REVIEW
Impact of food processing and detoxification treatments
on mycotoxin contamination
Petr Karlovsky
1
& Michele Suman
2
& Franz Berthiller
3
& Johan De Meester
4
&
Gerhard Eisenbrand
5
& Irène Perrin
6
& Isabelle P. Oswald
7,8
& Gerrit Speijers
9
&
Alessandro Chiodini
10
& Tobias Recker
10
& Pierre Dussort
10
Received: 14 April 2016 /Revised: 29 July 2016 /Accepted: 5 August 2016 / Published online: 23 August 2016
#
Abstract Mycotoxins are fungal metabolites commonly oc-
curring in food, which pose a health risk to the consumer.
Maximum levels for major mycotoxins allowed in food have
been established worldwide. Good agricultural practices, plant
disease management, and adequate storage conditions limit
mycotoxin levels in the food chain yet do not eliminate my-
cotoxins completely. Food processing can further reduce my-
cotoxin levels by physical removal and decontamination by
chemical or enzymatic transformation of mycotoxins into less
toxic products. Physical removal of mycotoxins is very effi-
cient: manual sorting of grains, nuts, and fruits by farmers as
well as automatic sorting by the industry significantly lowers
the mean mycotoxin content. Further processing such as mill-
ing, steeping, and extrusion can also reduce mycotoxin con-
tent. Mycotoxins can be detoxified chemically by reacting
with food components and technical aids; these reactions are
facilitated by high temperature and alkaline or acidic condi-
tions. Detoxification of mycotoxins can also be achieved en-
zymatically. Some enzymes able to transform mycotoxins nat-
urally occur in food commodities or are produced during fer-
mentation but more efficient detoxification can be achieved by
deliberate introduction of purified enzymes. We recommend
integrating evaluation of processing technologies for their im-
pact on mycotoxins into risk management. Processing steps
proven to mitigate mycotoxin contamination should be used
whenever necessary. Development of detoxification technol-
ogies for high-risk commodities should be a priority for re-
search. While physical techniques currently offer the most
efficient post-harvest reduction of mycotoxin content in food,
biotechnology possesses the largest potential for fu ture
developments.
Keywords Mitigation
.
Natural toxins
.
Physical methods
.
Chemical treatment
.
Biological detoxification
.
Decontamination
Introduction
Toxic secondary metabolites produced by fungi belong to the
most toxic contaminants regularly occurring in a wide range
of food commodities (Bennett and Klich
2003). Most
* Pierre Dussort
publications@ilsieurope.be
1
Molecular Phytopathology and Mycotoxin Research,
Georg-August-University Göttingen, Grisebachstrasse6,
37077 Göttingen, Germany
2
Barilla G. R. F.lli SpA, Advanced Laboratory Research, via Mantova
166, 43122 Parma, Italy
3
Christian Doppler Laboratory for Mycotoxin Metabolism,
Department IFA-Tulln, University of Natural Resources and Life
Sciences, Vienna, Konrad-Lorenz-Straße 20, 3430 Tulln, Austria
4
Cargill R&D Center Europe, Havenstraat 84,
B-1800 Vilvoorde, Belgium
Mycotoxin Res (2016) 32:179–205
DOI 10.1007/s12550-016-0257-7
Society for Mycotoxin Research and Springer-Verlag Berlin Heidelberg 2016. This article is published with open access at Springerlink.com
5
Department of Chemistry, Division of Food Chemistry and
Toxicology, Germany (retired), University of Kaiserslautern,
P.O.Box 3049, 67653 Kaiserslautern, Germany
6
Nestlé Research Center, Vers-chez-les-Blanc, PO Box 44,
1000 Lausanne 26, Switzerland
7
INRA, UMR 1331 ToxAlim, Research Center in Food Toxicology,
180 chemin de T ournefeuille, BP93173, 31027 Toulouse, France
8
Université de Toulouse, INP, UMR1331, Toxalim, Toulouse, France
9
General Health Effects Toxicology Safety Food (GETS),
Winterkoning 7, 34353 RN Nieuwegein, The Netherlands
10
International Life Sciences Institute-ILSI Europe, Avenue E.
Mounier 83, Box 6, 1200 Brussels, Belgium
countries responded t o th is threat by establishing an d
enforcing maximum levels for mycotoxins in food
(European Commission
2006; van Egmond et al. 2007).
Setting maximum levels is based on toxicity assessment and
exposure data but it also takes supply and demand into ac-
count. Raw materials are usually tolerated to have higher con-
tamination levels (except for products intended for direct hu-
man consumption) than finished products. The rationale be-
hind this is a dilut ion effect when formulating with non-
contaminated ingredients in preparation of the final product
as well as of the potential mitigation effects due to processing.
In both cases, the mycotoxin concentration in the finished
product will be lower than in the raw material.
Spoilage and toxin formation can occur already on the field
and during storage of agricultural commodities or processed
food. This article focuses on food, but results obtained on feed
will be considered when they can be used to estimate the
efficiency of mitigation strategies potentially useful for food.
A variety of fungal species mostly from the genera
Aspergillus, Penicillium, Fusarium, Alternaria,orClaviceps
are known to produce mycotoxins. Most important in terms of
toxicity and occurrence are aflatoxins B
1
,B
2
,G
1
,andG
2
(AFB
1
,AFB
2
,AG
1
,AFG
2
); ochratoxin A (OTA); fumonisins
B
1
,B
2
,andB
3
(FB
1
,FB
2
,FB
3
); deoxynivalenol (DON) and
other trichothecenes; zearalenone (ZEN); patulin (PAT); and
ergot alkaloids (EAs), which a re briefly characterized in
Table
1, while their chemical structures are shown in Fig. 1.
Harmful effects of mycotoxin-contaminated food can be
avoided by (i) preventing contamination, (ii) removing contam-
inated material from the food commodity, (iii) mitigating myco-
toxin content in food, and (iv) treating exposed individuals. In
some commodities , only part of the harvest enters the food chain.
Selection of charg es with low mycotoxin levels for consumption
while using the remainder for feed and energy production would
reduce the exposure of consumers to mycotoxins. Unfortunately ,
this is only possible in a few commodities and, even there, pro-
duction systems targe ting food markets, feed manufacturing, and
energy production are often so specialized that they cannot re-
place each other. The first priority therefore remains prevention
of toxin accumulation directly on the field (preharvest) or there-
after (transport and storage) (Kabak et al.
2006; Choudhari and
Kumari
2010). A variety of agricultural practices, e.g., growing
resistant crop varieties, crop rotation, soil tillage, chemical and
biological control of plant diseases, and insect control are avail-
able to minimize mycotoxin production on the field (Edwards
2004; Munkvold 2014; Mesterhazy 2014; Alberts et al. 2016).
Proper harvest and storage conditions are crucial to prevent fun-
gal growth and mycotoxin accumulation in harvested commod-
ities (Jacobsen
2014). Unfortunately , preharvest measures do not
guarantee the absence of mycotoxins in food or feed. Food pro-
cessing can impact mycotoxins in raw material by (i) physical
removal, (ii) chemical transformation which can result in metab-
olites of lower or higher toxicity, (iii) release from masked or
entrapped forms which may increase bioavailability, (iv) enzy-
matic detoxifica tion, and (v) adsorption to solid surfaces.
Physical and chemical mechanisms reducing mycotoxin content
often act together in the same food processing step. For instance,
sulfur dioxide used in corn wet milling to ease the separation of
germs, proteins, and starch possesses potential for chemical de-
toxification. Reduction of mycotoxin contamination was docu-
mented for cleaning; milling; brewing; fermentation; cooking;
baking; frying; roasting; flaking; alkaline cooking;
nixtamalization (soaking, cooking in an alkaline solution, and
hulling of grains); and extrusion. Concentrations of some myco-
toxins can be reduced substantially while others, such as DON,
are relatively resistant to degradation (Milani and Maleki
2014;
Karlovsky
201 1). Detoxification of grain mycotoxins during
food processing has recently been reviewed (Kaushik
2015).
As the last resort, consumers can be prophylactically treated with
binders in areas of chronically high aflatoxin exposure (Afriyie-
Gyawu et al.
2008;Wangetal.2008).
The following terms are used to describe the outcome of
mitigation treatments throughout this article: removal of myco-
toxins from raw materials and/or finished products,
transformation (modification of the chemical structure of the
molecule), detoxification (transformation which reduced the tox-
icity), and decontamination (removal or detoxification/inactiva-
tion). Effective decontamination should be irreversible, modified
forms of mycotoxins should be affected together with parent
compounds, the products should be non-toxic, and the food
should retain its nutritive value and remain palatable (Milani
and Maleki
2014). Processing procedures, agents, and microor-
ganisms must be allowed for use in food (Codex Alimentarius,
2015
). The interested reader is also referred to European
Commission
Regulation 2015/786, defining acceptability criteria
for detoxification processes applied to products intended for an-
imal feed (EC
2015). These criteria may serve as a model for the
assessment of mycotoxin detoxification technologies in food pro-
cessing. Compliance of a given detoxification process with those
criteria will be assessed by the European Food Safety Authority
(EFSA).
In this review, conventional food processing affecting my-
cotoxins as well as processes dedicated to decontamination are
covered. Applications of the techniques to selected commod-
ities are presented for illustration, knowledge gaps are
outlined, and recommendations for prioritizating mitigation
actions and further research are given.
Physical processing methods
Sorting
Unprocessed cereals in bulk trading often contain dust and
admixtures. Broken and damaged k ernels usua lly contain
most of mycotoxin contamination (Johansson et al. 2006)
180 Mycotoxin Res (2016) 32:179–205
though they constitute only 3–6 % of the bulk load (Whitaker
et al. 2003). The first processing of agricultural goods after
harvest often involves sorting, washing, or milling (Grenier
et al.
2014). Figure 2 summarizes the use of these techniques.
Sorting machines based on particle weight and size are in use
since the end of the n ineteenth century (Mayer
1898).
Originally, grains were sorted in bulk using centrifugation
force and flotation in air flow. In the 1960s, optical sorting
Tabl e 1 Major mycotoxins and their producers, affected crops, adverse health effects and guidance values
Mycotoxin Major producing fungi Main affected crops Principal adverse effects Health-based guidance value
(HBGV)
Aflatoxins FB
1
,
FB
2
,FG
1
,FG
2
;
metabolite
AFM
1
in milk
Aspergillus parasiticus,
A. flavus (JECFA
2001a)
Peanuts, nuts, maize, cottonseed,
wheat, barley, cocoa beans, rice,
copra, dried fruits, spices, figs,
crude vegetable oils (IARC
2012;
EFSA
2007;JECFA1999)
Extremely potent toxins and genotoxic
carcinogens (after metabolic
converstion to 8,9-epoxides in the
liver); classified as carcinogenic to
humans, AFM
1
as possibly
carcinogenic to humans (EFSA
2007;IARC2012;JECFA1999,
2001a)
Because of carcinogencity, exposure
should be kept as low as reasonably
achievable. No official HBGV
Ochratoxin A
(OTA)
Aspergillus alutaceus,
Aspergillus carbonarius,
Penicillium verrucosum
(EFSA
2006)
Grain, legumes, oleaginous seeds,
peanuts, cashews, dried fruits,
coffee, wine, grape juice, cocoa,
spices, meat products (JECFA
2001a;EFSA2006)
Nephrotoxic, renal tumors in rodents at
high doses (EFSA
2006,JECFA
2001a,IARC1993); classified as
carcinogenic in experimental
animals and possibly humans (IARC
1993)
PTWI 120 ng/kg BW/day (EFSA
2006)
and 100 ng/kg BW/day (JECFA
2001a)
Fumonisins B
1
,
B
2
,andB
3
(FB1, FB2,
FB3)
Fusarium verticillioides,
F. proliferatum,
Aspergillus niger (EFSA
2005;JECFA2001a,
2012)
Maize (Fusarium spp.), grapes
(A. niger) (EFSA
2005;JECFA
2001a, 2012)
Inhibit sphingolipid biosynthesis;
induction of apoptosis, tumors in
rodents (EFSA
2005;JECFA2001a;
SCF
2003,IARC2002), putative
teratogenicity; FB
1
classified as
possibly carcinogenic to humans
(IARC
2002)
Group PMTDI (JECFA
2001a, 2012)
and group TDI (SCF
2003)2μg/kg
BW/day for FB
1
,FB
2
,andFB
3
alone or in combination
Deoxynivalenol
(DON) and its
acetylated deri-
vates (3- and
15-acetyl-
DON)
F. graminearum,
F. culmorum (EFSA
2004, 2011a;JECFA
2001a, 2011)
Wheat, maize, barley, oats, rye; less
often rice, sorghum and triticale
(EFSA
2004, 2011a;JECFA
2001a, 2011)
Feed refusal, vomiting, and diarrhea;
reduced growth; thymus, spleen,
heart, liver, and immune system
affected at higher doses (EFSA
2004;IARC1993;JECFA2001a;
SCF
2002); not classifiable as to
carcinogenicity to humans, (IARC
1993)
TDI 1 μg/kg BW/day for DON (SCF
2002,EFSA2004); group PMTDI
1 μg/kg BW/day; ARfD 8 μg/kg
BW/day for DON and its acetylated
derivatives (JECF A
2011)
Other
trichothecenes,
e.g., T-2 toxin,
HT-2 toxin,
nivalenol
(NIV)
F. sporotrichioides,
F. langsethiae (JECFA
2001a), F. poae and
F. c er e al i s , F. culmorum
and F. graminearum
(EFSA
2013)
Cereals (EFSA
2011a) Acute effects of T-2 similar to high dose
radiation (diarrhea, hemorrhage,
hematotoxicity, and immune sup-
pression) (JECFA
2001a,EFSA
2011a); toxicological profile of NIV
similar (EFSA
2013); not classifi-
able as to carcinogenicity to humans
(IARC
1993)
Group TDI 0.1 μg/kg BW/day (EFSA
2011a) and group PMTDI 0.06 μg/
kg BW/day (JECFA
2001a
)forT-2
an
d HT-2 toxins combined.
TDI 1.2 μg/kg BW/day for NIV (EFSA
2013)
Zearalenone
(ZEN)
Fusarium spp. (JECFA
2000,EFSA2011b)
Worldwide in all types of grains;
highest levels in maize and wheat
bran (JECFA
2000,EFSA2011b)
ZEN and its metabolites interact with
α-andβ-estrogen receptors and en-
docrine disruptors (JECFA
2000,
EFSA
2011b)
PMTDI 0.5 μg/kg BW/day for ZEN,
recommended that the total intake of
ZEN and its metabolites should not
exceed the PMTDI (JECFA
2000);
TDI 0.25 μg/kg BW/d for ZEN
(EFSA
2011b)
Patulin (PAT) Byssochlamys spp.,
Penicillium spp.,
Asper gillus spp. (IARC
1986;JECFA1996)
Many fruits, strawberries, tomatoes,
olives, and cereals (IARC
1986;
JECFA
1996)
Gastrointestinal ulceration;
immunotoxicity and neurotoxicity in
animals; genotoxic (JECFA
1996);
inadequate evidence of carcinoge-
nicity in animals, not classifiable as
to its carcinogenicity to humans
(IARC
1986)
PMTDI 0.4 μg/kg BW/day (JECFA
1996)
Ergot alkaloids Claviceps spp., in Europe
mostly C. purpurea
(EFSA
2012,BfR2004)
True grasses; most important on
cereals (rye, wheat, triticale,
barley, millet, and oats) (EFSA
2012,BfR2004)
Interact with neurotransmitter
receptors; acute toxicity: convulsive
neurotroxicity, uterine hemorrhage,
and abortions; chronic toxicity:
vasoconstriction with ischemia and
necrosis of extremities (ergotism)
(EFSA
2012,BfR2004
)
V
arious EAs seem to have similar toxic
potency; group ARfD 1 μg/kg BW/
day and group TDI 0.6 μg/kg BW/
day; both apply to the sum of EAs
(EFSA
2012)
PTWI provisional tolerable weekly intake, PMTDI provisional maximum tolerable daily intake, TDI tolerable daily intake, ARfD acute reference dose
(for 1-day exposure)
Mycotoxin Res (2016) 32:179–205 181
was established. The operation principle is to direct streams of
grains along an array of optical sensors. When a grain differ-
ing in color is detected, the detector triggers a magnetic valve
and a jet of pressurized air removes the kernel from the stream
(Fraenkel
1962). This principle is still used today.
Contemporary grain sorters have a throughput of dozens of
tons grain per hour.
Aflatoxin contamination is usually heterogeneous so that
separating damaged kernels can effectively reduce contamina-
tion (Kabak et al.
2006). Grain sorting using UV light illumi-
nation for aflatoxin reduction is common. The observed bright
greenish-yellow fluorescence (BGYF) does not originate from
aflatoxins but from a kojic acid derivative following reaction
with endogenous peroxidase. In dried commodities, peroxi-
dase is inactivated an d the BGYF method does not work.
The quick and easy Bblack light test^ may therefore result in
both false positive and false negative findings (Bothast and
Hesseltine
1975). Although the test is not as reliable as orig-
inally hoped (Doster and Michailides
1998), it is widely used,
e.g., by Turkish companies exporting dry figs and nuts to the
EU. As an audition by the Food and Veterinary Office of the
EU confirmed, the efficiency of sorting is regularly verified by
laboratory analysis (EC
2013).
Distribution of ergot alkaloids (EAs) is even more hetero-
geneous than aflatoxins because intermediate contamination
does not exist at a single-kernel level. Sclerotia loaded with
EAs are efficiently removed from rye by opto-electronic
sorting (Young et al.
1983; Miedaner and Geiger 2015).
Because infection with Fusarium verticillioides often does
not cause symptoms (Munkvold and Desjardins
1988) and
Fig. 1 Chemical structures of
major mycotoxins and
modification due to food
processing. 1 de-epoxidation, 2
acetylation, 3 oxidation, 4
epimerization, 5 deamination, 6
glucosylation, 7 hydrolysis, 8
lactone cleavage (hydrolysis), 9
hydroxylation, 10 peptide
cleavage, 11 sulfonation, 12
reduction, 13 ether cleavage
182 Mycotoxin Res (2016) 32:179–205
correlation between fumonisin content and symptoms is weak
(Afolabi et al. 2007), grain sorting might not reduce fumonisin
content efficiently though successful attempts have been re-
ported (Pearson et al.
2004). Mycotoxins accumulating with-
out visible symptoms pose a limit to optical sorting as a my-
cotoxin mitigation strategy. This may explain why no re-
duction of aflatoxin content by sorting was found in a
recent study (Mutiga et al.
2014).
Sieving cleaning
Removing kernels with extensive mold growth, broken ker-
nels, and fine materials such as dirt and debris can be achieved
by sieve cleaning, which significantly lowers total mycotoxin
contamination. Removal of EAs from wheat grains by sieving
has been used as a plant quarantine treatment (Muthaiyan
2009). After sieving off corn screenings, it was determined
Fig. 2 Summary of physical and
chemical processes applicable to
food commodities in order to
mitigate targeted mycotoxins.
*Conversion to a more estrogenic
cis-form. **Experimentally
demonstrated on apple juice
Mycotoxin Res (2016) 32:179–205 183