ~'1 , , V'-'I 35, /,0. /*5. 50. 55. 60. 65." Outtiers
Mass fraction aflatoxin B 1 (/ug/kg)
Fig. 2. Frequency distribution of analysis results in IARC Check Sample Programme for aflatoxin B1 in maize (Friesen [1])
599 mycotoxin reference materials, as their availability might greatly improve the reliability of the analytical results. In collaboration with a group of European laboratories experienced in analyses for mycotoxins, the BCR Mycotoxin Programme was undertaken in 1982, with the objective to partici ) a n t s 20 S p r a y - d r i e d m i l k - m y c o t o x i n check sample:lgSL
-IARC
15 10 5
0.0
0.2 0./, 0.6 0.8 1.0
1.2 1./,
1.6 1.8 2.0 2.2 2./~ 2.5
Mass fraction a f l a t o x i n M 1 ( / u g / k g )
Fig. 3. Frequency distribution of analysis results in IARC Check Sample Programme for aflatoxin M~ in milk powder (Friesen [1])
Collection of good quality feedingstuffs
J IWO
J
improve the accuracy and thereby the comparability of mycotoxin measurements. Our laboratory became actively involved in this programme from the beginning. Among the matrices and mycotoxins selected at the onset of the BCR Mycotoxin Programme were peanut butter with aflatoxins B1, B2, GI, G2, peanut meal with aflatoxin B1, compounded animal feedingstuff with aflatoxin B1, milk powder with aflatoxin M1, maize and wheat with deoxynivalenol, maize and wheat with zearalenone, and wheat with ochratoxin A. Priority was given to the development of the aflatoxin reference materials. Within this group, the milk powders with certified contents of aflatoxin M1 (AFM1) got first attention, because there was experience with the preparation of milk powders with certain AFM~ contents and it was thought feasible to take account of analytical, stability, homogeneity and packaging requirements. The general experience gained could then be beneficial to the other BCR mycotoxin projects. The approach followed in the development and certification of mycotoxin reference materials will be outlined hereafter by giving some details of the milk powder project on AFM1, which has been successfully completed (Van Egmond and Wagstaffe [2, 3]) and then, very briefly, the progress with the development of the other mycotoxin reference materials will be mentioned.
Analysis of these feedingstuffs for aflatoxin B I Artificial contamination of feedingstuff briquettes
RIVM
with aflatoxin B 1 up to desired levels
2 Development of milk powder reference materials, certified for AFM1
I Feeding of contaminated feedingstuff briquettes
to
selected groups of dairy cows for 7 days
IWO
Collection of milk at days 6-7
I Spray-drying of milk according to usual Dairy Industry procedures
NIZO
Packing milkpowder in polyethylene
bags
l Filling and sealing into laminate sachets under N2
CCF
atmosphere (30 g of each)
r Labelling
and storage
at
-10 ~
RIVM
Fig. 4. Schematic presentation of the production and control of samples of milk powder, naturally contaminated with aflatoxin M1
Selection of suitable target-concentrations of AFM1 in milk powder was done on the basis of the normal contamination levels of AFM1 found in European milks, the existing legal limits and the 8-fold concentration that occurs during spraydrying of liquid milk. The selected concentrations were: 0, 0.1, 0.3 and 0.8 ~tg/kg. Before preparing these batches a preliminary batch was produced containing AFM1 at a level of ca. 0.9 gg/kg. This batch was prepared to demonstrate that: Milk powder with a desired AFM1 concentration can be obtained. AFM1 is homogeneously distributed both within and between units of a batch. AFM1 and milk powder are stable for several years to allow long term conservation. In addition, this preliminary batch provided material for method intercomparisons between laboratories.
Mass fraction in u g / k g
o N2, 2O~ 9 vacuum, 20~
1.0
_ _ - -.--o--_
0.5
, O
1
)
,
,
3
4
;
,
,
,
,
6
7
8
10
, 11 1 t i m e in m o n t h s
Fig. 5 Stability of aflatoxin M1 in milk powder during 12 months of storage at 20~
600 The production of the material was supervised by our Institute (RIVM) in collaboration with three Dutch Institutions: The Research Institute for Animal Feeding and Nutrition (IVVO). The Netherlands Institute for Dairy Research (NIZO). The Cooperatieve Condensfabriek Friesland (CCF). The procedure adopted is outlined in Fig. 4. After production of the milk powder, laminate sachets under nitrogen and in vacuo were stored at various temperatures over a 12 month period. No evidence of instability was found at any of the storage conditions. The results are exemplified by those recorded at 20~ (Fig. 5). The homogeneity of A F M t in the preliminary batch was confirmed by analysing 10 randomly drawn samples, which yielded a coefficient of variation of 4.1%. This is very low for such a level of contamination (Table 1). Thus it was concluded that it would be feasible and practical to produce milk powders with AFM1 in the described way, and preparations were made to produce the four main batches of milk powder. In the meantime two preliminary intercomparison studies of methods of analysis were carried out to identify and control the principal sources of analytical error, before the certification study would be undertaken. About 20 European laboratories took part. In each case, the laboratories were asked to use the method that they considered most accurate. In the first intercomparison study, participants were asked to : determine AFM1 in a plain CHC13 solution, determine AFM1 in a milk powder (conc. 0.8 ~tg/kg, unknown to participants), use their own aflatoxin M1 calibrant. The results were rather poor (Fig. 6A). There was insufficient agreement both within and between laboratories. It would not be possible to distinguish between samples with AFM1 contents at 0.3 and 0.8 ~tg/kg, which were choosen as two of the target levels for the actual certification. Therefore, a second intercomparison was designed to take
Table 1. Results of homogeneity measurements of AFM ~in preliminary batch of milk powder Mean: 0.64 gg/kg Range: 0.59-0.66 gg/kg
S.D.: C.V.:
0.026gg/kg 4.1%
1.4 1.2 1.0 0.8 05 O.Z, 0.2
12ii
~" 1s-t Intercomporison
0
o~
13. .C_. Z .g
~o s "6
+
1.0 0.8 0.6
0.4
"5 L
0.2
o ~E
1.0 0.8
4-
+
2 n.o_.dIntercomporison C
+4-
4-
4,
*
4-
"t"
06
R.284
0.4 0.2
+
+
4-
+
,+ ++
+
C e r t i f i c a t i o n of RM~ 2 8 4 + 2 8 5
Laboratories
Fig. 6 A - C . Comparison of the results of the preliminary intercomparisons (A and B) of methods for aflatoxin M~ with the certification results (C). Bars represent the confidence interval of each set mean
0.8 0
0
M 1 content
0% o
o
o
o
o
o
o
o
in pg/kg
o
o
o
o
o
o
0
(17
o
o
o
o
oo
o
0.6
o
(14
[]
RM
282
9
RM
283
a
RM
284
o RM
285
0.3 zx A
O.2
~A
n
m
m
m
mare
A
~
9e
oe
9
9
9
9
a
9
9
9
ooee
D 0
100
200
300
zx
n
Fig. 7
a=
0.1 9 e
zx
m m
A
400
500
sample
600
700
s e q u e n c e nr.
9
9
oo
I
al
800
900
oo
9
I [] 1000
9
9
r 1100
Graphical presentation of analytical data from the homogeneity study for aflatoxin M 1 in milk powders RM282, 283, 284 and 285
601 Table 2. Development stages of mycotoxin reference materials Material
Stage Feasibility
Certification
Completed +
Aflatoxin M1 in milk powder (0, 0.09, 0.31, 0.76 gg/kg) Aflatoxin B1 in peanutmeal (0, ca. 200 gg/kg)
+
Aflatoxin B1 in compounded animal feedingstuff (0, ca. 15 gg/kg)
+
Deoxynivalenol in wheat and maize (0, ca. 300 and ca. 500 ~tg/kg resp.)
+
Aflatoxins Bb B2, G~, G2 in peanut butter (sum of aflatoxins: ca. 10 gg/kg) Ochratoxin A in wheat
+a
+
Zearaleone in maize and wheat
+
a The aflatoxin concentrations in peanut butter will probably be determined by a group of laboratories without an official certification exercise account of the principal sources of error found in the first study and to further identify unsolved problems. In the second study, participants were asked to: determine AFM1 in a plain CHC13 solution, determine AFM~ in a CHC13 extract of milk powder, determine AFM1 in a milk powder (conc. 0.8 ~tg/kg, unknown to participants), carry out recovery experiments in parallel with the assay of the milk powder, use a common calibrant. The results of the second study showed some improvement in the interlaboratory agreement (Fig. 6 B). Although in itself it would not allow certification, the second study allowed identification of the most important sources of error and their elimination, before the final certification would be undertaken. At this time, the four main batches of milk powder had been produced and homogeneity studies had been completed. The variability in analysis results was normal for the four contamination levels with the exception of two outlying results (Fig. 7). Analysis of other samples with packing sequence numbers close to the samples in question gave confidence that the outlying results were incidental. It was concluded that the reference materials were acceptably homogeneous with respect to their AFM1 content and the final collaborative certification exercise was undertaken. This exercise was designed in the light of the experience of the preliminary intercomparisons. A rigorous scheme was applied for the certification exercise and particular attention was paid to : The provision of a common and carefully characterized AFM1 calibrant, independence of measurements (8 replicate measurements were required to be made over a period of 3 days and participants used a wide range of methods), the use of recovery experiments, made in parallel with the certification measurements, made by spiking the "blank" RM282.
The results obtained for two of the studied levels (AFM1 target concentrations 0.3 and 0.8 gg/kg) are presented in Fig. 6C. Not all results are presented. Those obtained with TLC with visual estimation did not allow sufficient precision for certification analyses, and therefore their results were not used in the calculation of the certified values. The agreement both between and within laboratories was quite good and had significantly improved as compared to the first and second intercomparison. The results were then further evaluated technically and statistically and final certified AFM1 contents were found to be: RM 2 8 2 : < 0 . 0 5 ~tg/kg, RM 283: 0.09 + 0.04/-0.02 ~tg/kg, RM 284: 0.31 + 0.06 ~tg/kg, RM 285: 0.76 _+ 0.05 gg/kg. The milk powders certified for AFM1 are the first mycotoxin reference materials available on a worldwide basis. Several of BCR's other projects within the Mycotoxin Programme are structured in more or less a similar way as the project on AFM1 in milk powder. Quite important are the intercomparisons of methods of analysis before actual certification is undertaken and the meetings of the Mycotoxin Expert Group in Brussels where open discussions take place to identify and eliminate possible sources of error. Table 2 presents the current state-of-affairs relative to the development of mycotoxin reference materials.
References
1. Friesen M (1984) Personal communication 2. Van Egmond HP, Wagstaffe PJ (1987) J Assoc Off Anal Chem 70:605-610 3. Van Egmond HP, Wagstaffe PJ (1988) J Assoc Off Anal Chem (in press) Received June 9, 1988