Mycotoxins in Food- Aflatoxin in milk

Mycotoxins are toxic products formed by filamentous fungi.  They pose a problem in food if the food has been infected with a fungus which then produced its toxic product.  As the infection can occur at many points in the food production chain, e.g., in the field or in storage, then it is very difficult to decide if a processed food product is contaminated and if it is, where the contamination arose from. In this case the contamination of milk with aflatoxin M₁ was considered.  Here the mycotoxin is ingested by the milk producing animal as aflatoxin B₁, which is then metabolically converted to a hydroxy derivative in the animal, called aflatoxin M₁. Samples of feeds, forage, maize and milk were taken at the dairy farms supplying milk to the dairy whose milk was being evaluated.  Samples of milk from various parts of the dairy were taken in the same time period as the supplying farm samples.  The feeds are to be screened for fungi and mycotoxins and the milk samples for aflatoxin M₁.   The milk from the farms supplying the dairy was positive for aflatoxin M₁ ranging from <1 - 10ppb by the Vicam milk assay and <1 to 1.5ppb by hplc.  Although this does not sound high, it is above regulated levels of several countries.  Work is now continuing to discover the source of the parent aflatoxin, aflatoxin B₁.

1. INTRODUCTION

Agricultural commodities are prone to infection by fungi, either through infection with plant pathogens or by contamination in storage if the material is not properly maintained.  Once this has occurred the fungi may produce toxic secondary metabolites called mycotoxins, which may then enter the food chain ending up in the final feed or food product [1].     Although over 300 mycotoxins are known [2] only a few are known to routinely contaminate commodities, the five major ones being, aflatoxin B₁; deoxynivalenol (DON); fumonisin B_1; ochratoxin A; and zearalenone.   Of these, aflatoxin B₁ (AFB1) is of particular importance, as it has been found in most feeds and foods and is highly carcinogenic [3].   Furthermore, when ingested by ruminants it is converted to aflatoxin M₁ (AFM1) under the influence of the cytochrome P₄₅₀ oxidase system found in rumen micro-flora and the animals own cells [4] (Figure 1).   This is secreted in the milk of lactating animals, which means that milk now contains a carcinogen, which although, less carcinogen than the parent mycotoxin AFB1, poses a serious threat to young animals and children consuming it.

The main chronic effect of both AFB1 and AFM1 is that they can cause liver cancer (hepato-carcinoma) in humans, especially children [5].    This effect is well documented for African rural communities ingesting AFB1 generally through the consumption of groundnuts (peanuts) as a staple [6].   The situation is, however, complicated by the fact that hepatitis B plays a part in this disease but it seems clear that aflatoxin is a prime role player in the promotion of the condition [7].   This is indicated by work showing that estimated intake of AFB1 parallels liver cancer incidence in a selected rural African population. As mention AFM1 is about 8 times less toxic than AFB1 but the carcinogenicity is still there and animal experiments show that it is a potent carcinogen [8].  Because of the concern with regards to AFM1 [9] young children and liver cancer, the permissible levels of AFM1 in milk are stringently legislated [10].   For example the European Union stipulates no more than 0.05ppb (0.05ug/L) in milk for human consumption [11] and the South African level is set at the same level of 0.05ppb [12].  To ensure that lower levels of aflatoxin are consumed by dairy cattle, the maximum level for aflatoxin B₁ in their feed is set at 5ppb in South Africa [13].

The question of where and when mycotoxins enter the food chain is a difficult one to answer, as fungi, their spores being ubiquitous in the environment, can infect and grow at any time on a food matrix, providing conditions are right[14].  This problem has been approached by considering the entry of fungi and mycotoxins along the whole food chain, starting in the field situation with growing crops to what is present in the prepared food on the plate [15].   The phrase from "farm to fork" has been used to popularise this approach, which can be applied as what is termed "Biotracer" [16].  That is tracing the target contaminant or component in commodities in the food chain to see where contamination enters the chain, as indicated in Figure 2 and how it can thus be controlled (e.g., as AFM1 in milk).  This approach is more difficult to pursue than it sounds, as feed components such as cereal grain may amount to many thousands of tons both in the field and in storage.  This raises the difficulties of sampling and actually "tracing" corresponding parts of the cargo through the chain.  One commodity that does not present quite the same difficulties as solid commodities is that of milk.  Here raw milk is pooled on the farm and then transported to the dairy where it is further bulked up.  Because it is a liquid it can be assumed that thorough mixing of the commodity takes place in the milk tankers and holding tanks in the dairy.   

Thus for milk being sampled in a biotracer type study, it is possible to take representative samples of the material from the milking parlour to milk on the supermarket shelf and then into the household or restaurant (Figure 2) provided one can be confident that the same batch is analysed through the whole chain.   Smaller dairies have more chance of being examined for a contaminant in the biotracer approach than a larger one with big through puts of material, as there are smaller volumes of milk to mix. However, it is probable that during a days output there may be variation in the stream so it is important to relate a group of farms out put with holding tank at the dairy and subsequent processed output.  At the retail end it is difficult to obtain milk off the shelf relating to that examined from the dairy, unless the dairy is prepared to divulge where the final product is to be sent to and the various identification codes used are divulged.  It is not unreasonable to assume that the milk in the retail store is the same as that in the dairy packing output for AFM1 contamination and hence taking this as the final level of contamination in milk sold to the public but of course  this might not be the case for other bio-traceable targets such as bacteria.

In addition to milk itself the material being fed to the dairy cattle, e.g., commercial compound feed, may be analysed for mycotoxins, in particular AFB1 [17] which is the most likely entry point of aflatoxin in the case of milk.    This can be taken a stage further back, if the farmer produces their own feed commodities such as cereals or silage these can be analysed in the field or point of production as shown in Figure 2 for signs of contamination.

The purpose of this study is to study milk entering a small South African dairy for contamination with AFM1 and to extend this the final product, the producing farms and the feed being given to the dairy cattle for AFB1.    The aims are several fold: to check the levels of AFM1 in South African milk over a year for the presence of AFM1 to check quality and  if there is seasonal variation in this; to find the point of entry of any AFM1, if found down to at least the farm level;  to biotrace the AFM1 and see what happens to it during the milk production process; to assist farmers in reducing any contamination and advise the dairy as to what may be necessary to quality control their product with respect to AFM1 contamination.    This will be achieved by sampling feed materials and milk at all stages in the best statistical manner and analysing them for aflatoxin using standard high performance liquid chromatography (HPLC) methods and a Vicam immuno-affinity/ fluorimetry method for comparison purposes.  In this presentation only the results for AFM1 in milk, as means of validating the study, will be reported.

Ten farms supplying as common dairy were selected for sampling plus one in a different area for comparison purposes.   Samples of silage and feed (500g) were randomly taken in triplicate the day fed prior to the milk sampling.  These were sealed in air tight bags, transported in a cold cooler box to the laboratory where they were stored prior to analysis in the refrigerator at 2^oC, which was carried out at the first opportunity.   Milk samples (10ml) were taken from farm bulk tanks, the tanker at the dairy and at various points along the processing chain including the final output line and selected stores in the surrounding area.  These were frozen, taken to the laboratory and held in a deep freeze until analysed.  The samples from farm 4 were not analysed for technical reasons.

The AFM1 analysis was done by the method described by the VICAM method for AFM1 in milk according to the manufacturer's instructions (VICAM VI series 4, Vicam, Waterton U.S.A.) using immuno-affinity column chromatography clean-up followed by fluorimetry. 

The clean up for confirmatory analysis of AFM1 was based on that according to the method of Manetta et al. 2005 [18] by HPLC.  In brief: a sample of milk was homogenised and centrifuged at 3000xg for 10min.   A sub-sample (10ml) of the aqueous phase was diluted with an equal volume of purified water and the AFM1 was purified from this by passing through a SPE-C₁₈ cartridge with wash washing and elution with dichloromethane/acetone (95/5v/v).  The eluate was dried under a stream of nitrogen and the residue dissolve in acetonitrile to a standard volume.

The high performance liquid chromatography (HPLC) [19] method, applied to both methods of clean-up, i.e., VICAM and SPE cartridge, in brief, was as follows: 50μl of sample or standard was injected into a Shimadzu liquid chromatography (LC 20AB) equipped with a Gemini C_18­  column (4.6x250mm, 5μm) and guard column (Waters, Milford USA) .  The AFM1 was separated using an isocratic elution solvent composed of water/methanol (55:45v/v) at a flow rate of 1ml/min over 30min.   The AFM1 was passed through a Kobra derivatization cell [19] and detected by a wavelength fluorescence detector set at 353 and 432nm. 

A standard curve was set up by injecting various concentrations of AFM1 standard (Sigma Co.) solution with three separate injections for each concentration.  Recoveries of AFM1 from milk were done by spiking blank milk samples with various levels of AFM1 and passing through the analytical systems.

The main purpose of this part of the overall study was to evaluate the quality of milk being produced on farms in South Africa in terms of AFM1 contamination in order that biotracer work on the mycotoxin could be carried out and to develop suitable methods of analysis.  Without the presence of any mycotoxin any such study would be at best hypothetical.  Table 1 shows that AFM1 was present in farm samples.  The levels range from 0 - 0.8ppb (µg/litre) by HPLC, which, although low, exceeded in some cases the European Union [11] allowable levels of 0.05ppb but not others, e.g., U.S. Food and Drug Administration [20].    This is, however, sufficient for further studies, including biotracer work.  It was of interest to see that certain farms at certain points were producing the higher levels.  The reason for this has to be explored by looking at feed material for AFB1.  Of further interest was the higher levels obtained by the VICAM fluorimetry levels (Table 1). As HPLC is the more definitive method [21] it would seem that the VICAM method has some interference factor within the milk leading to higher readings.  This is being investigated further.   The total volume of milk, obtained from the farms supplied to the dairy was 97992 litres.   If this was bulked separately at the dairy, the AFM1 concentration, as calculated from the levels of AMF1 in individual batches, would average out to 0.12ppb, which compares to similar levels found in bulked milk at the dairy.   (A set of 42 samples were randomly taken at various stages in the dairy operation; the AFM1 level ranged from 0.01 - 0.52ppb with a mean of 0.1ppb.) 

It would be possible to calculate the contribution made to the daily AFM1 output load by the selected farms in terms of total mg AFM1 contributed to the total milk bulk, which contained 12.0mg AFM1.  This is of some use in the quality monitoring of milk for if the levels of individual milk samples or sub-bulks beginning to exceed the daily limit of AMF1 found in the total then remedial steps will have to be taken to lower the level back to that, which is the usual norm, especially if the levels are exceeding legislatory limits. Clearly the contribution by any farm has to be weighted as the amount of AFM1 contributed not only depends upon the concentration of the mycotoxin in the milk but also the volume provided.  Farms providing large volumes of lower concentration of AFM1 will contribute the least mycotoxin to the final bulk and hence are preferable.   


TABLE 1: RESULTS OF WINTER MILK SAMPLING MAY 2010




In a full biotracer investigation it would be desirable to bulk the milk separately at the dairy and analyse this which would give the effects of transporting and any other procedures carried out at this level.  As there are practical difficulties in making this approach it is an example of the difficulties of biotracing specific batches of product.

Random samples of milk from retail outlets were sampled in order to check the quality of the milk output of various dairies under 13 brand names cover several types of milk (The results of the analysis are given in Table 2, most of which were less than 1ppb of AFM1, and, although above the regulation of 0.05ppb (12) are not a cause for alarm, especially as all the results given in this study are preliminary and require further investigation using the more sensitive Kobra cell detector. 


TABLE 2: FIRST SURVEILLANCE STUDY RESULTS, RETAIL MILK

4.CONCLUSION

From these preliminary results it is reasonable to assume that the milk emanating from the dairy being studies is safe in terms of AFM1 contamination.   Further studies will show where the source of any contamination is and what effects the dairy processing will have on the levels.  The results indicate that as most of the milk supplied is below any legislated levels, the effect of bulking up will dilute any supplies that have higher than usual amounts of the mycotoxin.

5. ACKNOWLEDGEMENTS

Thanks are due to the collaborating dairy and farmers who provided samples of milk, their time and advice with which this study could not have been done.


6. REFERENCES

 This paper is part of the lecture given at the Test & Measurement Conference & Workshop, Drakensburg Conference Centre in South Africa in  Novemeber 8 - 10^th,  2010 and the full article will be publish later.