Analytical approaches for MCPD esters and glycidyl esters ... · MCPD esters and glycidol esters in vegetable oils and some other foods. The focus is on the analytical methods based
Post on 14-Oct-2020
15 Views
Preview:
Transcript
Food Additives & Contaminants: Part A2012, 1–35, iFirst
Analytical approaches for MCPD esters and glycidyl esters in food and biological
samples: a review and future perspectives
C. Crewsa, A. Chiodinib*, M. Granvoglc, C. Hamletd, K. Hrncirıke, J. Kuhlmannf, A. Lampeng, G. Scholzh,R. Weisshaari, T. Wenzlj, P.R. Jastib and W. Seefelderh
aThe Food and Environment Research Agency (FERA), Sand Hutton, York YO41 1LZ, UK; bILSI Europe a.i.s.b.l., Avenue E.Mounier 83, Box 6, B-1200 Brussels, Belgium; cGerman Research Institute for Food Chemistry (DFA), Lise-Meitner-Straße 34,D-85354 Freising, Germany; dPremier Foods, Lord Rank Centre, Lincoln Road, High Wycombe HP12 3QR, UK; eUnilever,Olivier van Noortlaan 120, NL-3130 AC Vlaardingen, the Netherlands; fSGS Germany GmbH, Weidenbaumsweg 137D,D-21035 Hamburg, Germany; gFederal Institute for Risk Assessment, Thielallee 88–92, D-14195 Berlin, Germany;hNestle Research Centre, Vers-Chez-les-Blanc PO Box 44, CH-1000 Lausanne, Switzerland; iChemisches undVeterinaruntersuchungsamt, Schaflandstraße 3/2 und 3/3, D-70736 Fellbach, Germany; jInstitute for Reference Materials andMeasurements, European Commission, Directorate-General Joint Research Centre, Retieseweg 111, B-2440 Geel, Belgium
(Received 12 April 2012; final version received 8 August 2012)
Esters of 2 - and 3-monochloropropane-1,2-diol (MCPD) and glycidol esters are important contaminantsof processed edible oils used as foods or food ingredients. This review describes the occurrence and analysis ofMCPD esters and glycidol esters in vegetable oils and some other foods. The focus is on the analytical methodsbased on both direct and indirect methods. Methods of analysis applied to oils and lipid extracts of foods havebeen based on transesterification to free MCPD and determination by gas chromatography-mass spectrometry(indirect methods) and by high-performance liquid chromatography-mass spectrometry (direct methods). Theevolution and performance of the different methods is described and their advantages and disadvantages arediscussed. The application of direct and indirect methods to the analysis of foods and to research studies isdescribed. The metabolism and fate of MCPD esters and glycidol esters in biological systems and the methodsused to study these in body tissues studies are described. A clear understanding of the chemistry of the methods isimportant when choosing those suitable for the desired application, and will contribute to the mitigation of thesecontaminants.
Keywords: chloropropanols; MCPD; MCPD esters; glycidol; glycidyl esters; food processing; carcinogens;vegetable oil; deodorisation; LC-MS; food safety; chromatography
Introduction
This review offers a detailed perspective of develop-
ments in analytical methods for fatty acid esters
of chlorinated propanols and glycidol. Chlorinated
propanols, especially 3-monochloropropane-1,2-diol
(3-MCPD), have been known as food contaminants
for over 30 years. Glycidol has long been recognised as
a carcinogen, but has only relatively recently been
shown to be associated with 3-MCPD in foods. Serious
attention has turned to the esters of MCPD and
glycidol in the past few years following their discovery
in refined vegetable oils.A review of MCPD and glycidyl esters is important
on account of the high levels found in relation to their
free forms, making it important to find out more about
their occurrence and possible toxicity, to monitor
mitigation successes, and eventually to ensure the
future compliance with any exposure recommenda-tions. Analytical methods have advanced considerablyin recent years, thus there is considerable benefit inproviding an update of the current status and identi-fying knowledge gaps. This review was produced bymembers of the International Life Sciences InstituteTask Force on Process Related Compounds/NaturalToxins Task Force and Risk Assessment of Chemicalsas part of a joint activity on 3-MCPD esters infood products and is published in association with apaper describing the potential mitigation measures(Craft et al. Forthcoming 2012).
History of MCPD esters and glycidyl esters
A number of chlorinated propanols were discoveredas contaminants of acid-hydrolysed vegetable protein
*Corresponding author. Email: achiodini@ilsieurope.be
ISSN 1944–0049 print/ISSN 1944–0057 online
� 2012 ILSI Europe
http://dx.doi.org/10.1080/19440049.2012.720385
http://www.tandfonline.com
(acid-HVP), used as a flavour enhancer, by Velıseket al. (1979) and Davıdek et al. (1980). Acid-HVP ismanufactured by the hydrochloric acid hydrolysis ofvegetable material (typically the residue from oilseedpressing) carried out under conditions of high temper-ature. The chlorinated products are formed fromthe reaction of the acid with residual lipids. Themajor compound, 3-MCPD, was suspected to be acarcinogen, and its discovery led to the developmentand validation of analytical methods, followed by foodsurveys and toxicological testing. Recommendationsregarding acceptable levels of intake were issued byofficial bodies including the European CommissionScientific Committee on Food (European Commission2001a, 2001b) and the US Food and DrugAdministration (USFDA) (2008). Industry action hasled to a reduction of 3-MCPD in commercial foodproducts such that they now comply with regulationsand contribute to a reduction in exposure.
Esters of 3-MCPD (also referred as ‘‘boundMCPD’’) were also found in acid-HVP and in modelsystems when precursors such as triacylglycerols(TAG) were reacted with hydrochloric acid (Velıseket al. 1979; Davıdek et al. 1980, 1982). They were alsoreported in goat’s milk by Cerbulis et al. (1984), but nofurther effort was made to investigate their occurrenceuntil quite recently, when significant amounts of3-MCPD fatty acid esters were found in foods and inparticular in refined edible oils.
Structures of MCPD esters
MCPDs are chlorinated analogues of glycerol havinga chlorine atom in positions 1 or 2. MCPD esters cantake many forms, being either mono- or di-esters, asshown in Figure 1, of the range of fatty acids found inoils, and they can be expected to mimic the naturalTAG in the composition of their fatty acids. Thechloropropanol portion also exists as racemic mix-tures, and as both 2 - and 3-chloro propyl esters.
Dichloropropanol esters have not been reported inrefined oils and it is unlikely that they are formedduring current manufacturing processes.
Toxicity of MCPD esters
The toxicological significance of 3-MCPD esters andtheir contribution to dietary intake of 3-MCPD is notknown (Bakhiya et al. 2009). Dietary intake and fateare of concern because of the potential contributionto the dietary intake of 3-MCPD from the hydrolysisof 3-MCPD esters in vivo by lipases in the gastrointes-tinal tract. There has not yet been a toxicologicalstudy of 2-MCPD as pure standards have not beenreadily available until recently and because in termsof occurrence the free 2-MCPD is usually less abun-dant. It exceeds 3-MCPD only in acid-HVP that hasundergone hydrolysis to reduce 3-MCPD, a process towhich 2-MCPD is more resistant.
Glycidyl esters
Fatty acid esters of glycidol (3-hydroxy-1,2-epoxypro-pane) (Figure 2) are of concern as glycidol is known tobe carcinogenic in animal studies and is thought to actvia a non-threshold, genotoxic mechanism of action(Bakhiya et al. 2009; Habermeyer et al. 2011). Glycidolhas been classified as probably carcinogenic to humans(Group 2A) by the International Agency for Researchon Cancer (IARC) (2000) and is ‘‘reasonably antici-pated to be a human carcinogen’’ by the US NationalToxicology Program (NTP) (1990, 2007).
The current interest in glycidyl esters arose frominvestigations into the discrepancies between theresults for different methods for the determination of3-MCPD esters observed when the same sampleswere analysed by two methods. It was proposed andconfirmed that overestimation of the levels of 3-MCPDesters was the result of the chlorination by addedsodium chloride of additional compounds identifiedas fatty acid esters of glycidol (Kuhlmann 2008;Weisshaar and Perz 2010).
The Bundesinstitut fur Risikobewertung (BfR –German Federal Institute for Risk Assessment) hasconcluded that infants who are fed exclusively on
Cl
HO
HO
OH
Cl
HO
(a) (b)
Cl
HO
R1O
Cl
R1O
R2O
R2O
Cl
R1O
(c) (d) (e)
Figure 1. Structures of fatty acid esters of 3-MCPD:(a) 3-MCPD; (b) 2-MCPD; (c) 3-MCPD mono-ester;(d) 3-MCPD di-ester; (e) 2-MCPD mono-ester. R1,R2¼ acyl of fatty acid.
Figure 2. Structure of a fatty acid ester of glycidol(R¼ alk(en)yl).
2 C. Crews et al.
commercially prepared infant milk formula could takein harmful levels of glycidol (BfR 2009).
Determination of MCPD esters and glycidyl esters
The separation of fat components, especially acylgly-cerols, has classically been carried out using thin-layerchromatography (TLC), and this technique was thefirst used to separate chloropropanol esters frommodel systems and food lipids (Velısek et al. 1979;Davıdek et al. 1980). The chloropropanol ester frac-tions were extracted from the TLC plates with solventand determined by gas chromatography-mass spectro-metry (GC-MS).
TLC methods combined with silica column chro-matography were used by Gardner et al. (1983) toshow the presence of 3-MCPD esters in adulteratedcooking oils in Spain. The esters were determineddirectly by GC-MS. A similar TLC/silica columnchromatography clean-up technique was applied totarget 3-MCPD esters in goat’s milk by Cerbulis et al.(1984). In this case, acid-catalysed transesterificationwas used to produce fatty acid methyl esters and free3-MCPD, which were both determined by GC-MS.
Hydrolysis methods were used more widely in 2004.Divinova et al. (2004) used acid-catalysed transester-ification (methanolysis) to release free 3-MCPD fromfat extracts of food, and phenylboronic acid (PBA)derivatisation to measure the total and the free3-MCPD (in the whole portion of food) by GC-MS,from which the bound level was calculated by differ-ence. Hamlet and Sadd (2004) used enzymatic cleavageto release free 3-MCPD from bound forms in fatextracts of cereal products. GC-MS with heptafluor-obutyrylimidazole (HFBI) derivatisation was used tomeasure 3-MCPD in both the whole sample and thehydrolysed fat portion.
Following concerns regarding the possible forma-tion of 3-MCPD during acid-catalysed transesterifica-tion, a method based on transesterification underalkaline conditions using sodium methoxide wereproposed (Weisshaar 2008). This method was sub-jected to further study and modification to preventdegradation of 3-MCPD under the alkaline conditionsand to deal with the issue of formation of 3-MCPDfrom the sodium chloride added as a part of theprocedure. Alkaline-catalysed transesterification hasbeen adopted as an official method by the DeutscheGesellschaft fur Fettwissenschaft (DGF – GermanSociety for Fat Research), however acid-based proce-dures are still widely accepted.
Separation of 2- and 3-MCPD mono-esters fromdi-esters is of considerable interest but has so far beenapplied only on a limited scale using silica columnfractionation (Zelinkova et al. 2007). Analysis ofmono- and di-ester fractions can clearly provide more
information of use in toxicological studies, as can thedetermination of individual esters.
Esters of glycidol can be determined indirectly byits conversion to free MCPD. Subtraction of the freeMCPD derived fromMCPD esters from the sum of thefree MCPD released by transesterification of theMCPD esters and the glycidyl esters gives the level ofthe glycidyl ester content. A non-stoichiometric factorreflecting the transformation of glycidol to 3-MCPDmust be applied (DGF 2011a).
Indirect methods can give only a measure of thetotal MCPD ester content of a food sample, or thesum of MCPD ester and glycidyl esters. It might beassumed that direct methods, where there is notransesterification stage, should be both easier andprovide more detailed information, but in practice themethodology has proven challenging to apply andthe large number of potential MCPD esters meansdata analysis can be difficult.
Direct methods were initially based on GC-MSanalysis of fractions isolated by TLC and/or columnchromatography (Reece 2005; Zelinkova et al. 2007),but the use of liquid chromatography-mass spectro-metry (LC-MS) has proved more popular and conve-nient. Procedures have ranged from direct injectionof oil solutions to the incorporation of solid-phaseextraction clean-up. Glycidyl esters and MCPD esterscan be determined simultaneously (Haines et al. 2011).For the independent determination of glycidyl esters,LC-MS is the method of choice, typically with gelpermeation chromatography clean-up (Dubois et al.2011; Granvogl and Schieberle 2011b).
MCPD esters are assumed to be associated with thenaturally occurring acylglycerols in foods, and solventextraction of the non-polar oil fraction of foods is thefirst step in most analytical methods, although somepolar element is preferred to aid extraction of themono-esters. In the early years quantification washampered by a lack of standards, but these are nowavailable from a small number of suppliers.
The direct determination of glycidyl esters is easierthan that of MCPD esters on account of the singleposition for fatty acid esterification and the fact thatthe esters can be separated by gas chromatography(GC) or liquid chromatography-mass spectrometry(LC-MS) after isolation of a non-polar lipid fractionand separation from the acylglycerols fraction. Indirectmethods are linked to the determination of MCPDesters and based on conversion of glycidyl esters to free3-MCPD which is determined in the usual ways.
Analytical methods for monitoring
Indirect methods
The principle of indirect determination of MCPDesters is the conversion of a number of individual
Food Additives & Contaminants: Part A 3
MCPD esters of fatty acids into a single compound,3- or 2-MCPD, that is quantified. To date, severalindirect methods have been developed for the analysisof MCPD esters, the vast majority as 3-MCPD esters.
The analytical protocol of these methods comprisesa uniform series of steps: addition of an internalstandard (either free or esterified form of isotopicallylabelled 3-MCPD) to the sample, transesterification(commonly performed either in acid or alkalinemedium), neutralisation of the reaction mixture andsalting out (using different neutralising reagents andsalts), derivatisation of the cleaved 3-MCPD/2-MCPD,and GC-MS analysis.
Analytical protocol for indirect methods
Since even small variations within each individualanalytical step may have a significant impact on thespecificity, repeatability, reproducibility, truenessand other parameters of the method, the analyticalprotocol is treated in detail in this section.
MCPD cleavage
The indirect determination of bound MCPD in oils orlipid extracts commences with the cleavage of MCPDesters. Free MCPD can be released enzymatically(Hamlet and Sadd 2004), but the majority of themethods employ a chemical route using a transester-ification step. Transesterification in the presence ofmethanol (methanolysis) results in the conversion ofTAGs and partial acylglycerols into fatty acid methylesters (FAMEs) and glycerol. Simultaneously, 3- and2-MCPD esters are converted to free 3- and 2-MCPD,respectively. The transesterification step can be carriedout both in acid (Divinova et al. 2004) and alkaline(Weisshaar 2008) media.
The alkaline-catalysed transesterification approachis convenient owing to its short duration (with reactiontimes up to 10min). The major drawback of thisapproach is that 3-MCPD is unstable in alkalinesolutions giving rise to glycidol. Two independentstudies have shown degradation of 3-MCPD to berapid under these conditions; the 3-MCPD recoveryfound after 1min of transesterification was 83–95%(Hrncirık et al. 2011), 75% after 3min (Kuhlmann2011a), about 50% after 9min (Kuhlmann 2011a), and40–41% after 10min (Hrncirık et al. 2011). It can beexpected that the poor stability of 3-MCPD in alkalinemedia affects the sensitivity of these methods signifi-cantly, thus there may be scope to optimise theprotocol to achieve maximum recovery by minimisingtransesterification time. However, assuming that thedecomposition of 3-MCPD (released from its nativeesters) and its deuterated analogue (used as internalstandard) occurs at the same rate, the transesterifica-tion time should not have an impact on the trueness
of the method. Nevertheless, the levels of bound3-MCPD found at very short transesterification times(1–2min) were surprisingly 10–20% higher than thoseobtained at longer times (5–10min) (Hrncirık et al.2011). The authors attributed this phenomenonto different conversion rates amongst the various3-MCPD forms (e.g. mono-esters being converted to3-MCPD faster than di-esters), and, as a consequence,suggested choosing a deuterated di-ester form of 3-MCPD as internal standard because of the similaritywith the majority of the native 3-MCPD esters.
In order to eliminate the problems related to unde-sirable conversion of MCPD to glycidol, Kuhlmann(2011a) proposed performing the alkaline-catalysedtransesterification at a temperature below �22�C,which resulted in a higher recovery of 3-MCPD atthe expense of a prolonged transesterification time(18 h). In contrast to alkaline-catalysed transesterifica-tion at ambient temperature, no degradation of3-MCPD was observed during acid-catalysed transes-terification (Hrncirık et al. 2011).
Neutralisation and salting out
Salting out, a step that follows transesterification,facilitates the extraction of lipophilic compounds(FAMEs) from the transesterified mixture. The mostcommon salting out agents used in the analysis of3-MCPD are sodium chloride, sulphate salts andsodium bromide (Divinova et al. 2004; Weisshaar2008; Hrncirık et al. 2011; Kuhlmann 2011a). Theseare added to the reaction mixture either after or duringthe neutralisation.
This step is particularly critical for the specificity ofmethods based on alkaline transesterification. In 2008,the specificity of a newly developed method based onalkaline catalysis and salting out by sodium chloride(Weisshaar 2008) was questioned by demonstratingthat the content of 3-MCPD esters was overestimatedin some samples due to the additional 3-MCPD formedde novo during sample preparation (Kuhlmann 2008).The compound leading to overestimated levels wasconfirmed to be glycidol, which can be present inrefined oils as esters of fatty acids (Weisshaar andPerz 2010).
This finding became the basis for an attempt todevelop a new methodology for the indirect estimationof the level of glycidyl esters in oils. Its principle wasthe calculation of the difference between two indepen-dent determinations: the first one (A) leading to thequantification of the sum of 3-MCPD and glycidylesters (after their conversion to 3-MCPD), while thesecond one (B) was specific to 3-MCPD esters. Suchan approach provides meaningful results only if theformer determination (A) ensures a complete conver-sion of glycidol into 3-MCPD, whilst in the specificmethod (B) any such glycidol conversion is avoided.
4 C. Crews et al.
While the latter can be achieved relatively easy eitherby replacing sodium chloride by other salts duringsalting out (e.g. DGF 2011a; Kuhlmann 2011a) or bythe deliberate decomposition of glycidyl esters priorto transesterification (DGF 2009), the achievementof a complete conversion of glycidol is more difficultbecause it is dependent on both the type of salts andthe conditions applied during the salting out step.
This holds particularly true for pH. Hrncirık et al.(2011) demonstrated that by modifying the pH duringsalting out, the glycidol conversion could vary greatly –it was limited in neutral solutions, but considerablyhigher at pH 2. By monitoring the pH of ‘‘neutralised’’mixtures derived from the application of differentmethods, it can be noticed that the pH varies substan-tially. Therefore, although these methods are based onthe assumption of a complete conversion of glycidol,it is very probable that this will not be always the caseand that the results will depend on the method used.
Unlike the methods employing alkaline transester-ification, the methods based on acid-catalysed transes-terification are not affected by variations in theconditions during salting out (Hrncirık et al. 2011).Glycidyl esters are irreversibly degraded during theacid-catalysed transesterification and do not interferewith the determination of MCPD esters (Hrncirık et al.2011).
Derivatisation and analysis
Due to the low volatility (b.p. 213�C) and high polarityof 3-MCPD, derivatisation prior to the GC analysisis preferable as it greatly improves sensitivity by theformation of a volatile 3-MCPD derivative andspecificity by enabling monitoring of specific fragmentions of the derivative. Several straightforward methodshave been proposed and used since the methodologyfor the analysis of free 3-MCPD was developed in theearly 1990s. The most common derivatisation agentused is PBA, followed by HFBI. Both derivatisationagents represent good alternatives, although HFBI issensitive to moisture and the procedure must be carriedout under anhydrous conditions. It should be notedthat both derivatisation methods enable simultaneousdetermination of both 3 - and 2-MCPD. Alternativemethods used to derivatise free 3-MCPD, such as theformation of dioxolane compounds with acetone, havenot yet been applied to MCPD ester determination.
The most common methods used
Different combinations of the analytical stepsdescribed above, and also additional modificationsto each individual step led to the development of anumber of methods. Table 1 provides an overviewand a basic description of the most common indirectmethods that have been published to date.
Methods based on acid-catalysed transesterification
Methods based on acid-catalysed transesterificationare derived from the pioneering work of Divinova et al.(2004), who applied methanolysis by a mixture ofsulphuric acid and methanol. Since then several mod-ifications have been adopted (Zelinkova et al. 2006;Seefelder et al. 2008; Hrncirık et al. 2011). A certaindisadvantage of this analytical approach is the rela-tively long time of transesterification (16 h), meaningthat it is typically carried out overnight. All acid-transesterification based methods show good robust-ness, i.e. the variation of analytical conditions has alow impact. Comparable results have been obtainedregardless of the changes in salting out (Ermacora andHrncirık 2012a) or the derivatisation procedure(Seefelder et al. 2008). More recently the same authorsincluded an additional step (conversion of glycidylesters to 3-monobromobropanediol (3-MBPD) estersinto their protocol) that enabled glycidyl esters to bequantified simultaneously with 3- and 2-MCPD esters(Ermacora and Hrncirık 2012b).
These methods are specific to 3- and 2-MCPDesters. Recently, the interference of chloride ions andglycidyl esters was evaluated under the conditions ofthe acid catalysed method (Hrncirık et al. 2011). Thestudy demonstrated that 3-MCPD esters were over-estimated only when unrealistically high levels ofchloride ions were added to the samples. Glycidylesters did not interfere with the determination of3-MCPD esters, which corroborates claims of the highspecificity of these methods. Very recently, Miyazakiet al. (2012) successfully employed a lipase fromCandida rugosa for the hydrolysis of 3-MCPD estersand glycidyl esters. Glycidol was subsequently con-verted to 3-MBPD, and following the derivatisationby PBA it was quantified next to 3-MCPD by thestandard GC-MS analysis.
Method based on enzymatic hydrolysis
An alternative to acid-catalysed transesterification isthe enzymatic cleavage of 3-MCPD esters. Hamletand Sadd (2004) applied lipase-catalysed hydrolysis of3-MCPD esters using a lipase from Aspergillus oryzae(24 h incubation at neutral pH). Coupled with deriva-tisation by HFBI and subsequent GC-MS analysis, themethod was used for the determination of MCPDesters in cereal products. Despite a good sensitivity andrepeatability, no further applications of enzymes forthe analysis of MCPD esters have been reported.
Non-specific methods based on alkaline-catalysedtransesterification
The transesterification by sodium methoxide (durationup to 10min) introduced by Weisshaar (2008) led toconsiderable shortening of the sample preparation
Food Additives & Contaminants: Part A 5
Table
1.IndirectanalyticalmethodsforMCPD
andglycidylesters.
Methoddenomination
Analyte
Internalstandard
Transesterification
(tim
e)Derivatisation
agent
Comments
References
Acidic
3-M
CPD
esters
PP-3-M
CPD-d5
Sulphuricacid/m
ethanol
(16h)
PBA
Divinovaet
al.(2004),
Zelinkovaet
al.(2006)
Acidic
3-M
CPD
esters
P-3-M
CPD-d5
Sulphuricacid/m
ethanol
(16h)
PBA/H
FBI
Seefelder
etal.(2008)
Acidic
3-and2-M
CPD
esters,
glycidylesters
PP-3-M
CPD-d5,
glycidyl-P-d5
Sulphuricacid/m
ethanol
(16h)
PBA
Hrncirıket
al.(2011),
Erm
acora
andHrncirık
(2012a,2012b)
Enzymatic(A
spergillus
oryzae)
3-and2-M
CPD
esters
5-�-C
holestane
Lipase
(24h)
HFBI
Applied
tobakerygoods
HamletandSadd(2004)
Enzymatic(C
andida
rugosa)
3-M
CPD
andglycidyl
esters
3-M
CPD-d5,
3-M
BPD-d5
Lipase
(0.5h)
PBA
Miyazakiet
al.(2012)
Alkaline,
non-specific
(DGF
C-III
18
(09)A)
Sum
of3-M
CPD
and
glycidylesters
3-M
CPD-d5
Methoxide/methanol
(5–10min)
PBA
DGF(2009)
Alkaline,
non-specific
(DGF
C-V
I17(10))
Sum
of3-M
CPD
and
glycidylesters
3-M
CPD-d5
Methoxide/methanol
(5–10min)
PBA
DGF(2011a)
Alkaline,
withpre-treat-
ment(D
GF
C-III
18
(09)B)
3-M
CPD
esters
3-M
CPD-d5
Methoxide/methanol
(5–10min)
PBA
Method
withdrawn
in2011
DGF(2009)
Alkaline,
chloride-free
(DGF
C-V
I18(10))
3-M
CPD
esters
PP-3-M
CPD-d5
Methoxide/methanol
(3–5min)
PBA
DGF(2011b)
Alkaline,
mild(SGS)
3-and2-M
CPD
esters,
glycidylesters
PP-3-M
CPD-d5,
3-M
BPD-d5
NaOH/m
ethanol(18h)
PBA
Kuhlm
ann(2011a)
6 C. Crews et al.
time. The method, combined with salting out withsodium chloride, derivatisation with PBA and analysisby GC-MS was adopted as the method DGF C-III18 (09) A (DGF 2009). Following the validation bycollaborative trial (Fiebig 2011), the method wasmodified and adopted as DGF C-VI 17 (10) (DGF2011b). The major drawback of this method is that itis not specific to bound 3-MCPD, as it determinesthe sum of 3-MCPD esters and glycidyl esters.
Specific methods based on alkaline-catalysedtransesterification upon acid pre-treatment
In order to overcome the drawback of the lackof specificity mentioned above, an additional pre-treatment step by sulphuric acid/propanol mixturewas introduced to selectively eliminate glycidyl esters.After the acid pre-treatment, the protocol followed themethod DGF C-III 18 (09) A (DGF 2009) describedin the section on ‘‘Occurrence of glycidyl esters.’’ Themethod was introduced as DGF C-III 18 (09) B, but itwas withdrawn early in 2011 as the pre-treatment stepwas proven to be insufficient for reliable determinationof 3-MCPD esters (Fiebig 2011; Shimizu et al. 2011).
Specific methods based on alkaline-transesterificationand chloride-free salting out
Another way to tackle the issue of specificity is tocombine alkaline-based transesterification with saltingout, in which a salt other than sodium chloride is used.This approach has been applied, with various modifi-cations, by several laboratories and finally adoptedas a DGF C-VI 18 (10) method (DGF 2011a). Thesemethods are specific to 3-MCPD, but the degradationof 3-MCPD during the transesterification step (dis-cussed in section ‘‘MCPD cleavage’’ in detail) remainsa point of concern.
Specific methods based on mild alkaline-transesterification
Performing the alkaline transesterification at lowtemperature virtually eliminates the undesirable con-version of MCPD to glycidol (Kuhlmann 2011a),although the transesterification time is prolonged to18 h (and is thus comparable with acid-catalysedtransesterification). In addition to the determinationof both MCPD isomers, the method enables thesimultaneous determination of the total glycidyl estercontent. This seems to be a promising approach, butthe complex analytical protocol and strict applicationof subfreezing conditions required in the first part ofthe sample preparation increase the chance of errors.So far the method has been performed only by a singlelaboratory, and its performance requires evaluationby collaborative trial.
Methods for indirect determination of glycidyl esters
The choice of a method for the indirect determinationof glycidyl esters is relatively limited. One option is adifferential approach that employs the application oftwo independent analyses: one that determines the sumof 3-MCPD esters and glycidyl esters using sodiumchloride for salting out (DGF 2011b) and a specific onethat determines only 3-MCPD esters. From the differ-ence of both determinations, a total amount of glycidylesters can be calculated; however, due to the reasonsdiscussed in section on ‘‘Direct methods,’’ the accuracyof such an approach still requires to be verified.Another, more elaborated alternative is the use of abromide salt that enables the conversion of glycidol to3-MBPD, as applied in the mild alkaline method,enzymic method and the method based on acid-catalysed transesterification (Table 1).
Kusters et al. (2011) developed a method for thesimultaneous determination of 3-MCPD esters andglycidyl esters in foods based on conversion ofthe glycidol moiety into 3-methoxypropane-1,2-diol(3-MPD) by acidic methanolysis prior to rapid alka-line-catalysed transesterification at ambient tempera-ture and incorporating an isotope-labelled 3-MCPDester as internal standard. The method was validatedfor various fat-rich foodstuffs.
Proficiency tests
In 2009 the Joint Research Centre (JRC) of theEuropean Commission conducted a proficiency teston 3-MCPD ester analysis in oils (Karasek et al. 2010).The test provided the first snapshot that reflected thesituation regarding the analysis of 3-MCPD esters atthat time. Participants of the test, which included,amongst others, commercial, industrial and officialcontrol laboratories, were not restricted in terms of themethod used. Two samples were analysed in the test:a virgin olive oil spiked with 3-MCPD dioleate (bound3-MCPD level 4.58mg kg�1) and a fully refined palmoil (assigned value of bound 3-MCPD: 8.77mgkg�1).In total, 29 out of 34 participants (85%) providedresults which were considered satisfactory (absolutez-score 52) for the spiked sample. The outcome of theanalysis of refined palm oil was much less favourablein that only 56% of participants complied with theperformance criteria. Further analysis of the dataindicated an obvious impact of the analytical methodon the outcome. All results obtained by acid-transesterification based methods were satisfactory(five datasets, in the range of 6.7–9.1mgkg�1).Specific alkaline-transesterification based methods,either those using acid pre-treatment or thosechloride-free, provided much larger variation (3.0–16.5mg kg�1). However, 11 out of 15 datasets werefound compliant. Nine laboratories which appliedthe non-specific alkaline-catalysed transesterification
Food Additives & Contaminants: Part A 7
based method failed, presumably due to the interfer-ence of glycidyl esters present in the refined oil sample.
Another more recent proficiency test was organisedby Food Analysis Performance Assessment Scheme(FAPAS) (2011). Twenty-six datasets were obtainedfor the analysis of 3-MCPD esters in a refined palm oilsample (assigned value¼ 4.7mg kg�1). Compared withthe previous proficiency test, the results showed onlylimited improvement of the situation: 16 datasets(62%) complied with the minimum performance crite-ria (absolute z-score 52) with reported values in therange of 3.5–6.1mg kg�1, 10 laboratories failed,in seven cases due to the same reason as in the caseof the JRC proficiency test, i.e. the use of non-specificalkaline-transesterification based methods.
Gaps and limitations
There has been enormous progress in the developmentof indirect methods for the determination of MCPDesters in the last decade, and in particular in last fouryears. Advances in the methodology have led to theimprovement of the performance (e.g. sensitivity) ofmethods used and to a better understanding of thelimitations of these methods. Substantial effort on thedevelopment of methods based on different principles,and their further modifications, has resulted in a largenumber of different methods being currently available.The fact that these methods differ in their scope andperformance is a cause of some concern. It would behighly desirable to harmonise current analytical meth-odology for MCPD and glycidyl esters and to identifythose methods that meet certain performance criteria.Such harmonisation will not only improve the qualityof the results, but also simplify their communication.To achieve this objective, it seems inevitable thatselected methods will be validated within internationalcollaborative studies and ultimately adopted as officialmethods.
Direct methods
The first data reported on the 3-MCPD ester contentof edible oils were generated by transesterification,isolation and derivatisation of the free 3-MCPDmoiety followed by GC-MS determination. However,it was soon apparent that the data generated werenot comparable and that results depended much onthe procedure applied to specific food matrix types(Karasek et al. 2010). For this reason it would beadvantageous to develop analysis methods that donot require chemical reactions and that allow thedirect measurement of the esters. Well-establishedsample preparation procedures, which are traditionallyemployed in lipid analysis, could be applied dueto the similarities between 3-MCPD esters andacylglycerides.
A challenge for the direct determination of MCPDesters is the large number of substances that have to beincluded into the assays. The distribution pattern ofthe 3-MCPD esters and glycidyl esters is expectedto follow the profile of the fatty esters of the productinvestigated. This complexity of the analyte composi-tion requires selective separation and detection meth-ods, and compared with the indirect methods, largerinvestment in reference materials.
Description of the methods
In the following paragraphs, analysis methods for thedirect determination of glycidyl esters and/or 3-MCPDesters are compiled and discussed. Informationwas extracted from the scientific literature and frompresentations given at recent conferences on this topic.An overview of the methods is given in Table 2.It should be emphasised that method performancecharacteristics presented, such as limit of detection(LOD) and limit of quantification (LOQ), are based onthe respective esters, be they glycidyl esters or 3-MCPDesters and not the free compounds.
The analyte pattern of glycidyl esters is rathersimple as they contain only one fatty acid moiety.Dubois (2011) concluded from theoretical consider-ations that the glycidyl fatty acid ester content in mostedible oils can be sufficiently well characterised by theanalysis of seven glycidyl fatty acid ester species. Theseare the glycidyl esters of lauric, myristic, palmitic,stearic, oleic, linoleic and linolenic acids.
Masukawa et al. (2010) described the determina-tion of five different glycidyl esters (in particularpalmitic, stearic, oleic, linoleic and linolenic acid ester)in edible oils by liquid chromatography atmosphericpressure chemical ionisation mass spectrometry(LC-APCI-MS) in positive-ion mode. The authorsalso considered the application of tandem massspectrometry, but abandoned this approach due tothe lack of selective fragment ions. Sample preparationof the oil samples was based on extraction of theanalytes with acetonitrile followed by dual stagesolid-phase extraction (SPE), employing firstly a C18and secondly a silica cartridge. The authors reportedthat the application of the SPE cartridges in this orderprovided superior results compared with the applica-tion in reverse order. A sub-2mm particle size C18column and an ultra-high-performance liquid chroma-tography system (UHPLC) were employed for theseparation of the glycidyl esters.
The direct injection of a diluted TAG-rich oilsample into the LC-MS system led to fast deteriorationof system performance. The recoveries of glycidylesters spiked into an edible oil sample were dependenton the species and were in the range from about 75%to about 95%. The instrument detection limit (IDL)was estimated for each of the five glycidyl esters to be
8 C. Crews et al.
Table
2.DirectanalyticalmethodsforMCPD
andglycidylesters.
Methoddenomination
Analyte(s)
Calibration
Sample
preparation
Chromatographic
column
Mobilephase
composition
Comments
References
LC-M
S,SIM
,APCI
inpositive-ionmode
Glycidylesters
Internalstandardisation
withglycidyl-P-d31
Dilutionof250mgoilin
5ml
acetone
ODS-A
MC18,3mm
ID�150mm,3mm
particle
size
(YMC-Pack)
SolventA:MeO
H/A
CN/
H2O¼4.25/4.25/1.5;
solventB:acetone
Collisonand
Blumhorst(2011),
Blumhorstet
al.
(2011)
LC-TOFMS
Seven
glycidyl
esters
Standard
addition
Dilutionofsample
incyclo-
hexane/ethylacetate,GPC
clean-uponBio-BeadsS-
X3,GEcontainingfrac-
tionevaporatedand
reconstitutedin
acetone;
additionalclean-upforoils
withhighDAG
andMAG
content:SPEon500mg
silica
(eluentCH
2Cl 2)
AcquityHSST3,
2.1mm
ID�50mm,
1.8mm
particle
size
(Waters)
SolventA:
MeO
H:H
2O¼3:1;
solventB:iso-PrO
H,
both
containingform
icacid
ComparisonofLC-M
S/
MSwithLC-
HRTOFMS;glycidyl
laureate
andglycidyl
myristate
included
Duboiset
al.(2011)
LC-TOFMS
Seven
glycidyl
esters;17
3-M
CPD
mono-and
di-esters
Standard
addition
Dualsample
preparation:(a)
glycidylesterand3-
MCPD
monoester:dual
stageSPE:firstly2gC18
(eluentACN),then
500mg
silica
(eluentCH
2Cl 2);(b)
3-M
CPD
diesters:column
chromatographyon3g
silica
gel
(eluentCH
2Cl 2)
AcquityHSST3,
2.1mm
ID�50mm,
1.8mm
particle
size
(Waters)
SolventA:
MeO
H:H
2O¼3:1;
solventB:iso-PrO
H(both
containingammo-
nium
form
ate
and
form
icacid)
Differentsample
prepa-
rationforMCPD
mono-anddi-esters
Dubois(2011)
LC-M
S,SIM
,APCI
inpositive-ionmode
Seven
glycidyl
esters
Isotopedilutionwith
deuterated(2)and
13C-labelled(5)ana-
logues
ofeach
analyte
Dilutionofsample
inn-pen-
tane:diethylether¼95:5,
clean-upbyfractionation
onsilica
gel
column,
evaporationofsolvent,
andreconstitutionofresi-
duein
acetonitrile
2mm
ID�150mm,
3mm
particle
size
(Phenomenex)
SolventA:0.1%
form
icacidin
H2O;solvent
B:0.1%
form
icacidin
ACN
Highsolventconsump-
tion,poorrecoveries
forsomeanalytes,
butcompensation
withlabelledstan-
dardsforeach
indi-
vidualstandard
Granvogland
Schieberle(2011a)
LC-TOFMS:ESI,SIM
ofsodium
ion
adducts*,positive-
ionmode
Fiveglycidyl
esters,tw
enty
3-M
CPD
ester
mono-and
di-esters
Internalstandardisation
withOO-M
CPD-d5
andglycidyl-P-d31
Dilutionofsample
inauto-
samplervialwithHPLC
solventB
containing
internalstandards
LunaC18column,
3mm
ID�50mm,
3mm
particle
size
(Phenomenex)
SolventA:MeO
H:0.26mM
NaAcin
MeO
H:A
CN¼8:1:1;
solventB:
CH
2Cl 2:0.26mM
NaAc
inMeO
H:A
CN¼8:1:1
‘‘Dilute-and-shoot’’
approach;problems
withinstrumentsta-
bility,corrosionof
ionsourceparts
Haines
etal.(2011)
LC-TOFMSESI,SIM
ofsodium
ion
adducts*,positive-
ionmode
Fiveglycidyl
esters,nine
3-M
CPD
ester
mono-and
di-esters
Internalstandardisation
with1-L-3-M
CPD-
d5,PP-3-M
CPD-d5
andglycidyl-O-d31
Liquid/liquid
partitioningof
edible
oilin
n-hexaneand
acetonitrile,SPEofn-
hexanephase
onSep-Pak
PlusSI,500mg(W
aters),
eluentsCHCl 3,andSPEof
acetonitrile
phase
onSep-
PakPlusC18,500mg
(Waters),eluents
acetoni-
trileandethylacetate,
mergingofeluents,evapo-
rationofsolventand
reconstitutionofresiduein
acetonitrile
AcquityUPLC
BEH
C18,2.1mm
ID�50mm,1.7mm
particle
size
SolventA:0.2mM
l�1
sodium
form
ate
inH
2O:M
eOH¼85:15;
solventB:0.2
mM
l�1
sodium
form
ate
inMeO
H:H
2O¼97.5:2.5
Horiet
al.(2012)
(continued
)
Food Additives & Contaminants: Part A 9
Table
2.Continued.
Methoddenomination
Analyte(s)
Calibration
Sample
preparation
Chromatographic
column
Mobilephase
composition
Comments
References
LC-M
S/M
S,SRM,
APCIin
positive-ion
mode
Seven
glycidyl
esters,tw
enty
3-M
CPD
ester
mono-anddi-
esters
Internalstandardisation
Dilutionofsample
inTBME:diethylether¼4:1,
clean-uponsilica
SPE
cartridge,
eluentn-hex-
ane:diethylether¼94:6
Shim
adzu
UFLC,
ODS4.6mm
ID�150mm,
5mm
particle
size
SolventA:10mM
ammo-
nium
form
ate
and0.1%
form
icacidin
MeO
H;
solventB:iso-PrO
H
MacM
ahonet
al.
(2011)
LC-M
S,SIM
,APCIin
positive-ionmode
Fiveglycidyl
esters
Externalcalibration;
standardsin
MeO
H/
iso-PrO
H(1/1)
Double
SPE:firstcolumn:
Sep-PakVacRC
C18,
500mg(W
aters),second
column:Sep-PakVacRC
Silica,500mg(W
aters)
evaporationofeluentand
reconstitutionofresiduein
MeO
H/iso-PrO
H(1/1)
ODS,4.6mm
ID�150mm,
5mm
particle
size
SolventA:ACN/M
eOH/
H2O¼17/17/6;solvent
B:iso-PrO
H
Methodsuitableforfats
andoils;lower
recoveriescompared
withMasukawa
etal.(2010)
Masukawaet
al.
(2011)
LC-M
S,SIM
,APCIin
positive-ionmode
Glycidylesters
Externalcalibration;
standardsin
MeO
H/
iso-PrO
H(1/1)
Double
SPE:firstcolumn:
Sep-PakVacRC
C18,
500mg(W
aters),second
column:Sep-PakVacRC
Silica,500mg(W
aters),
evaporationofeluentand
reconstitutionofresiduein
MeO
H/iso-PrO
H(1/1)
AcquityUPLC
BEH
C18,2.1mm
ID�100mm,1.7mm
particle
size
(Waters)
SolventA:ACN/M
eOH/
H2O¼17/17/6;solvent
B:iso-PrO
H
Fast
deteriorationof
system
perform
ance
withTAG-richoil
samples
Masukawaet
al.
(2010)
LC-O
rbitrapMS,ESI,
SIM
ofammonium
ionadducts,positive-
ionmode
Twelve3-M
CPD
mono-and
di-esters
Internalstandardisation
withPP-3-M
CPD-d5
and1-P-3-M
CPD-d5
Dualsample
preparation:(a)
3-M
CPD
monoester:SPE
on500mgaminopropyl
cartridge(eluentn-
hexane:CH
2Cl 2:diethyl
ether¼89:10:1);(b)3-
MCPD
di-esters:column
chromatographyon1.8g
silica
gel
(eluentn-hex-
ane:ethylacetate¼85:5)
Kinetex
minibore
C8,
2.1mm
ID�50mm,
1.7mm
particle
size
(Phenomenex)
SolventA
10mM
ammo-
nium
form
ate
inwater;
solventB
10mM
ammonium
form
ate
inMeO
H
Low
detectionand
quantificationlimits
for3-M
CPD
di-
esters,poorionisa-
tionefficiency
for
3-M
CPD
monoesters
Moravcovaet
al.
(2012)
10 C. Crews et al.
DART-O
rbitrapMS,
SIM
ofammonium
ionadducts,positive-
ionmode
Twelve3-M
CPD
mono-and
di-esters
Internalstandardisation
withPP-3-M
CPD-d5
and1-P-3-M
CPD-d5
Dualsample
preparation:
(a)3-M
CPD
monoester:
SPEon500mgaminopro-
pylcartridge(eluent
n-hexane:CH
2Cl 2:
diethylether¼89:10:1);
(b)3-M
CPD
di-esters:
columnchromatography
on1.8gsilica
gel
(eluent
n-hexane:ethyl
acetate¼85:5)
––
Higher
LODsand
LOQscompared
withUHPLC
method,veryshort
analysistime
Moravcovaet
al.
(2012)
LC-M
S/M
S,SRM,ESI
inpositive-ionmode
Fourteen
3-M
CPD
mono-and
di-esters
Standard
addition
Dilutionofsample
inCH
2Cl 2:M
eOH¼1:1
LunaC18column,
3mm
ID�50mm,
3mm
particle
size
(Phenomenex)
SolventA:3mM
NH
4Ac
inMeO
H:A
CN¼9:1;
solventB:
CH
2Cl 2:A
CN:3mM
NaAcin
MeO
H:A
CN¼8:1:1
Severematrix
effects
Pinkstonand
Stoffolano(2011)
LC-M
S,SIM
,APCIin
positive-ionmode
Fiveglycidyl
esters
Externalcalibration;
andinternalstan-
dardisationwith
C17:0-G
E,standards
inMeO
H/iso-PrO
H(1/1)
Double
SPE:firstcolumn:
Sep-PakVacRC
C18,
500mg(W
aters),second
column:Sep-PakVacRC
Silica,500mg(W
aters),
evaporationofeluentand
reconstitutionofresiduein
MeO
H/iso-PrO
H(1/1)
ODS,4.6mm
ID�150mm,5mm
particle
size
SolventA:MeO
H;solvent
B:iso-PrO
HFurther
developmentof
methodpresentedby
Masukawaet
al.
(2011);interferences
experienced;
improved
LODs
Shiro,Kondo,
Kibune,
etal.
(2011)
LC-M
S,SIM
,APCIin
positive-ionmode
Fiveglycidyl
esters
Externalcalibration;
standardsin
MeO
H/
iso-PrO
H(1/1)
Double
SPE:firstcolumn:
Sep-PakVacRC
C18,
500mg(W
aters),second
column:Sep-PakVacRC
Silica,500mg(W
aters),
evaporationofeluentand
reconstitutionofresiduein
MeO
H/iso-PrO
H(1/1)
ODS,4.6mm
ID�150mm,5mm
particle
size
SolventA:
MeO
H:H
2O¼92:8;sol-
ventB:iso-PrO
H
Shiro,Kondo,
Masukawa(2011)
Food Additives & Contaminants: Part A 11
below 20 pg on-column. However, a drawback of themethod was the application of external calibration foranalyte quantification, which is not able to compensatefor matrix effects encountered.
In the following paper, Masukawa et al. (2011)proposed a slightly modified analysis procedure for thedetermination of the five aforementioned glycidylester species. The main changes were the dilution ofthe sample in a mixture of acetone and chloroformthat also enabled the analysis of solid fats, and theapplication of conventional high-performance liquidchromatography (HPLC) instead of UHPLC. Thelatter modification aimed to broaden the group ofpotential users, as it does not require specialisedinstrumentation. However, the authors noted thatapplying the initial clean-up procedure with thechange of UHPLC to HPLC resulted in lower analyterecoveries. They attributed this to interferences stem-ming from residual TAGs. The detection capability ofthis method was improved compared with the originalmethod by a factor of three, which might be theconsequence of the higher injection volume applied inthe HPLC procedure. The recovery values of glycidylesters spiked into an edible oil were slightly above100% for the modified procedure. Repeatabilityrelative standard deviations for all analytes had amaximum of 4.0%.
The same group of authors proposed a furthermodification of the method, which resulted in short-ened analysis times and again improved detectioncapability (Shiro, Kondo, Kibune, et al. 2011). TheIDLs of this third method were about eight times lowerthan the IDLs of the original method. The LOD of theanalytes in edible oil were estimated to range between25 and 33 mg kg�1. Another improvement was theelimination of chlorinated solvents. The acetone/chlo-roform mixture was replaced by tert-butyl methylether/ethyl acetate (4:1 v/v), and the eluent of thenormal phase SPE clean-up was n-hexane/ethyl acetate(95:5 v/v). The application of glycidyl heptadecanoateas internal standard did not improve analytical preci-sion compared with external calibration. In a presen-tation given at the 102nd AOCS Annual Meeting (1–4May 2011 in Cincinnati, OH, USA) the authorsdescribed interferences experienced with the thirdanalysis procedure (Shiro, Kondo, Masukawa 2011).In particular glycidyl linoleate eluted with almost thesame retention time as an unknown substance, whichseemed to be extracted from plastic parts of the SPEmanifold. A slight modification of the HPLC mobilephase composition remediated this problem withoutaffecting method performance for the other analytes.
UHPLC with high resolution time of flight massspectrometry (UHPLC-TOFMS) analysis with internalstandardisation with one isotope-labelled glycidyl esterand two isotope-labelled 3-MCPD esters was per-formed by Hori et al. (2012) for the simultaneous
determination of five glycidyl esters, three 3-MCPDmono-esters and six 3-MCPD di-esters in edible oils.Their sample preparation encompassed liquid–liquidpartitioning of the edible oil in n-hexane and acetoni-trile, and subsequent solid-phase extraction on silicaand C18 cartridges. In contrast to the methoddescribed above, the two cartridges were used inparallel. The n-hexane phase was loaded onto thesilica cartridge, whereas the acetonitrile phase wasapplied to the C18 cartridge. The collected eluents werecombined, evaporated and reconstituted in acetonitrileprior to the measurement. Interferences as describedby Shiro, Kondo, Masukawa (2011) were not reported,which could be attributed to the higher mass spectro-metric resolution of the applied TOF instrument.
None of the analysis procedures mentionedabove has been validated for the direct determinationof glycidyl laurate and glycidyl myristate, which areexpected to occur in some oils such as palm kernel oilor coconut oil. This gap was recently filled by Duboiset al. (2011). The method comprised gel permeationchromatography (GPC) of the diluted edible oilsample, which provides sufficient separation of glycidylesters from the bulk matrix. GPC clean-up wascomplemented by SPE in silica for samples containingpalm kernel oil or coconut oil. The authors consideredthe SPE clean-up necessary in order to separate theglycidyl ester fraction from co-eluted short chaintriacyl- and diacylglycerols (DAG), as it improvedanalytical precision and stability of the instrument. Themeasurements of the sample extracts were performedby both LC-TOFMS and liquid chromatography-tandem mass spectrometry (LC-MS/MS).
The latter technique did not show advantagescompared with LC-TOFMS as collision induced dis-sociation (CID) provided only non-selective fragments.Quantification was performed by external quantifica-tion as well as by standard addition, and internalstandardisation by five isotope-labelled glycidyl esters.The authors identified standard addition as the mostsuitable quantification method. This was becausematrix effects could be compensated for in the absenceof isotope-labelled analogues which were not yetcommercially available. Recoveries and precisionof the analyses were assessed on spiked safflower oilsamples. Recoveries of all analytes were above 70%in almost all experiments, whereas intermediate preci-sion was usually better than 25%. Recoveries for fiveof the seven glycidyl esters were quite stable overthe concentration range (0.1–7.5mg kg�1) tested. Onlyglycidyl laurate and glycidyl myristate had highervariability.
These two glycidyl fatty acid esters are also part ofthe scope of an analysis procedure developed at theDeutsche Forschungsanstalt fur LebensmittelchemieLeibniz Institut (DFA – German Research Centre forFood Chemistry). Granvogl and Schieberle (2011a)
12 C. Crews et al.
reported at the 102nd AOCS Annual Meeting amethod that enables the determination of sevenglycidyl ester species by stable isotope dilution analysisLC-MS. Sample clean-up consisted of fractionation onsilica gel 60 (16 g) containing 7% of water and stepwiseelution of the analytes with pentane/diethyl ether (95/5v/v). The fraction between 150 and 250ml wascollected, evaporated to dryness and reconstituted inacetonitrile. A pentafluoropentylpropyl stationaryphase and a water/acetonitrile mobile phase gradientwere employed for the HPLC separation of the glycidylesters. Analytes were detected by positive-ion APCI-MS in selected ion monitoring mode (SIM).
Recovery of the analytes was rather poor. Thelowest recovery value was for glycidyl linoleate,at about 34%. The advantage of the application ofisotope dilution is that analyte losses during samplepreparation are corrected for. On the other hand thestandards cannot influence LOD and LOQ. Details onmethod performance characteristics other than recov-ery were not provided. A similar procedure applyingsilica SPE cartridges instead of silica gel columnsfor sample clean-up was developed at the USFDA(MacMahon et al. 2011). They applied a standard C18column and methanol containing ammonium formateand formic acid for chromatographic separation,and APCI-MS/MS for the detection of the analytes.Method performance was evaluated on two differentdays, two spiking levels and for three different matri-ces. The recoveries reported ranged between 79% and114%, whereas the relative standard deviation of theanalyses was between 3% and 16%.
Haines et al. (2011) and Collison and Blumhorst(2011) applied the least degree of sample preparationfor the determination of glycidyl esters in vegetableoils. It consisted only of the dilution of the edible oilin an organic solvent prior to the injection into the LCsystem. The method also enabled the determinationof a range of 3-MCPD esters, which will be describedbelow. The determination of the analytes wasperformed by electrospray ionisation LC-TOFMS inSIM. Analytes were separated on a C18 stationaryphase applying a mobile phase gradient consisting ofmethanol, acetonitrile, and methylene chloride.Sodium acetate was added to the mobile phase inorder to generate sodium ion adducts of the analytes.The limits of detection of glycidyl esters in palm oilwere about 100 mg kg�1, with the exception of glycidylmyristate for which the LOD was about 290 mg kg�1.Hence, the LODs were about three times higher thanthose reported by Shiro, Kondo, Kibune, et al. (2011).The precision of the analysis procedure, expressed asrelative standard deviation, was reported to lie between5% and 10%. Unfortunately, no information wasprovided on the recovery of the analytes from differentoils, which could vary significantly due to matrixeffects.
A major drawback of the dilute-and-shoot
approach chosen by Haines et al. (2011) is the use of
low levels of sodium salts in the mobile phase to drive
the formation of the sodium adduct ion. The high load
of matrix results in the need for frequent instrument
cleaning and maintenance. Additionally, mobile phase
components have caused corrosion of some ion source
parts such as the ESI nebuliser needle. In a recent
update of the analysis procedure, methylene chloride
was replaced by non-chlorinated solvents, the sample
was diluted in acetone prior to injection, and LC-
APCI-MS was applied instead of LC-ESI-TOFMS
(Blumhorst et al. 2011; Collison and Blumhorst 2011).
The LOD varied for the seven glycidyl fatty acid esters
between 40 mg kg�1 of oil and 160 mg kg�1 of oil.Recovery values were determined for two edible oils
at three different fortification levels. They indicated
significant matrix effects, with a tendency to over-
estimation of the content of some analytes.The determination of 3-MCPD esters alone, or in
combination with glycidyl esters, provides additional
challenges compared with the determination of glycidyl
fatty acid esters alone. The challenges result not only
from the additional number of analytes that has to be
covered, but also by the strong similarity of the
analytes to major matrix constituents, in particular
monoacylglycerols (MAGs) and DAGs.Reports in literature on the direct determination of
MCPD esters are very limited. This might be attribut-
able to the lack of suitable commercial reference
materials in the early days of this topic. Laboratories
active in this field were until recently obliged to
synthesise reference materials themselves. Details of
the synthesis of glycidyl esters and/or MCPD esters
can be taken from several publications (Masukawa
et al. 2010; Granvogl and Schieberle 2011a; Haines
et al. 2011). However, the commercial supply of
reference materials has recently improved significantly.
A broad range of both native and stable isotope-
labelled glycidyl and MCPD esters is available
from different suppliers. A non-exhaustive list of
suppliers comprises (in alphabetical order) Chiron AS(Trondheim, Norway), Toronto Research Chemicals
(Toronto, ON, Canada), and Wako Pure Chemical
Industries, Ltd (Osaka, Japan).The Lebensmittelchemisches Institut des
Bundesverbandes der Deutschen Sußwarenindustrie
(LCI) (2008) published an equation for the calculation
of the number of possible 3-MCPD esters depending
on the number of different fatty acids contained in a
particular edible oil (LCI 2008). Hence, six different
fatty acids would theoretically provide 96 different
3-MCPD mono- and di-ester combinations, including
all possible stereoisomers. It is not realistic to expect to
be able to determine routinely all possible 3-MCPDesters. Hence, a compromise and agreement has to be
Food Additives & Contaminants: Part A 13
found as to which analytes should be included in theanalysis.
Gaps and limitations
The papers presented so far have to be seen in viewof the lack of such agreement and the limitationsprovided by the limited availability of reference mate-rials. Haines et al. (2011) included five 3-MCPD mono-esters and fifteen 3-MCPD di-esters in their analysisprocedure, whereas MacMahon et al. (2011) presentedchromatograms of twenty-two 3-MCPD esters,Moravcova et al. (2012) analysed three 3-MCPDmono-esters and nine 3-MCPD di-esters, andPinkston and Stoffolano (2011) focused on in total14 mono- and di-esters of 3-MCPD, which wereaccording to them the most relevant. Hori et al.(2012) developed an analysis method for three3-MCPD mono-esters and six 3-MCPD di-esters.Dubois (2011) concluded from his considerationsthat the determination of ten 3-MCPD di-esterswould be sufficient to cover between 90% and 99%of the expected total 3-MCPD di-ester content,depending of the type of edible oil. Exceptions areprovided by coconut oil and palm kernel oil, whichcontain high amounts of short chain fatty acids.Additionally, seven 3-MCPD mono-esters should beincluded in the assay.
The analytical approaches applied for the determi-nation of 3-MCPD esters in edible oils differ widely.Haines et al. (2011) applied the dilute-and-shootapproach described above. They used deuteratedglycidyl palmitate as internal standard for 3-MCPDmono-esters and deuterated 3-MCPD-dioleate for3-MCPD di-esters. The LODs were between 15 and60 mg kg�1 for MCPD mono-esters, between 4 and10 mg kg�1 for 3-MCPD di-esters containing atleast one unsaturated fatty acid, and 25 mg kg�1 for3-MCPD di-esters containing only saturated fattyacids. The precision of the method, expressed as relativestandard deviation, was in the range of 5–10%.However, questions again arose concerning the stabil-ity of the analysis system and the matrix effectsencountered.
Pinkston and Stoffolano (2011) applied an analyt-ical procedure similar to that of Haines et al. Theirmodifications consisted of an ammonium salt mobilephase additive in place of the sodium salt used byHaines et al. (2011) and the application of tandemquadruple mass spectrometry operated in selectedreaction monitoring mode instead of LC-TOFMS.Use of the ammonium salt mobile phase additiveavoided the degradation of instrument performance,and associated cleaning and maintenance, described byHaines et al. (2011). Remarkably, they experiencedsevere matrix effects in the analysis of vegetable oilsamples, which caused them to choose quantification
by standard addition. The LOQs achieved were inthe best case about 20–50 mg kg�1 for mono-estersand 50–100 mg kg�1 for di-esters. However, LOQsof di-esters rose for some oil matrices to 500–1000 mg kg�1. For this reason, they recommend thatsample clean-up be included in future methods.
The disadvantages of using sodium salts in themobile phase were also tackled by Moravcova et al.(2012) who investigated alternatives. The addition ofammonium acetate to the mobile phase and conse-quently the measurement of ammonium ion adductswas identified as best solution and provided for3-MCPD di-esters detection limits comparable withthose of sodium ion adducts. The lower limits of theworking range of their UHPLC-OrbitrapMS methodwas for the 3-MCPD di-esters studied in the rangeof 2–5mg kg�1. Mass resolution was set to 50.000FWHM. The total run time of the UHPLC methodwas 8min. This was further reduced by applyingDART-OrbitrapMS for measurement of the 3-MCPDesters. Cycle times per sample were shortened to about20 s. Sample preparation for 3-MCPD di-ester analysisconsisted of column chromatography over silica geland was identical for both measurement techniques.However the sensitivity of the DART-OrbitrapMSanalysis was lower than that of UHPLC-OrbitrapMS.3-MCPD mono-esters were isolated from the matrixaccording to the procedure described by Seefelder et al.(2008). Neither the UHPLC nor the DART basedmeasurement method was sufficient sensitive for thedetermination of 3-MCPD mono-esters in real lifesamples.
A more complex analysis procedure was presentedby Dubois (2011). It consisted of the separate deter-mination of MCPD mono-esters and MCPD di-esters.Double SPE was applied for the isolation of MCPDmono-ester, whereas MCPD di-esters were separatedfrom the bulk by column chromatography on silica gel.Both MCPD ester fractions were measured by LC-TOFMS. Dubois confirmed the need to apply eitherisotope dilution with stable isotope-labelled glycidyland MCPD esters, or standard addition for compen-sating for matrix effects (Dubois 2011; Dubois et al.2011). Isotope dilution would be easier and faster thanstandard addition, but is hampered by the limitedavailability of labelled analogues of the analytes.An effect that cannot be compensated for by eithermethod is caused by differences of responses ofco-eluting isomers, which could cause significant bias.Dubois reported a difference of response between1-palmitoyl-3-MCPD and 2-palmitoyl-3-MCPD of40% (Dubois 2011). In addition the two estersco-eluted with palmitoyl-2-MCPD.
A more recent study (Dubois et al. 2012) comparedthe performance of an indirect and a direct method.The indirect procedure used acid-catalysed trans-methylation and HFBI derivatisation. The direct
14 C. Crews et al.
method had two separate extraction procedures, oneused double SPE cartridges to isolate 2 - and 3-MCPDmono-esters and glycidyl esters and the second useda single silica column to isolate 2- and 3-MCPDdi-esters. The esters were measured by LC-TOFMSand quantified using matrix-matched standard addi-tions. There was no significant bias when the methodswere used to analyse numerous samples of palm oils.The indirect method was considered more suitablefor routine application. Consequently, the trueness ofanalysis results has to be questioned. Unfortunately,the analytical community lacks a supply of edible oilwith certified glycidyl ester and MCPD ester content.For this reason, analysts have compared their resultswith those gained by other methods, including methodsbased on the indirect determination of glycidyl andMCPD esters. So far it has not been made clear whichmethod operated under which conditions providesresults closest to the true value.
Collison and Blumhorst evaluated the results of thedirect determination of glycidyl esters (Blumhorst et al.2011; Collison and Blumhorst 2011) in five edible oilsagainst the results obtained with a slightly modifieddouble SPE procedure proposed by Shiro, Kondo,Kibune, et al. (2011). The results gained by the twomethods gave similar results. However, their associatedmeasurement uncertainties values did not agree.
Discussion
It is not clear which method gives the ‘‘better’’ results.Questions regarding over- and underestimation of theanalyte contents by one or the other method have tobe tackled and answered in order to progress towardsa generally accepted analysis procedure.
The development of analytical methods for thedetermination of individual glycidyl esters in edibleoils is more advanced than the development of directanalysis methods for the determination of MCPDesters in oil. This might be explained by the simplertask and the easier access to reference materials, whichwere scarcely available in the early days of this topic.
Comparison of direct and indirect analytical methods
Introduction and background
Whereas initial analytical methods for MCPD andglycidyl esters were mainly based on the detection of3-MCPD and glycidol that could be released fromthe esters (indirect analysis), more recent advances inMCPD and glycidyl ester analysis have targeted thedirect determination of the individual compounds.This trend may be explained by the fact that indirectanalytical methods do not provide any detailed infor-mation on the chemical structure of the differentesters. This type of information may, however, become
relevant when the chemical structure of the estersmight turn out to have an impact on their biologicalfate or toxic potential, and, therefore, differentiationwould become necessary. The major triggers of thedevelopment of direct methods however are the con-cerns expressed on the precision and trueness ofcommonly applied indirect analytical methods forMCPD esters, whose performances are thought to becompromised by the analytical steps as mentionedabove. Even though promising, more accurate directanalytical methods have a limited applicability in aroutine environment mainly due to the relativelyhuge number of analytical standards they require.Comparison of direct and indirect analytical methodsmay therefore serve as an alternative for evaluating thetrueness of indirect analytical methods since certifiedreference materials are not yet available.
The following paragraphs provide an overviewon studies dealing with the comparison of direct andindirect analytical methods for MCPD and glycidylesters and gives furthermore recommendations whichmay be taken into consideration for promoting fitfor purpose and commonly recognised methods forMCPD and glycidyl esters.
Overview on comparative studies
Data on the comparison of direct and indirect analyt-ical methods have mainly been presented in the frameof conferences. Table 3 provides an overview on theoutcome of current studies.
Direct and indirect methods for glycidyl esters
The first comparison of direct and indirect analyticalapproaches for glycidyl esters was carried out byShimizu et al. (2011). The authors concluded thatthe results obtained from direct and indirect measure-ment of the glycidyl esters were not comparable.Commercially available samples in particular showedlower glycidyl ester levels or even negative results whenapplying the indirect method. For direct determina-tion, glycidyl esters were isolated by double SPEand measured by liquid chromatography coupled to asingle quadrupole mass spectrometer (LC-MS)equipped with an APCI interface. Calibration of thedirect analytical method was carried out externally(no internal/labelled standards). For measuring sam-ples indirectly, the DGF differentiating method in itsformer version C-III 18 (09) (DGF 2009) was applied,including the use of sodium chloride in the salting outand derivatisation step. The authors suggested that thepresence of sodium chloride in this procedure mayhave led to the formation of additional amounts of3-MCPD through the means of partial acylglycerides(MAG and DAG) or through residues of glycidylesters which were incompletely destroyed during the
Food Additives & Contaminants: Part A 15
Table
3.Comparisonofdirectandindirectanalyticalmethods.
Methodsdenomination
Analytescompared
Oilmatrices
included
Indirect
method
withdrawn,
yes/no
2-and3-M
CPD
differentiatedby
directmethod,
yes/no
Comments
References
DirectLC-M
S–double
SPE
versusIndirectDGF
C-III
18
(09)A;B
GE
4cookingoil,2palm
,2spread
Y–
Non-comparable
andinconclusive
results
Shim
izuet
al.(2011)
DirectLC-M
S-TOF;‘‘dilute
and
shoot’’versusindirectDGF
C-V
I18(10)A;B
GEandMCPD
esters
4palm
,1blend
NN
Significanthigher
resultsofMEs
bytheindirectmethod.
Comparable
resultsforthe
absence
ofglycidylester
Hinrichsen(2011)
DirectLC-M
S;double
SPEversus
directLC-M
S;‘‘dilute
and
shoot’’
GE
1palm
,1soybean,
1corn
––
Incanola
oiltheresultsmatched
nicely,in
palm
oilnot.A
special
shapeofthespraysourceis
recommended
for2.3.3-D
Collisonand
Blumhorst(2011)
DirectLC-M
S;‘‘dilute
andshoot’’
versusindirectDGF
C-III
18
(09)part
A
Sum
ofGEandMCPD
esters
2palm
,2blend
NN
Non-comparable
and
inconclusiveresults
Haines
etal.(2011)
DirectLC-M
S/LC-M
S2,SPE
versusindirectDGF
C-III
18
(09)A;Bversusindirect
GC-M
S,alkalinemild(SGS)
GE
2avocado,1sunflower,
1palm
,2DAG
rich,
1TAG
rich
2.3.1-I:Y;
2.3.3-I:N
–DirectGEsdeterminationin
oil
samplesshowed
bettercorrela-
tionwithSGS‘‘3-in-1’’than
withDGF
C-III
18(09)
Granvogland
Schieberle(2011a)
DirectLC-M
S2;‘‘dilute
and
shoot’’versusindirect
MCPD
esters
9vegetable
NN
Comparable
resultsforseven
of
nineinvestigatedvegetable
oil
samples.Indirectapproach
rather
isrelatedto
theDGF
C-V
I18(10)methodthanto
aWeisshaarmethod
Pinkstonand
Stoffolano(2011)
A)DirectLC-M
S2;GPC
andSPE
versusB)directLC-M
S/LC-
MS2,SPE(D
FA)versusC)
indirectGC-M
S,acidic
versus
D)indirectGC-M
S,alkaline
mild(SGS)versusE)indirect
DGF
C-V
I18(10)A;B
GEandMCPD
esters
A),B),C),D)10palm
;D),E)10soybean,
13sunflower,53
palm
N–
A),B),D)comparable
resultsfor
GEsfrom
somedirectmethods
versusindirect‘‘SGSindirect
3-in-1’’method.C),D),E)
comparable
resultsforallthree
indirectmethodsregarding3-
MCPD.Comparable
resultsfor
2-M
CPD
determined
indirectly
bytw
omethods(2-M
CPD
was
notevaluatedwithin
methodE)
Kuhlm
ann(2011b),
Granvogland
Schieberle
(2011b)
DirectLC-TOF;double
SPE
and
SGC
formono-M
Eanddi-ME
versusindirectGC-M
S,acidic
MCPD
esters
30palm
N(Y
)indirectly
Analysisofmono-estersin
30palm
oilsamplesbydirectandindi-
rect
methodsgavesimilar
results
Dubois(2011)
16 C. Crews et al.
acidic pre-treatment. Differences in the results from
both methods may therefore be largely explained by
the drawbacks of the withdrawn C-III 18 (09) method.
Also, the absence of internal standards for the direct
analysis of glycidyl esters prevented compensation for
losses of the analytes during sample preparation.A method based on GPC separation combined with
an additional clean-up by solid-phase extraction on
silica-gel and LC-MS detection has been validated
including four 13C-labelled (glycidyl palmitate, oleate,
linoleate and linolenate) standards for the analysis of
seven glycidyl esters in various edible oils by (Dubois
et al. 2011). The method has been compared with the
indirect analytical method for glycidyl esters known as
‘‘SGS 3-in-1 method’’ (Kuhlmann 2011a), which is
based on alkaline catalysed release of glycidol, fol-
lowed by a transformation of glycidol to MBPD,
derivatisation with PBA and analysis by GC-MS.
Results obtained from the analysis of 10 samples of
refined palm oil showed a good correlation of results,
however, accompanied by a systematic bias
(Kuhlmann 2011b).The same samples were further analysed by a stable
isotope dilution analysis based direct analytical
method developed at the DFA (Granvogl and
Schieberle 2011a) using seven 2H- or 13C-labelled
standards (glycidyl laurate and linolenate). In this
method LC-APCI-MS was used for detection of the
individual esters. Samples were cleaned over silica gel
columns. Even though results generally agreed well
there was no full correlation between any of the three
methods (Figure 3).In a further study, the DFA method was compared
with the ‘‘SGS 3-in-1 method’’ as well as to the DGF
C-III 18 (09) method. Regarding the DFA and theSGS methods, the comparison of results for sevensamples of different types of oils (two avocado, onepalm oil, one sunflower oil, one rice husk oil and twoDAG-rich oils) showed good correlation for all sam-ples except for the rice husk oil where the glycidolcontent obtained with the indirect method exceeded thevalue resulting from the direct approach by approxi-mately 30%. The correlation to the DGF C-III 18 (09),which was in the meantime withdrawn, was lesssatisfying and levels seemed to be underestimatedwith this method (Granvogl and Schieberle 2011a).
Direct and indirect methods for MCPD esters
The first comparison of direct and indirect analyticalmethods for MCPD esters was presented by Haineset al. (2011). The DGF method (C-III-18 (09)) appliedto measure the total MCPD equivalents (withoutdifferentiating MCPD and glycidyl esters) wasreported to give results that were consistently higherthan with the LC-TOFMS method that measuredMCPD and glycidyl esters directly. The LC-TOFMSmethod included two labelled standards (one labelledglycidyl ester and one labelled 3-MCPD di-ester).Samples were injected together with the labelledstandards after dilution without any clean-up.2-MCPD esters were not included in the analysis.
The new DGF method (C-VI 18 (10)) containedimprovements, through the implementation of alabelled 3-MCPD di-ester (d5-3-MCPD-1,2-bis-palmi-toyl ester), the use of chloride-free salts during saltingout in part B, and performing the derivatisationprocedure at room temperature in an organic solvent.
Figure 3. (Colour online). Comparison of 10 palm oil samples analysed by two direct methods (developed by DFA (Granvogland Schieberle 2011b) Nestle Research (Dubois et al. 2011) and one indirect analytical method developed by SGS (Kuhlmann2011a). The amounts of glycidyl esters detected by the different methods are expressed as total amounts of glycidol present in theglycidyl esters (mg kg�1).
Food Additives & Contaminants: Part A 17
However, neither of the modifications was sufficientto explain the huge discrepancies between the levelsof MCPD and glycidyl esters measured directly andindirectly. The authors (Haines et al. 2011) highlightedthe problem of rapid sensitivity losses due to fatresidues and the presence of sodium in the mobilephase. It is worth mentioning that the direct measure-ment approach did not consider the presence of2-MCPD esters, which have the same molecularweights as their analogous 3-MCPD esters. 2-MCPDesters are likely to co-elute with the 3-MCPD estersduring column chromatography, however, their signalresponse may be different (Dubois 2011).
Comparison data from a follow-up study werepresented at the 2011 OVID/BLL conference in Berlin/Germany (Hinrichsen 2011). This time, the new DGFmethod (C-VI 18 (10)) had been applied to measureMCPD and glycidyl esters indirectly. Comparison ofresults for ester-bound 3-MCPD from this new studyto those of the LC-TOFMS method mentioned aboveshowed that positive results in palm oil samples thathad been specially treated to remove 3-MCPD andglycidyl esters were obtained only by indirect analysis,and the direct analysis could not confirm their presenceabove a limit of 0.1mg kg�1. Both direct and indirectapproaches found glycidyl esters only at very low levelsif at all.
Recently, the development of a direct analyticalmethod for MCPD esters using a ‘‘dilute and shoot’’approach combined with LC-MS/MS for measurementusing standard additions with labelled and non-labelled standards was put forward (Pinkston andStoffolano 2011). The method, which did not consider2-MCPD esters, was compared with a modified versionof the original ‘‘Weisshaar method’’ where sodiumchloride is substituted by sodium sulphate (comparablewith DGF C-VI 18 (10) B). Results of nine vegetableoil samples presented showed in general a good
correlation but seemed to be biased, and valuesobtained from direct measurement were below thoseobtained by the applied indirect analysis. These effectsmay be explained by the fact that for direct analysisonly a limited set of reference compounds was avail-able. Furthermore, some 3-MCPD di-esters weredifficult to quantify due to some matrix interferences,indicating the possible need for implementation of aclean-up step prior to LC-MS/MS analysis.
Dubois (2011) recently presented the developmentof two direct analytical methods, the first targeting thedirect analysis of intact MCPD mono-esters andthe second one targeting the direct analysis of MCPDdi-esters. MCPD mono- and di-esters were isolated viadouble SPE (C18 and silica) or silica gel columns,respectively, and analysed by LC-TOFMS. Standardaddition quantification was used, with labelled internalstandards. The method was compared with an indirectanalytical method for 3- and 2-MCPD esters, whichwas based on the acid-catalysed transesterificationof the MCPD esters (Divinova et al. 2004), Extrelut�
extraction, derivatisation with HFBI, and GC-MS.For comparison, 30 different oil samples (mainly palmoil based) were analysed. The presence of 2-MCPDesters in the oils had been taken into consideration.The correlation between results showed a slope of0.975 and constant bias of �0.12mg g�1 indicatingthat the two methods provided very similar results(Figure 4).
Gaps and limitations
So far only a very limited number of data for thecomparison of direct and indirect analytical methodsfor MCPD and glycidyl esters are available. With theexception of the study from Dubois on MCPD esters(Dubois 2011), the indirect methods subjected to inter-method comparison have been mainly alkaline based,
Figure 4. Determination of MCPD esters in 30 oil samples by the indirect method (acid-catalysed transesterification, HFBIderivatisation) and the direct method of Dubois et al. (2011). Regression calculations do not include coconut and palm kerneloil as these oils could not be quantified reliably by the present direct method (lack of specific analytical standards).
18 C. Crews et al.
which have been frequently modified and improved.With the exception of the study by Dubois, the directmethods used for comparison have not take intoconsideration the presence of 2-MCPD esters. Sincethese compounds carry the risk of co-elution with3-MCPD esters on liquid chromatography, andhave different signal intensities, a misinterpretation ofresults from these studies cannot be excluded.
Since direct analytical methods for glycidyl estershave shown to be relatively easy to handle andreference material more easily available, the compar-ison of direct and indirect analytical approaches mayin the future become more relevant for the MCPDester analysis. The indirect analysis of MCPD estersmay be impaired by the presence of glycidyl esters andby unidentified co-occurring compounds. The prepa-ration of adequate reference materials may thereforebecome difficult. Once indirect methods for MCPDesters that meet certain performance criteria have beenidentified and harmonised, their trueness could beassessed on naturally contaminated samples, and themethods compared with appropriate direct procedures.
Discussion
For glycidyl esters, in comparison to the directanalytical methods for MCPD esters direct analyticalmethods for glycidyl esters have been shown to requirea minimum of time and effort and a reasonably smallnumber of analytical standards (Dubois et al. 2011;Granvogl and Schieberle 2011b). Since direct analysisis not considered to be impaired by errors originatingfrom transesterification or derivatisation procedures itis favoured over indirect approaches for the quantifi-cation of glycidyl esters.
For MCPD esters, correlation without any signif-icant bias can so far be demonstrated only forcomparison of a direct LC-TOFMS method (samplepreparation over C18 and/or silica material) alsotaking into consideration the presence of 2-MCPDesters, versus an indirect method for 3 - and 2-MCPD
esters based on the cleavage of the MCPD esters underacidic conditions (Figure 5).
Indirect analytical methods require significantlyfewer chemical standards, to be less cumbersome, beeasily applicable to all type of different vegetable oilsand to allow easy distinction between 3- and 2- ‘‘esterbound MCPD.’’ Data showing a similarly goodcorrelation between other direct methods and alkalinebased transesterification methods such as the newDGF method (DGF C-VI 18 (10) Part B) or the ‘‘SGS3-in-1 method’’ are currently not available. However, acollection of data for different type of vegetable oils (26samples) showed an excellent correlation between the2 - and 3-MCPD esters determined by the ‘‘SGS 3-in-1method’’ and the acidic based transesterificationmethod mentioned above (Figure 5).
Analytical methods and occurrence data
Introduction
Early findings showed that refined palm oil is likely tobe the major dietary source of both MCPD esters andglycidyl esters. Samples of refined oils measured todate have shown high levels of the contaminants andthe oil is popular with food manufacturers. It is acommon ingredient of infant formulae and infantfollow-on formulae, which are often the major foodsource for this vulnerable population. Variable rangesof 3-MCPD esters have been found in certain foodgroups. They are not normally found in unrefinedvegetable oils or infant and baby foods in cans andjars. Higher but still relatively low levels have beenreported for infant formula, oil-based creamers, andheated cereals such as malt and bread crust. Thehighest levels have been found in refined vegetable oils,and particularly in palm oil.
Refined palm oils contain about 1–10mgkg�1 of3-MCPD esters with fractionated palm oils having less,e.g. palm oleins contained 0.4–0.6mg kg�1. A singlesurvey of 2-MCPD in 20 refined palm oils showed a
Figure 5. Comparison of values obtained from 26 refined vegetable oil samples analysed by the SGS 3-in-1 method and anacid-catalysed transesterification method (Divinova et al. 2004; Robert et al. 2004).
Food Additives & Contaminants: Part A 19
range of about 0.2–6mgkg�1. Glycidyl esters in palmoils range from 0.3 to 10mg kg�1 with up to15mgkg�1 in palm olein. Other refined oils generallycontain lower levels (typically 1–2mgkg�1) but withexceptional samples having about 20mgkg�1. Otherfoodstuffs analysed for MCPD and glycidyl esters arethose associated with high temperature processing or ahistory of free MCPD occurrence. Moderately highlevels of 3-MCPD esters have been reported in cookedpotato products (0.02–1.2mgkg�1), infant formulae(0.1–0.6mg kg�1), and toasted cereals (0.04–1.4mgkg�1). Only low levels have been reported inunrefined oil (50.3mg kg�1), most breads (50.011–0.042mgkg�1), and infant and baby food in cans andjars (50.01mg kg�1).
Tables 4–6 collate occurrence data published in2004–2011. Different analytical methods were usedfor the generation of these data and, thus care isrequired when considering the diversity and reliabilityof the data.
Analytical methods used for the occurrence data
The methods used for the first isolation and identifi-cation of 3-MCPD esters were based on TLC followedby GC-MS. They were applied to study the composi-tion of acid-HVP and to characterise the products ofthe reactions of fats and fat components with hydro-chloric acid in an investigation of the chlorinatedpropanols found in acid-HVP.
Following a revival in interest in the esters,methods based on cleavage of the acid/chloropropanolbond were devised and later methods enabling thedirect determination of individual esters have beenused on a small scale.
The TLC based method was not intended toprovide more than an indication of levels of MCPDesters that might be found in foods where free MCPDhad been encountered. They were carried out beforecharacterised standards and internal standards hadbeen prepared and without regard to the methodperformance and, thus quantification may be regardedas providing approximate data. The results obtainedshowed a very wide range of MCPD ester levels in thesame food type, and high levels (up to 6.4mgkg�1) inmeats that have not been evident in later analyses.
The first acid methanolysis based on indirectdetermination of MCPD esters without prior TLCisolation of fractions used in the analysis of foodstuffswas reported by Svejkovska et al. (2004). Both mono-and di-esters of 3-MCPD were found at levels of up toabout 35mgkg�1 of the fat portion (in French friesand dark malts). The detection limit was about1mgkg�1 of fat.
At the same time, enzyme cleavage was appliedby Hamlet and Sadd (2004) on the analysis of baked
cereal foods giving results comparable with theacid-catalysed transesterification methods used shortlyafter.
The alkaline transesterification causes rapiddecomposition of 3-MCPD esters, and early resultsusing this approach are probably unreliable (Hrncirıket al. 2011; Kuhlmann 2011a). Reduction of thetemperature improves the recovery after alkalinetransesterification considerably (Kuhlmann 2011a).
In the DGF method CIII 18b (2009), glycidyl esterswere determined indirectly in oils from measurementsof the difference between quantification of the sumof 3-MCPD and glycidyl esters, and determination of3-MCPD esters alone by avoiding sodium chlorideaddition, or by decomposition of glycidyl esters priorto transesterification.
Very many factors do or might have an effect onthe level of MCPD esters and glycidyl esters reported.These include:
. Fat extraction procedure.
. Methanolysis method – acid or basic orenzymatic.
. Methanolysis procedure reagent composition/concentration, reaction time and temperature.
. Addition of salt, or the choice of salt (foralkaline methanolysis).
. pH of the salting out solution (for alkalinemethanolysis).
. Method used to isolate and determine theMCPD produced.
However, it has been shown that high levels ofchloride can be used in acid methanolysis withoutoverestimation of 3-MCPD esters, and that glycidylesters do not interfere with the determination(Ermacora and Hrncirık 2012a). It has been shownthat the derivatisation procedure used for the determi-nation of the released free MCPD does not affect theresult (Seefelder et al. 2008).
Occurrence of 3-MCPD esters
Data on the occurrence of 3-MCPD esters in differentfoodstuffs are listed in Table 4. Data obtained with the2008 version of DGF method C-III-18 have not beenincluded in the table, because with this method, thesum of 3-MCPD and glycidol is determined. Althoughthousands of analyses of ester-linked 3-MCPD havebeen performed in recent years, only a small part of thedata have been published in peer reviewed journals orofficial reports. Data of 3-MCPD esters in thermallytreated foodstuffs other than fats and oils are sporadicand very heterogeneous due to differences in compo-sition and different thermal treatment of the singlefoodstuffs. Noticeable are high amounts found insalami and other microbiologically ripened foods;although the formation processes responsible are not
20 C. Crews et al.
Table
4.Occurrence
infood–ester-linked
3-M
CPD.
Class
Food
Method
Number
Range
(mgkg�1)
Comments
References
Cereals
Barley
Acidcleavage
80.044–1.386
Dolezalet
al.(2009)
Cereals
Biscuits
Acidcleavage
60.249–0.696
FSA
(2010)
Cereals
Breadassorted,UK
Acidcleavage
55
0.011–0.042
FSA
(2010)
Cereals
Breadtoasted
Enzymaticcleavage
70.06–0.16
Hamletet
al.(2004),HamletandSadd
(2004)
Cereals
Breakfast
cereals
Acidcleavage
50.04–0.888
FSA
(2010)
Cereals
Coffee
surrogates
Acidcleavage
50.145–1.184
Divinovaet
al.(2007)
Cereals
Crackers
TLC-G
C-M
S,acidcleavage
30.1–1.14
Reece
(2005),Svejkovskaet
al.(2004)
Cereals
Crispbread
TLC-G
C-M
S,acidcleavage
20.42–0.58
Reece
(2005),Svejkovskaet
al.(2004)
Cereals
Doughnuts
TLC-G
C-M
S,acidcleavage
20.42–1.21
Reece
(2005),Svejkovskaet
al.(2004)
Cereals
Rye,
wheat,flour
Acidcleavage/enzymatic
cleavage
34
50.005–1.02
Roasted
and
unroasted
Hamletet
al.(2004),HamletandSadd
(2004),Dolezalet
al.(2005,2009)
Coffee
Coffee
TLC-G
C-M
S,acidcleavage
26
50.1–0.39
Green,roasted
andsoluble
coffee
Reece
(2005),Svejkovskaet
al.(2004),
Dolezalet
al.(2005)
Dairy
Coffee
creamer
Acidcleavage
15
0.13–0.73
Karsulınovaet
al.(2007)
Dairy
Cream
(aerosol)
Acidcleavage
10
0.05–0.73
Karsulınovaet
al.(2007)
Dairy
Cheese
TLC-G
C-M
S,acidcleavage
5n.d.–1.28
Feta,Parm
esan,
processed
cheese
Reece
(2005),Svejkovskaet
al.(2004)
Infantandbabyfood
Cereal
Acidcleavage
55
0.011–0.23
FSA
(2010)
Infantandbabyfood
Infantbiscuits
Acidcleavage
20.11–0.306
FSA
(2010)
Infantandbabyfood
Jarred
food
Acidcleavage
55
0.011
FSA
(2010)
Infantandbabyfood
Humanbreast
milk
DirectGC-TOF-M
S126
50.011–0.076
Zelinkovaet
al.(2008)
Infantandbabyfood
Infantform
ula
Acidcleavage
11
50.075–0.588
Zelinkovaet
al.(2009a)
Infantandbabyfood
Milk,growing-up
Acidcleavage
30.062–0.291
Zelinkovaet
al.(2009a)
Infantandbabyfood
Infantform
ula
(fatportion)
DGF
C-III
18(09)
56
0.57–4.1
Fatcontent
23–28%
Weisshaar(2011),includingadditional
unpublished
data
Malt
Maltandbeer
Acidcleavage
20
0.004–0.65
Svejkovskaet
al.(2004),Divinovaet
al.
(2007),Dolezalet
al.(2009)
Meat
Chicken
grilled
TLC-G
C-M
S,acidcleavage
40.26–0.74
Reece
(2005),Svejkovskaet
al.(2004),
FSA
(2010)
Meat
Ham
TLC-G
C-M
S,acidcleavage
4n.d.–2.64
Smoked
and
unsm
oked
Reece
(2005),Svejkovskaet
al.(2004)
Meat
Salami
TLC-G
C-M
S,acidcleavage
60.88–6.41
Reece
(2005),Svejkovskaet
al.(2004),
Zelinkovaet
al.(2006)
Meat(fish)
Mackerel,herring
TLC-G
C-M
S,acidcleavage
30.28–1.08
Smoked
and
pickled
Reece
(2005),Svejkovskaet
al.(2004)
Miscellaneous
Bouilloncube
Acidcleavage
50.38–0.67
Karsulınovaet
al.(2007)
Miscellaneous
Nuts
roasted
Acidcleavage
4n.d.–0.5
Svejkovskaet
al.(2004),Zelinkovaet
al.
(2006)
Miscellaneous
Sweetspreads(fatportion)
DGF
C-III
18(09)
20
50.15–3.9
Fatcontent
30–45%
Weisshaar(2011),includingadditional
unpublished
data
Oilsandfats
Anim
alfats,unrefined
DGF
C-III
18(09)
25
50.1–0.14
Weisshaar(2011),includingadditional
unpublished
data
Oilsandfats
Cocoabutter
DirectLC-TOF
25
0.5
Haines
etal.(2011)
(continued
)
Food Additives & Contaminants: Part A 21
Table
4.Continued.
Class
Food
Method
Number
Range
(mgkg�1)
Comments
References
Oilsandfats
Fryingoils(fresh
andused)
DGF
C-III
18(09)
51
50.15–16.2
Weisshaar(2011),includingadditional
unpublished
data
Oilsandfats
Margarine(fatportion)
DGF
C-III
18(09)
22
50.15–7.7
Fatcontent
40–80%
Weisshaar(2011),includingadditional
unpublished
data
Oilsandfats
Mayonnaise(fatportion)
DGF
C-III
18(09)
17
50.15–1.04
Fatcontent
19–80%
Weisshaar(2011),includingadditional
unpublished
data
Oilsandfats
Palm
shortening/olein
DirectLC-TOF
50.4–0.6
Haines
etal.(2011)
Oilsandfats
Refined
coconutoil
Acidcleavage
21.418–1.694
Karsulınovaet
al.(2007)
Oilsandfats
Refined
palm
kernel
oil
Acidcleavage
30.85–1.40
Karsulınovaet
al.(2007)
Oilsandfats
Refined
palm
oil
Acidcleavage
41.39–4.17
Karsulınovaet
al.(2007)
Oilsandfats
Refined
vegetable
oils(notfrying)
DGF
C-III
18(09)
153
50.15–18.8
Weisshaar(2011),includingadditional
unpublished
data
Oilsandfats
Refined
vegetable
oils
Alkali/Br�
32
50.1–5.2
Kuhlm
ann(2011a)
Oilsandfats
Refined
hazelnut/walnutoil
Alkali/Br�
61.2–19.0
Kuhlm
ann(2011a)
Oilsandfats
Refined
oliveoil
Acidcleavage
55
0.3–2.462
Oliveoiland
olivepomace
oil
Zelinkovaet
al.(2006)
Oilsandfats
Refined
palm
oil
Alkali/Br�
20
1.1–10.0
Kuhlm
ann(2011a)
Oilsandfats
Refined
seed
oils
Acidcleavage
55
0.3–1.234
Zelinkovaet
al.(2006)
Oilsandfats
Refined
seed
oils
Alkali/Br�
15
50.1–2.1
Kuhlm
ann(2011a)
Oilsandfats
Refined
salm
onoil
Alkali/Br�
70.7–13.0
Kuhlm
ann(2011a)
Oilsandfats
Refined
vegetable
fats/oils
Acidcleavage
11
0.897–2.435
Differentiation
monoester–
di-ester
Seefelder
etal.(2008)
Oilsandfats
Unrefined
oils
Alkali/Br�
420
50.05
Kuhlm
ann(2011a)
Oilsandfats
Unrefined
vegetable
oils
DGF
C-III
18(09)
122
50.15–0.31
Weisshaar(2011),includingadditional
unpublished
data
Oilsandfats
Vegetable
oils
DirectLC-TOF
12
50.5
Haines
etal.(2011)
Oilsandfats
Virgin
oils
Acidcleavage
14
50.1
to5
0.3
Zelinkovaet
al.(2006)
Potato
products
Crisps
Acidcleavage
30
0.098–0.606
Dolezalet
al.(2008)
Potato
products
Crisps
Acidcleavage
20
0.048–1.186
FSA
(2010)
Potato
products
Crisps
Acidcleavage
16
0.229–1.008
Zelinkovaet
al.(2009b)
Potato
products
French
fries
Acidcleavage
20
0.035–0.397
FSA
(2010)
Potato
products
French
fries
Acidcleavage
16
0.1–0.258
Zelinkovaet
al.(2009b)
Potato
products
Homecooked
potato
products
Acidcleavage
17
0.014–0.225
FSA
(2010)
Potato
products
Mashed
Acidcleavage
30.038–0.275
Dolezalet
al.(2008)
Potato
products
Pre-fryingproducts
Acidcleavage
16
0.027–0.064
Zelinkovaet
al.(2009b)
22 C. Crews et al.
Table
5.Occurrence
infood–ester-linked
glycidol.
Class
Food
Method
Number
Range
(mgkg�1)
Comments
References
Infantandbabyfood
Infantform
ula
(fatportion)
DGF
C-III
18(09)
56
50.15–3.0
Fatcontent23–28%
Weisshaar(2011),includingaddi-
tionalunpublished
data
Miscellaneous
Sweetspreads(fatportion)
DGF
C-III
18(09)
20
50.15–2.1
Fatcontent30–45%
Weisshaar(2011),includingaddi-
tionalunpublished
data
Oilsandfats
Anim
alfats,unrefined
DGF
C-III
18(09)
25
50.1
Weisshaar(2011),includingaddi-
tionalunpublished
data
Oilsandfats
Cocoabutter
DirectLC-TOF
250.2
Haines
etal.(2011)
Oilsandfats
Cookingoil
DGF
C-III
18(09)
35–24
Shim
izuet
al.(2011)
Oilsandfats
Cookingoil
SPEþLC-M
S3
3–28
Shim
izuet
al.(2011)
Oilsandfats
Fryingoils(fresh
andused)
DGF
C-III
18(09)
51
50.15–10.7
Weisshaar(2011),includingaddi-
tionalunpublished
data
Oilsandfats
Margarine(fatportion)
DGF
C-III
18(09)
22
50.15–5.0
Fatcontent40–80%
Weisshaar(2011),includingaddi-
tionalunpublished
data
Oilsandfats
Mayonnaise(fatportion)
DGF
C-III
18(09)
17
50.15–0.33
Fatcontent19–80%
Weisshaar(2011),includingaddi-
tionalunpublished
data
Oilsandfats
Palm
shortening/olein
DirectLC-TOF
60.4–15.6
Haines
etal.(2011)
Oilsandfats
Refined
vegetable
oils(not
frying)
DGF
C-III
18(09)
153
50.15–4.1
Weisshaar(2011),includingaddi-
tionalunpublished
data
Oilsandfats
Refined
vegetable
oils
Alkali/Br�
32
50.1–3.1
Kuhlm
ann(2011a)
Oilsandfats
Refined
hazelnut/walnutoil
Alkali/Br�
60.5–1.4
Kuhlm
ann(2011a)
Oilsandfats
Refined
palm
oil
DGF
C-III
18(09)
36–8
Shim
izuet
al.(2011)
Oilsandfats
Refined
palm
oil
SPEþLC-M
S3
8–10
Shim
izuet
al.(2011)
Oilsandfats
Refined
palm
oil
Alkali/Br�
20
0.30–1.80
Kuhlm
ann(2011a)
Oilsandfats
Refined
seed
oils
Alkali/Br�
15
50.10–0.60
Kuhlm
ann(2011a)
Oilsandfats
Refined
salm
onoil
Alkali/Br�
750.10–1.20
Kuhlm
ann(2011a)
Oilsandfats
Unrefined
oils
Alkali/Br�
420
50.025
Kuhlm
ann(2011a)
Oilsandfats
Unrefined
vegetable
oils
DGF
C-III
18(09)
122
50.10
Weisshaar(2011),includingaddi-
tionalunpublished
data
Oilsandfats
Vegetable
oils
DirectLC-TOF
10
50.2–3.7
Haines
etal.(2011)
Oilsandfats
Sunflower
oil
DirectLC-M
S4
1.2–2.1
GranvoglandSchieberle
(2011b)
Oilsandfats
Rapeseedoil
DirectLC-M
S3
0.2–0.36
GranvoglandSchieberle
(2011b)
Oilsandfats
Avocado
DirectLC-M
S4
2.5–7.8
GranvoglandSchieberle
(2011b)
Oilsandfats
Olive,
refined
DirectLC-M
S3
1.4–4.1
GranvoglandSchieberle
(2011b)
Food Additives & Contaminants: Part A 23
yet known. Data on oils, fats and foods containing
refined fats clearly show that the fat refining process is
the most important factor causing the formation of
3-MCPD esters. Refined palm oil and palm oil
fractions were identified as the most problematic
food ingredients for the dietary exposure to 3-MCPD
esters due to high contamination levels and the
widespread use of these fats.
Occurrence of glycidyl esters
Data on the occurrence of glycidyl esters are listed in
Table 5. Except for a handful of oil samples, all results
were obtained with indirect methods. The data clearly
show that the refining process is responsible not only
for the formation of 3-MCPD esters, but also for the
formation of glycidyl esters. Glycidyl esters have been
found only in refined vegetable oils and in products
containing them, including infant formulae, where
levels of up to about 3mgkg�1 have been reported.
Refined palm oils contained about 0.5–10mgkg�1 of
glycidyl esters with some identified frying oils contain-
ing almost 30mgkg�1. Other refined oils generally
contained up to 5mgkg�1.Data describing glycidyl ester levels in thermally
treated products except where derived from added
refined fats and oils have not been published so far.
Occurrence of 2-MCPD esters
Data on the occurrence of 2-MCPD esters, determined
in about 60 vegetable oils in a single study with an
indirect method (Kuhlmann 2011a), are listed in
Table 6. Ranges were between 0.2 and 6mgkg�1 for
refined palm oils and usually 50.5mgkg�1 for other
refined oils. Some other oil samples contained up to
11mgkg�1. Data on the occurrence of 2-MCPD esters
in thermally treated products excluding fats and oils
have not been published up to now.
Gaps and limitations
It is evident that, up to now, only a small number of
reliable occurrence data have been made available in
peer reviewed journals or official reports. Current
knowledge of formation related to occurrence has been
summarised by Hamlet et al. (2011). Apart from theproblems of reliability and comparability of analytical
methods mentioned above, the most important gaps
are that there are only few and incidental occurrence
data on 3-MCPD esters in thermally treated foodstuffs
except refined fats and oils, and none on glycidyl esters
and 2-MCPD esters. The development of reliable
quantitative extraction procedures for different types
of food is a necessary precondition to closing this gap.Only few occurrence data of glycidyl esters are
available and only a handful of data of 2-MCPD esters
are available. As 2- and 3-MCPD can easily be
separated by GC-MS after derivatisation, it should
be possible to determine their esters simultaneously in
the same sample using indirect methods. Only very few
data have been published on 3-MCPD ester levels in
commercial food samples outside Europe.Analytical difficulties exist in that unless an official
method is available, because most investigators use
different approaches or even subtle changes that might
affect the result. The situation is compounded by the
paucity of well validated methods and the changes that
have been made to the official DGF (2009) over a short
period of time.
Discussion
The variety of factors that influence the quantitative
results of food sample analysis is so great that it is
difficult to analyse their effect for every survey
reported. Important factors are not restricted to the
analytical method. Conditions of edible oil refining
temperature and duration vary considerably, as can the
proportion of ingredient refined oils used in the foods,
Table 6. Occurrence in food – ester-linked 2-MCPD.
Class Food Method NumberRange
(mg kg�1) Comments Reference
Oils and fats Refined evening primrose oil Alkali/Br� 3 50.1–1.0 Kuhlmann (2011a)Oils and fats Refined grape kernel oil Alkali/Br� 4 0.4–2.4 Kuhlmann (2011a)Oils and fats Refined olive oil Alkali/Br� 4 50.1–0.4 Kuhlmann (2011a)Oils and fats Refined peanut oil Alkali/Br� 4 50.1–0.4 Kuhlmann (2011a)Oils and fats Refined palm oil Alkali/Br� 20 0.2–5.9 Kuhlmann (2011a)Oils and fats Refined rapeseed oil Alkali/Br� 45 50.1–0.3 Kuhlmann (2011a)Oils and fats Refined safflower oil Alkali/Br� 4 50.1–0.4 Kuhlmann (2011a)Oils and fats Refined soybean oil Alkali/Br� 45 50.1–0.1 Kuhlmann (2011a)Oils and fats Refined sunflower oil Alkali/Br� 45 50.1–0.3 Kuhlmann (2011a)Oils and fats Refined walnut/hazelnut oil Alkali/Br� 6 0.5–11 Kuhlmann (2011a)Oils and fats Refined salmon oil Alkali/Br� 7 0.1–0.3 Kuhlmann (2011a)
24 C. Crews et al.
and variations between different samples of the foodtype might be due just to this.
Analytical methods in biological samples
Background
Until very recently, analytical methods to analyseMCPD esters and glycidyl esters in biological samplesobtained from animal studies were simply not avail-able. Other than reports of 3-MCPD esters beingfound in fresh, unprocessed goat’s milk published inthe 1980s (Cerbulis et al. 1984), analytical methodsaddressed exclusively free 3-MCPD or glycidol andtheir metabolites after direct dosing of animals, andthose metabolites were determined mainly in urinesamples (see below). Only more recently, whenparticularly high levels of MCPD esters, and laterglycidyl esters, were found in refined vegetable oils thatpotentially contribute considerably to human foodexposure, has the fate of these compounds in theorganism received renewed attention. Their structuralsimilarity with lipids suggested the potential (partial)release of free 2-MCPD, 3-MCPD, or glycidol in thehuman gastrointestinal tract, but until very recently, itwas not known to what extent this release occurs, if itchanges the kinetics of their distribution, or if theesters themselves cause or contribute to specificadverse effects.
The metabolic fate and the potential toxicity of 3-MCPD and glycidyl esters as compared with their freecounterparts are being investigated in two majorstudies, a sub-chronic 90-day study (European FoodSafety Authority (EFSA) 2011) and a biokinetic studyperformed by the BfR. The fate and potential adverseeffects of 2-MCPD esters in animals or humans havenot at all been addressed to date.
Published studies involving analytical methodslinked to MCPD esters, glycidyl esters and theirmetabolites in biological samples (from animals orhumans, considering also non-processed milk as bio-logical sample) can be categorised as follows:
. In vivo studies addressing the metabolism offree 3-MCPD and glycidol.
. Occurrence of 3-MCPD esters in fresh, non-processed milk (goat, human).
. Analysis of 3-MCPD, glycidol, their esters,and their metabolites in tissues, blood andurines from animals treated with 3-MCPDand glycidyl esters.
. Analysis of biomarkers of exposure inhumans.
The following section briefly summarises thepathways involved in 3-MCPD and glycidol metabo-lism. Then those analytical methods for the parentcompounds and their metabolites with relevance to the
subject and issues today are discussed in more detailand summarised in Table 7.
Metabolic fate of 3-MCPD and glycidol
Most data on metabolism date back to the 1970s whenanalytical methods were far less developed than today,and more qualitative than quantitative. These studiesinvolved (1) following the fate of the radioactivity inanimals after administration of a radiolabelled parentcompound and characterisation/identification of theradioactivity in general or specific radiolabelled metab-olites in body fluids or tissues, or (2) analysis of theparent compound or a specific metabolite directly witha specific analytical method.
Early studies in the 1970s addressed the metabolicfate of 3-MCPD and the possible involvement ofglycidol as an obligatory reactive intermediate in itsdetoxification in rats and mice (Jones 1975; Jones et al.1978). The first urinary metabolites identified were thecysteine conjugates generated through glutathione(GSH) conjugation of glycidol as a possible interme-diate: these were the cysteine conjugate S-(2,3-dihy-droxypropyl)cysteine (DHPC) and the correspondingmercapturic acid N-acetyl-S-(2,3-dihydroxypropyl)cys-teine (DHPMA) that were identified in rat urine. It wasalso suggested that traces of glycerol were formed andexcreted in urine. A significant proportion of the 3-MCPD administered intraperitoneally was excretedunchanged in the urine, and significant quantities ofthe dose were exhaled as CO2 in rats and mice (Jones1975).
The conjugation to glutathione has been shown forseveral other mono- and dihalopropanols suggestingglycidol as a metabolic intermediate (Jones andFakhouri 1979). In addition, the only metabolicstudy available on 2-MCPD (Jones 1973) also identi-fied the mercapturic acid DHPMA as a urinarymetabolite, suggesting a common intermediate with3-MCPD, i.e. the epoxide glycidol. However, thetoxicities of 2 - and 3-MCPD are very different, andadditionally very different from the toxicity of glycidolby itself (reviewed in Schilter et al. 2011) suggestingthat glycidol is not relevant or insignificant as a (free)toxic intermediate. In parallel to the studies character-ising the metabolisation through GSH conjugation, anoxidative degradation pathway of 3-MCPD wasdiscovered trough transformation to the intermediates�-chlorolactaldehyde and �-chlorolactic acid resultingin excretion/exhalation of oxalic acid, CO2 and Cl�
(Jones et al. 1978).Metabolism studies on glycidol are limited, but
available data indicate that 3-MCPD and glycidolshare overlapping metabolic pathways. Glycidol isconverted to glycerol and exhaled as CO2 inrats and mice and produces the same GSH conjugates
Food Additives & Contaminants: Part A 25
Table
7.Analyticalapproaches
tostudyingthelevelsofMCPD,glycidol,theiresters
andmetabolitesin
biologicalsamples.
Analyte(s)
Internalstandard
Matrix/extraction
Transesterification
(tim
e)Derivatisation
Instrumentation
Comments
References
3-M
CPD
None
Humanurine;
3�ethyl
acetate
extraction
–NoneorBSTFA
GC-ECD
orGC-M
S(EI;SIM
)Calibrationusingspiked
humanblankurine
DeRooijet
al.
(1996)
3-M
CPD
3-M
CPD-d5
Raturine,
blood;silica
gel
column,ethylacetate
extraction
–HFBA
GC-M
S(N
CI;SIM
)Calibrationusingspiked
pooledblankratblood
orurine
Berger-Preiss
etal.(2010)
3-M
CPD,�-chloro-
lactic
acid,
DHPMA
3-M
CPD-d5,�-
chlorolactic
acid-d3,S-
DHPMA-d5
Raturine;
sample
dilution
1:100or1:1000
––
LC-M
S/M
S(N
IM;
SRM)
Additionally,anim
alswere
directlydosedwithdeu-
teratedcompounds
EFSA
(2011)
DHPMA
Benzylm
ercapturic
acid-d5
Humanurine;
acidification
topH
1.5–2with6N
HCl,lyophilisation,
neutralisationwith2N
NaOH,lyophilisation
anddesaltingin
methanol/acetone
methanolicHCl
(30min)
BSTFA
GC-M
S(EI;SIM
)Sensitivityinsufficientfor
humanurinesamples
from
workersexposedto
epichlorohydrin;inter-
feringsignals
DeRooijet
al.
(1997)
DHPMA
13C2-D
HPMA
Humanurine;
acidification
topH
2.5
withform
ate/
form
icacid,SPEwith
form
icacid/m
ethanol
elution
––
HIL
IC-ESI-MS/M
Sin
NIM
andMRM
Calibrationusingspiked
blankpooledhuman
urines
Eckertet
al.
(2010)
3-M
CPD
ester
3-M
CPD
Goatmilk;lyophilisation,
petroleum
ether
extrac-
tion,silica
gel
column
withhexane-benzene
1N
methanolicHCl
(16hRT)or
H2SO
4
–TLC
onsilica
gel
Gor
MS-D
CIGLC
for
FA
methylester
analysis
Cerbuliset
al.
(1984)
A)3-M
CPD;B)
3-M
CPD
ester
(determined
as
free
3-M
CPD
after
transesterifi-
cation);C)3-
MCPD
ester
determined
as
single
fattyacid
esterderivatives
3-M
CPD-d5;
PP-3-M
CPD,
2-dipalm
itate
Humanmilk;oxalate/etha-
nol/hexane-diethylether
extraction,solvent
exchangeto
hexane,
silica
gel
column,disso-
lutionin
THF
A)–;B)Sulphuric
acid/m
ethanol
(16h40�C);C)–
A)PBA;B)
PBA;C)–
GC/M
S(EI;SIM
)or
GCxGC
TOF-M
S(EI)
Determinationoffree
and
bound3-M
CPD
accord-
ingto
Zelinkovaet
al.
(2006)
Zelinkovaet
al.
(2008)
A)3-M
CPD;B)sum
ofglycidyland
3-M
CPD
esters
3-M
CPD-d5
A)cellculture
supernatants
dilutionwithsodium
phosphate
buffer;matrix
clean-upwithhexane-
acetone;
solvent
removal;B)cellculture
supernatants;lyophilisa-
tion,dissolutionin
t-butylm
ethylether/ethyl
acetate
A)–;B)NaOCH
3/
methanol
PBA
GC/M
S(EI;SIM
)A)accordingto
Divinova
etal.(2004);B)
Accordingto
DGF
MethodC-III-18(09)
Opt.A
Buhrkeet
al.
(2011)
N-(2,3-dihyroxypro-
pyl)valineof
globin
(2H5)N
-2,3-dihy-
droxypropyl-
adductsofglobin
(diH
OPrV
al)
Lysederythrocytes;globin
precipitation,ethylace-
tate
extraction,dissolu-
tionin
form
amide
–pentafluorophenyl
isothiocyanate,
diethylether
extraction,acety-
lationwithTEA
GC-M
S/M
S(N
CI)
N-terminalvalinebymodi-
fied
Edmandegradation.
Adduct
canbeform
edfrom
severalcompounds,
e.g.ECH,glycidol,
3-M
CPD,other
halohydrins
Landin
etal.
(1996)
as 3-MCPD, i.e. DHPC and DHPMA, in the urine ofrodents and humans (reviewed in Nomeir et al. 1995;Eckert et al. 2010). The metabolic pathways for 3-MCPD, 2-MCPD and glycidol are summarised inFigure 6 (adapted from Jones 1975; Jones et al. 1978;Lynch et al. 1998).
Some microorganisms are able to dehalogenate 3-MCPD to glycidol (van den Wijngaard et al. 1989),which was thought not to be a relevant reaction inhumans (Lynch et al. 1998). Recently, the possibility ofthe formation of a mercapturic acid conjugate from the�-chlorolactaldehyde intermediate has been raised(Habermeyer et al. 2011), but no experimental datasupporting this hypothesis was provided.
It is not entirely clear which metabolite(s) contrib-ute(s) to or cause the toxicities (renal, testicular) of 3-MCPD. While available data indicate the contributionof the oxidative pathway in the development ofreversible infertility in males of various animal speciesby inhibition of glycolysis in the epididymal tract(reviewed in Jones 1983), the causal involvement ofoxalic acid in the kidney carcinogenicity of 3-MCPD inrats was discussed but remains questionable.
The toxicity of glycidol has been studied since theearly 1970s (reviewed in Nomeir et al. 1995). Initialstudies on metabolism indicated the conversion ofglycidol to �-chlorohydrin (metabolite of 3-MCPD) by
stomach hydrochloric acid. This hypothesis was thor-oughly tested by Nomeir et al. (1995) demonstratingthat glycidol-derived radioactivity was mainly excretedin urine and exhaled as CO2. Urinary metaboliteprofiling identified one major and three lesser metab-olites (in total 15 metabolites were identified). Theindividual metabolites were not characterised further,but �-chlorolactic acid was quantitatively insignificantand independent of the route of administration indi-cating the absence of a significant conversion ofglycidol to 3-MCPD by hydrochloric acid in thestomach.
Metabolites of both the oxidative and the GSHconjugation pathway were recently confirmed in a90-day toxicological study with 3-MCPD and itsdipalmitate ester in rats (EFSA 2011). Free (but notbound) 3-MCPD, �-chlorolactic acid, and DHPMAwere analysed in urines at the end of the study.Interestingly, dosing of either free 3-MCPD or thedipalmitate ester resulted in similar profiles of thesemetabolites indicating that 3-MCPD is rapidly releasedfrom its ester and enters into the same metabolicpathways as 3-MCPD. The study also showed that theoxidative pathway is likely of minor importance, sinceonly low levels of �-chlorolactic acid close to the LOQwere found, while the mercapturic acid conjugateDHPMA was readily quantifiable at all dose levels of
Figure 6. Metabolic pathways for 3- and 2-MCPD and glycidol (adapted from Jones 1975; Jones et al. 1978; Lynch et al. 1998).GST, glutathione S-transferase; EH, epoxide hydrolase; NAT, N-acetyltransferase.
Food Additives & Contaminants: Part A 27
3-MCPD and its di-ester. The toxic effects of free andester-bound 3-MCPD were surprisingly similar andindicated no particular adverse effects specific to ester-bound 3-MCPD (palmitic di-ester). However, theseverity of the adverse effects was milder whendosing equimolar amounts of the di-ester indicatingthat the release kinetics may play a critical role in theirtoxicity. A current biokinetic study in rats conductedby the BfR is likely to confirm a relatively quickand complete release of the free compounds from both3-MCPD and glycidyl esters. These results seem tosubstantiate initial assumptions of the esters as addi-tional sources of dietary exposure to free MCPD andglycidol, respectively.
Analytical methods for 3-MCPD, glycidol,and their metabolites
The studies conducted in the 1970s were pivotal inidentifying the metabolic fate of 3-MCPD and glycidol.The methods used mostly involved TLC of urinarysamples (with or without solvent extraction) afteradministration of radioactively labelled compounds toexperimental animals. Identification of the analyteswas usually by comparison to the Rf values ofauthentic reference compounds, and quantificationwas by radioactivity scanning of the TLC plates orscintillation counting. Calcium oxalate was further-more identified by microscopic examination of diureticurine from treated animals (Jones 1975; Jones et al.1978; Nomeir et al. 1995). These methods are unlikelyto be of methodological relevance today; therefore, thefollowing sections focus on more modern state-of-the-art methods and include information on limits ofdetection and quantification that are likely applicabletoday.
3-MCPD, glycidol, and their metabolites in urineand blood
Non-radioactive, quantitative analysis of 3-MCPD (asa metabolite of epichlorohydrin) in rat urine was firstreported by De Rooij et al. (1996). Urine samples wererepeatedly extracted with ethyl acetate after addition ofinternal standard. Pooled organic phases were sub-jected directly to gas chromatography with electron-capture detection (GC-ECD). Alternatively, the ethylacetate extracts were silylated and subjected to GC-MSwith SIM detection. Calibration curves were obtainedfrom control urines spiked with authentic 3-MCPD.3-MCPD was identified by the presence of 5 charac-teristic ions in the mass spectrum of the trimethylsilylderivatives. The LOD of underivatised 3-MCPD was2 mgml�1 of urine, after TMS derivatisation the LODwas 0.4mgml�1 of urine and as low as 0.05mgml�1 ofurine when using higher injection volumes. Themethod was not sensitive enough to determine the
cysteine conjugates of 3-MCPD, DHPC, and DHPMA
in human urines (De Rooij et al. 1997). The methods
developed were proposed for use in occupational
settings for workers exposed to epichlorohydrin.Berger-Preiss et al. (2010) were recently able to
improve significantly the LOQ of 3-MCPD analysis in
urine and blood samples, which was achieved by GC-
MS with SIM in negative chemical ionisation mode
(GC-MS-NCI), with a reported LOQ of 2 ngml�1. The
method was designed for use in biokinetics studies in
rats. Urinary samples were acidified with acetic acid,
spiked with deuterated internal standard, absorbed on
to silica gel, and extracted with ethyl acetate. Analytes
were derivatised with heptafluorobutyric acid anhy-dride (HFBA). Calibration curves were established by
spiking blank rat blood or urine with different
concentrations of 3-MCPD and 3-MCPD-d5.
Precision, recovery, and accuracy were determined.
Blood samples were analysed in a similar way: water
and internal standard were added to blood mixed with
EDTA and samples were processed by addition of
silica gel as for urinary samples.In the recent 90-day toxicological study (EFSA
2011), urines were analysed directly for free 3-MCPD,
�-chlorolactic acid, and DHPMA by LC-MS/MS using
ionisation in negative ion mode and detection in
selected reaction mode (SRM). Samples were diluted1:100 or 1:1000, spiked with internal standards (3-
MCPD-d5 and S-DHPMA-d5), and injected. Limits of
detection were 1.73mg l�1 for 3-MCPD, 0.71 mg l�1 forS-DHPMA, and 9 mg l�1 for �-chlorolactic acid.
Eckert et al. (2010) recently reported the develop-
ment of a quantitative analytical method for the
mercapturic acid metabolite of glycidol (and 3-
MCPD) and DHPMA in human urine using hydro-
philic interaction liquid chromatography with tandem
mass spectrometry in negative electrospray ionisation
mode (HILIC-ESI-MS/MS). Urine samples were acid-
ified to a pH of 2.5 using ammonium formate buffer
and formic acid and internal standard (13C-DHPMA)
was added. Samples were processed by SPE. Afterelution, extracts were dried, reconstituted in solvent
and injected on to the LC-MS/MS. Two mass transi-
tions were used as quantifier and qualifier. Precision,
recovery, accuracy and the LOD for DHPMA
(5.5 mg l�1) were determined. The methods were
designed to study urinary metabolites of glycidol and
other alkylating chemicals in occupational settings and
in the general population (Eckert et al. 2011). Since
DHPMA is a common metabolite of 3-MCPD (and 2-
MCPD) and glycidol as well as their esters (and other
compounds, e.g. epichlorohydrin), it is not possible to
conclude which compounds humans were exposed to
without concurrent analysis of more specific bio-
markers of exposure or the levels of individual com-pounds in foods that were ingested.
28 C. Crews et al.
MCPD and glycidyl esters
To date only indirect methods have been available forthe determination of 3-MCPD esters in biologicalsamples.
3-MCPD esters in fresh milk (goat, human)
The first study to analyse 3-MCPD esters in a‘‘biological sample’’ dates back to 1984, whenCerbulis et al. (1984) described a class of compoundsseparated from the triacylglycerol fraction from freshnon-processed goat milk by TLC of the lipid fractionextracted by petroleum ether. It was not found inpeanut, corn, sunflower or safflower oils. They intro-duced the procedure of the release of free 3-MCPD bytransesterification overnight with methanolic hydro-chloric acid at room temperature followed by GC orTLC on silica gel G. 3-MCPD was identified by GC-MS. Alternatively, transesterification was carried outwith sulphuric acid as a catalyst. GC analyses of themethyl esters obtained from the transesterificationshowed that the major fatty acids present in 3-MCPDesters from goat’s milk were capric (C10:0), lauric(C12:0), palmitic (C14:0), stearic (C18:0), and oleicacid (C18:1), similar to the profile of goat milk TAG.The authors later identified 3-MCPD di-esters inhuman milk, but not in bovine milk or butter(Kuksis et al. 1986). As discussed above, this method(acid-catalysed transesterification) seems to be robustand specific, and the risk of interference from glycidylesters is considered low.
Zelinkova et al. (2008) analysed a series of humanbreast milk samples using GC-MS using SIM. Milk fatwas separated by extraction with potassium oxalate,ethanol and hexane-diethyl ether and re-extracted withethanol/hexane-diethyl ether. The extract was twicepartitioned with water, and the organic phase driedover anhydrous sodium sulphate, before dissolution inhexane. To analyse 3-MCPD di-esters, the milk fat wasspiked after extraction with 3-MCPD-d5-dipalmitate,applied to a silica gel column, washed with lightpetroleum ether, and eluted with diethyl ether. Thedried eluate containing 3-MCPD di-esters was re-dissolved in tetrahydrofuran (THF) and subjected toGC-MS. The method of Zelinkova et al. (2006) wasused to determine bound and free 3-MCPD by GC-MS. For free 3-MCPD the milk fat was extracted withhexane-acetone and derivatised with PBA for GC-MSdetermination. For bound 3-MCPD, the milk fat wasdissolved in THF, transesterified with sulphuric acid/methanol, neutralised with sodium bicarbonate, spikedwith 3-MCPD-d5 and derivatised with PBA. The LODwas determined to be 100 mg kg�1 milk fat with arelative standard deviation (RSD) of 5.9% and satis-factory linearity (r2¼ 0.9998). The calculated levels ofbound 3-MCPD in the milk samples ranged from 511to 76 mg kg�1. Di-esters prevailed over mono-esters,
and the fatty acids determined represented thoseusually found in human milk (lauric, palmitic, oleic).Based on these results, it is not possible to conclude towhich compounds individuals were exposed (esters orfree 3-MCPD). Concurrent analysis of free and boundMCPD and glycidol in the feed and food (andpotentially also in urine), respectively, would berequired. The fact that both in the goat and thehuman milk studies the fatty acid profile of bound 3-MCPD is similar to the endogenous milk fatty acidprofile may point to an endogenous (re-)synthesis ofthe fatty acid esters of 3-MCPD.
Free and bound 3-MCPD in cell culture supernatants
Parallel determination of free 3-MCPD and ester-bound 3-MCPD has been carried out in a recentin vitro study using the Caco-2 cell line as a model forhuman intestinal cells (Buhrke et al. 2011). In thisstudy, free 3-MCPD was determined from cell culturesupernatants by diluting the supernatant 10-fold with20% of sodium chloride in 20mM sodium phosphatebuffer at pH 6.0 and subsequent GC-MS analysisaccording to the protocol of Divinova et al. (2004)using 3-MCPD-d5 as internal standard. For theanalysis of ester-bound 3-MCPD, cell culture super-natants were lyophilised and resuspended in a mixtureof t-butylmethylether and ethyl acetate (80/20 v/v).This solution was used for ester cleavage and subse-quent 3-MCPD determination according to option Aof the later withdrawn DGF Standard Method CIII18b (DGF 2009) as described above. For analysis of 3-MCPD esters in biological samples a modification ofthe method would be desirable to comply with actualdevelopments.
Glycidyl esters
Comparable with the analysis of 3-MCPD esters,methods are available to determine glycidyl estersindirectly after ester cleavage and subsequent determi-nation of free glycidol. However, due to the highreactivity of the epoxide it is not simple to determineglycidol or glycidyl esters directly in biological sam-ples, because these substances can react with numerouscompounds present in biological systems. Therefore,indirect methods such as the determination of the urinemetabolite DHPMA (see above) or of haemoglobinadducts (see below) are employed for the estimation ofglycidol or glycidyl ester concentrations in biologicalsamples. Current improvements of indirect analyticalmethods minimise the conversion between glycidol and3-MCPD so that indirect methods should becomeapplicable in the near future for biological samples aswell. Otherwise, direct methods for the determinationof 3-MCPD and glycidyl esters omitting the step of
Food Additives & Contaminants: Part A 29
ester cleavage that are currently being developed couldbe applied to biological samples.
Biomarkers of exposure – haemoglobin adducts
As mentioned above, glycidol can hardly be directlydetected in biological systems due to its high reactivity.Glycidol belongs to the group of alkylating agents andcan therefore covalently bind to numerous biomole-cules such as proteins or DNA. As an example,glycidol can react with the N-terminal valine residueof haemoglobin. The resulting glycidol–haemoglobinadduct can be used as a biomarker to monitor theburden of glycidol in human blood. The generalprocedure to analyse haemoglobin adducts of alkylat-ing agents, such as ethylene oxide or 1,3-butadiene, hasbeen reviewed by Boogaard (2002). A specific methodfor the determination of haemoglobin adducts ofepichlorohydrin that are identical to the ones formedfrom glycidol has been published by the group ofLandin et al. (1996). Briefly, erythrocytes were isolatedfrom blood samples and subsequently the globinfraction was isolated from the erythrocytes. Afteraddition of the deuterated internal standard, a mod-ified Edman degradation procedure was carried out inthe presence of pentafluorophenyl isothiocyanate forderivatisation of N-(2,3-dihydroxypropyl)valine, whichis the product of glycidol covalently bound to theN-terminal valine residue of haemoglobin. The result-ing pentafluorophenylthiohydantoin is extracted andfurther modified to block the remaining hydroxylfunctions. Finally, the extract is analysed via GC-MSusing negative chemical ionisation (NCI), and theamount of the glycidol–haemoglobin adduct is calcu-lated from the derivatised Edman degradation productof the sample in relation to the corresponding Edmandegradation product of the deuterated internalstandard.
A recent study investigated the levels of bloodglycidol haemoglobin adducts (using the above men-tioned method by Landin et al. 1996) in humans whoconsumed DAG edible oils containing small amountsof glycidyl esters (Honda et al. 2011) over a period offour months. Background adducts were measurable inall, but no increased blood haemoglobin adducts weredetected in DAG oil consumers in comparison to non-consumers. On the contrary, background levels wereslightly higher in non-consumers. On the other hand, inthe study by Landin et al. (1996) smokers had higherlevels compared with non-smokers. However, the GElevels in DAG oils consumed were low, the sample sizewas small, and other dietary factors or potential othersources of glycidol haemoglobin adducts were notcontrolled.
As with many other biomarkers of exposure, thequantitative relationship between the biomarker and
the actual exposure is not straightforward.For example, erythrocytes are rather long-lived cells,therefore, the occurrence of haemoglobin adductsrepresents an average exposure over the full lifetimeof the cells, which is approximately 120 days.In addition, glycidol haemoglobin adducts are notspecific for the exposure to glycidol or its esters, butmight be formed through other compounds, such asepichlorohydrin. As long as the qualitative andquantitative relationship between the compound thatan organism is exposed to and a potential biomarker ofexposure is not known, its applicability has certainlylimitations.
Gaps and limitations
Up to now, analytical methods to determine 3-MCPDand glycidyl esters have not been widely applied tosamples obtained from biological studies, and it is notyet established if the methods currently available foroils and foods are equally applicable to biologicalsamples, e.g. if the sample sizes obtained from animalstudies are quantitatively sufficient, or if the extractionmethods developed for oil or food matrices aretransferable to urine, blood, or tissues. The methodsemerging now as reliable and specific, be they direct orindirect, should be applied to biological samples to testtheir performance in these matrices.
Because MCPD, glycidol and their esters arepotentially quickly metabolised in the organism, itmay not be sufficient to determine the levels of theparent compounds in biological samples, but relevantmetabolites or biomarkers of exposure should beidentified. These metabolites or biomarkers shouldnot only reflect the actual exposure, but shouldoptimally also be connected to the putative adverseeffects relevant for human exposure through the diet.Mechanistic studies that are aimed at characterisingthe kinetics of the release of 3-MCPD or glycidol fromtheir esters must, therefore, be sufficiently specific.
Furthermore, the identification of analytical meth-ods for potential biomarkers in humans should alsotake into account that humans are likely exposed tomixtures of these compounds in foods. Refined oilshave been shown to contain esters of glycidol, 3-MCPD, and 2-MCPD at the same time. Many otherfood sources are more likely to contain the freecompounds, so the potential use of a metabolite thatis common to all of these compounds, such asDHPMA, as a biomarker of human exposure mightnot be adequate. The requirements for a relevanthuman biomarker are, however, far beyond the scopeof this paper and will not be discussed further.
The relevant ‘‘biological samples’’ (organs, bodyfluids) suitable to be analysed for specific purposes(mechanistic study, monitoring exposure, etc.) needs to
30 C. Crews et al.
be identified. Relevant metabolites and specific bio-markers of exposure need to be identified.
The appropriate analytical methods for thesemetabolites and biomarkers, to be applied to thebiological samples need to be developed and validated.Testing is needed of the applicability of direct andindirect methods for MCPD, glycidol and their estersin biological samples (sensitivity, specificity, etc.).Sample sizes (the quantity available from animalstudy or from humans) need to be determined, andextraction techniques for different tissues and bodyfluids to be designed to reduce matrix effects.
Consideration should be given to the study of theexposure to mixtures of MCPD, glycidol, their estersand resulting metabolites. Recent animal studies in ratsincluding that by EFSA (2011) will help to furtherrefine remaining questions and to narrow down andprioritise future research needs.
Conclusions
Research activity in this field over recent years hasbeen both intensive and productive. Comprehensivestudies of the reactions involved in the indirect analysisof MCPD esters have increased our understanding ofthe mechanisms involved, but many variations havebeen proposed and the need for harmonisation hasbecome apparent. Indirect methods for both boundMCPD and bound glycidol largely meet the require-ments for quick and routine analysis, and so indirectmethods are likely to be the ones used for routinescreening and food surveys. It can be consideredprudent that the wide scale application of suchmethods for this purpose has awaited better charac-terisation and validation. Indirect methods that pro-vide data on the content of both 2- and 3-MCPDesters, and hopefully glycidyl esters, are required andwill need validation by collaborative trial.
Direct methods for MCPD esters are likely to beused in a less routine fashion and the variability inperformance between different instruments and meth-ods means that a range of methods is likely to remain.These methods provide detailed information for for-mation and metabolism studies. The difficulties in thechromatographic separation of 2-MCPD esters from 3-MCPD esters in the direct methods remain to be dealtwith. For glycidyl esters the small number of esters andthe uncertainty of their analysis by the indirectmethods means that direct methods will continue tobe prominent. The growing commercial availability ofreference standards will benefit the scope and perfor-mance of the direct methods, but for both direct andindirect approaches there is a strong requirement forcharacterised reference materials and inter-laboratoryvalidation of one/two direct analytical method forglycidyl esters which include isotopically labelled
internal standards. The method(s) may be thoroughlycompared with indirect analytical methods such as the‘‘SGS 3-in-1 method,’’ which have already shown to becorrelated closely to the direct analytical methodsavailable. This may give support to laboratories whichare not equipped with highly sophisticatedinstruments.
The variation of results and doubts about methodperformance have to an extent been responsible for thelack of large-scale surveys of occurrence of MCPDesters and glycidyl esters. Much occurrence data havebeen derived from method testing experiments andstudies of formation. More data are required for onoils other than palm, and for composite foodstuffs and,where possible, this must be related to the processingconditions that affect their formation, in particularedible oil refining. It is very likely that the wide rangesof results reported truly represent a wide range ofconcentrations present in the foods analysed, and interms of using the data tabulated, the differences in theresults due to, for example, the presence or absence ofglycidyl esters is comparatively insignificant. Data of ahigher quality will undoubtedly be produced oncereliable analytical methods have been validated byinternational collaborative trials and applied to well-planned and -executed surveys.
Metabolic studies have shown that MCPD, glycidoland their esters are potentially quickly metabolised inthe organism. Both 3-MCPD and its dipalmitateexhibit nephrotoxicity and testicular toxicity, with theesters having milder effects than free 3-MCPD. Furtherstudies on the relevant metabolites and biomarkers ofexposure are required to calculate actual exposure andto characterise the kinetics of 3-MCPD and glycidolrelease from their esters. Monitoring of metabolitescommon to MCPD, glycidol and their esters might notgive sufficient information.
Very promising advances have been made towardsmitigation of MCPD ester and glycidyl ester formationin palm oil. Studies of all stages of palm oil productionhave indicated where contaminant formation can bereduced. It was considered that a multifacetedapproach might be beneficial, beginning with a reduc-tion of chloride application in fertiliser, continuingwith the use of plant varieties low in partial glycerideprecursors, the selection of young fruit of good quality,ending with technical changes to the refining steps, andthe removal by use of inorganic adsorbents.
The determination of 3-MCPD and glycidyl estersin biological studies using methods akin to those usedfor food is being investigated with caution. They havepromise but will need to be associated with methods todetermine the relevant biomarkers and metabolites.
It is over 30 years since the discovery of 3-MCPDand its esters in hydrolysed vegetable proteins, and theattention given to the free chloropropanol has in recentyears been superseded by interest in its esters in refined
Food Additives & Contaminants: Part A 31
vegetable oils. Dramatic reduction of the levels of freeMCPD in hydrolysed vegetable protein was achievedby close cooperation of the food industry withgovernment food safety agencies and research scien-tists, and the current repetition of this cooperation willhopefully lead to an effective mitigation of thisproblem.
Acknowledgements
This work was commissioned by the Process-relatedCompounds & Natural Toxins Task Force and RiskAssessment of Chemicals in Food Task Force of theEuropean branch of the International Life SciencesInstitute (ILSI Europe). Industry members of the taskforces are Ajinomoto Europe, Bunge Europe, Cargill,Danone, DSM, Kikkoman Foods Europe, Kraft FoodsEurope, Luigi Lavazza, Mars, Nestle, Pepsico International,Premier Foods, Procter & Gamble, Soremartec Italia –Ferrero Group, and Unilever. This review wascoordinated by Dr Alessandro Chiodini and Dr PratimaRao Jasti, Scientific Project Managers at ILSI Europe.For further information about ILSI Europe, pleaseemail: info@ilsieurope.be or telephone þ32 2 771 00 14.
Note
The opinions expressed herein and the conclusions of thisreview are those of the authors and do not necessarilyrepresent the views of ILSI Europe or those of its membercompanies. Colin Crews, Michael Granvogl, Jan Kuhlmann,Alfonso Lampen and Rudiger Weisshaar received an hono-rarium from ILSI Europe for their participation in thisreview and reimbursement of their travel and accommoda-tion costs for attending the related meetings. AlessandroChiodini and Pratima Rao Jasti are employed by ILSIEurope.
References
Bakhiya N, Abraham K, Gurtler R, Appel KE, Lampen A.2009. Toxicological assessment of 3-chloropropane-1,2-
diol and glycidol fatty acid esters in food. Mol Nutr Food
Res. 55(4):509–521.Berger-Preiss E, Gerling S, Apel E, Lampen A,
Creutzenberg O. 2010. Development and validation of an
analytical method for determination of 3-chloropropane-1,2-diol in rat blood and urine by gas chromatography-
mass spectrometry in negative chemical ionization mode.
Anal Bioanal Chem. 398:313–318.Blumhorst MR, Venkitasubramanian P, Collison MW. 2011.
Direct determination of glycidyl esters of fatty acids in
vegetable oils by LC-MS. J Am Oil Chem Soc.88:1275–1283.
Boogaard PJ. 2002. Use of haemoglobin adducts in exposuremonitoring and risk assessment. J Chromatogr B.
778:309–322.Buhrke T, Weisshaar R, Lampen A. 2011. Absorption and
metabolism of the food contaminant 3-chloro-1,2-propa-
nediol (3-MCPD) and its fatty acid esters by human
intestinal Caco-2 cells. Arch Toxicol. 85(10):1201–1208.
Bundesinstitut fur Risikobewertung (BfR). 2009. Initial
evaluation of the assessment of levels of glycidol fatty
acid esters detected in vegetable fats [Internet]. 2009. Berlin
(Germany): BfR. Available from: http://www.bfr.bund.de/
cm/349/initial_evaluation_of_the_assessment_of_levels_
of_glycidol_fatty_acid_esters.pdf/Cerbulis J, Parks OW, Liu RH, Piotrowski EG, Farrell HM.
1984. Occurrence of diesters of 3-chloro-1,2-propanediol in
the neutral lipid fraction of goats’ milk. J Agric Food
Chem. 32:474–476.Collison MW, Blumhorst M. 2011. Advances in LC-MS
analysis of glycidyl esters. Paper presented at: 102nd
AOCS Annual Meeting & Expo; 1–4 May 2011;
Cincinnati, OH, USA.
Craft BD, Chiodini A, Garst J, Granvogl M.
Forthcoming 2012. Fatty acid esters of monochloropro-
panediol (MCPD) and glycidol factors affecting the
formation of MCPD esters and glycidyl esters in refined
edible oils. Food Addit Contam A. DOI: 10.1080/
19440049.2012.709196.Davıdek J, Velısek J, Kubelka V, Janıeek G. 1982.
New chlorine containing organic compounds in protein
hydrolysates. Paper presented at: EuroFood Chem 1;
17–20 February 1981; Vienna, Austria.Davıdek J, Velısek J, Kubelka V, Janıeek G, Simicova Z.
1980. Glycerol chlorohydrins and their esters as products
of the hydrolysis of tripalmitin, tristearin and triolein with
hydrochloric acid. Z Lebensm Unters Forsch. 171:14–17.
De Rooij BM, Boogard PJ, Commandeur JNM,
Vermeulen NPE. 1997. 3-Chloro-2-hydroxypropylmercap-
turic acid and alpha-chlorohydrin as biomarkers ofoccupational exposure to epichlorohydrin. Environ
Toxicol Pharmacol. 3:175–185.
De Rooij BM, Commandeur JN, Ramcharan JR,
Schuilenburg HC, van Baar BL, Vermeulen NP. 1996.
Identification and quantitative determination of 3-chloro-
2-hydroxypropylmercapturic acid and alpha-chlorohydrin
in urine of rats treated with epichlorohydrin. J
Chromatogr B. 685:241–250.Deutsche Gesellschaft fur Fettwissenschaft (DGF). 2009.
Deutsche Gesellschaft fur Fettwissenschaf: DGF Standard
Method C III 18 (09) – Determination of ester-bound 3-
chloropropane-1,2-diol (3-MCPD esters) and 3-MCPD
forming substances in fats and oils by means of GC-MS.
Deutsche Einheitsmethoden zur Untersuchung von
Fetten, Fettprodukten, Tensiden und verwandten
Stoffen. Stuttgart (Germany): Wissenschaftliche
Verlagsgesellschaft. Available from: http://www.dgfett.de/
methods/Deutsche Gesellschaft fur Fettwissenschaft (DGF). 2011a.
Deutsche Gesellschaft fur Fettwissenschaft: DGF
Standard Method C-VI 18 (10) Fatty-acid-bound 3-
chloropropane-1,2-diol (3-MCPD) and 2,3-epoxipropane-
1-ol (glycidol). Determination in oils and fats by GC/MS
(Differential measurement). Deutsche Einheitsmethoden
zur Untersuchung von Fetten, Fettprodukten, Tensiden
und verwandten Stoffen. Available from: http://
www.dgfett.de/methods/c_vi_18_%2810%29-english.pdf/Deutsche Gesellschaft fur Fettwissenschaft (DGF). 2011b.
Deutsche Gesellschaft fur Fettwissenschaft, DGF
Standard Method C-VI 17 (10) Ester-bound 3-chloropro-
pane-1,2-diol (3-MCPD ester) and glycidol (glycidyl ester).
32 C. Crews et al.
Summation method for the determination in fats and
oils by GC-MS. Deutsche Einheitsmethoden zur
Untersuchung von Fetten, Fettprodukten, Tensiden und
verwandten Stoffen. Available from: http://www.dgfett.de/
methods/c_vi_17_%2810%29_english.pdf/Divinova V, Dolezal M, Velısek J. 2007. Free and bound
3-chloropropane-1,2-diol in coffee surrogates and malts.
Czech J Food Sci. 25:39–47.Divinova V, Svejkovska B, Dolezal M, Velısek J. 2004.
Determination of free and bound 3-chloropropane-1,2-diol
by gas chromatography with mass spectrometric detection
using deuterated 3-chloropropane-1,2-diol as internal
standard. Czech J Food Sci. 22:182–189.
Dolezal M, Chaloupska M, Divinova M, Svejkovska B,
Velısek J. 2005. Occurrence of 3-chloropropane-1,2-diol
and its esters in coffee. Eur Food Res Technol.
221:221–225.Dolezal M, Ilko V, Zelinkova Z, Velısek J. 2009. 3-MCPD
and its esters in malts and roasted cereals. Paper presented
at: EURO FOOD CHEM XV, Food for the Future – The
Contribution of Chemistry to Improvement of Food
Quality; 5–8 July 2009; Copenhagen, Denmark.Dolezal M, Zelinkova Z, Velısek J. 2008. Analysis of potato
product lipids for 3-MCPD esters. Paper presented at: 6th
Euro Fed Lipid Congress; 7–10 September 2008; Athens,
Greece.Dubois M. 2011. Analysis of MCPD and glycidyl-esters in
edible oils. Paper presented at: 102nd AOCS Annual
Meeting & Expo; 1–4 May 2011; Cincinnati, OH, USA.Dubois M, Tarres A, Goldmann T, Empl AM,
Donaubauer A, Seefelder W. 2012. Comparison of indirect
and direct quantification of esters of monochloropropa-
nediol in vegetable oil. J Chromatogr A. 1236:189–201.
Dubois M, Tarres A, Goldmann T, Loeffelmann G,
Donaubauer A, Seefelder W. 2011. Determination of
seven glycidyl esters in edible oils by gel permeation
chromatography extraction and liquid chromatography
coupled to mass spectrometry detection. J Agric Food
Chem. 59(23):12291–12301.Eckert E, Drexler H, Goen T. 2010. Determination of six
hydroxyalkyl mercapturic acids in human urine using
hydrophilic interaction liquid chromatography with
tandem mass spectrometry (HILIC-ESI-MS/MS).
J Chromatogr B. 878:2506–2514.Eckert E, Schmid K, Schaller B, Hiddemann-Koca K,
Drexler H, Goen T. 2011. Mercapturic acids as metabolites
of alkylating substances in urine samples of German
inhabitants. Int J Hyg Environ Health. 214(3):196–204.Ermacora A, Hrncirık K. 2012a. Evaluation of an improved
indirect method for the analysis of 3-MCPD esters based
on acid transesterification. J Am Oil Chem Soc.
89:211–217.
Ermacora A, Hrncirık K. 2012b. Indirect determination of
2-/3-MCPD and glycidyl esters in oils/fats: a new method
based on acid transesterification. Paper presented at:
102nd AOCS Annual Meeting & Expo; 29 April–2 May
2012; Long Beach, CA, USA.
European Commission. 2001a. Opinion of the Scientific
Committee on Food on 3-monochloro-propane-1,2-diol
(3-MCPD) updating the SCF opinion of 1994 adopted on
30 May 2001. Available from: http://ec.europa.eu/food/fs/sc/scf/out91_en.pdf/
European Commission. 2001b. Commission Regulation (EC)
No. 466/2001 of 8 March 2001 setting maximum levels for
certain contaminants in foodstuffs. Off J Eur Comm.
L77(1):1–13.European Food Safety Authority (EFSA) (prepared by
Barocelli E, Corradi A, Mutti A, Petronini PG). 2011.
Comparison between 3-MCPD and its palmitic esters in a
90-day toxicological study. Parma (Italy): University of
Parma. Scientific report submitted to EFSA CFP/EFSA/
CONTAM/2009/01.Fiebig HJ. 2011. Determination of ester-bound 3-chloro-1,2-
propanediol and glycidol in fats and oils – a collaborative
study. Eur J Lipid Sci Technol. 113:393–399.Food Analysis Performance Assessment Scheme (FAPAS).
2011. Proficiency Test nr. 2631. Available from: http://
www.fapas.com/
FSA. 2010. Survey of process contaminants in retail foods –
Year 3: 2009. Food Standards Agency Food Survey
Information Sheet 03/10. Available from: http://
www.food.gov.uk/multimedia/pdfs/fsis0310.pdf/Gardner A, Yurawecz M, Cunningham W, Diachenko G,
Mazzola E, Brumley W. 1983. Isolation and identification
of C16 and C18 fatty acid esters of chloropropanediol in
adulterated Spanish cooking oils. Bull Environ Contam
Toxicol. 31:625–630.
Granvogl M, Schieberle P. 2011a. Quantitation of glycidyl
esters via stable isotope dilution analysis. Paper presented
at: 102nd AOCS Annual Meeting & Expo; 1–4 May 2011;
Cincinnati, OH, USA. Available from: http://www.aocs.
org/files/ResourcesPDF/3%5FAOCS%5FCincinnati%5
F030511.pdf/Granvogl M, Schieberle P. 2011b. Development of direct
quantification method of glycidyl esters in edible fats and
oils via stable isotope dilution analysis. Paper presented at
the final conference of the FEI project; 21 November 2011;
Berlin, Germany. Available from: http://www.bll.de/
themen/kontaminanten/abschlussveranstaltung-3-mcpd/
Habermeyer M, Guth S, Eisenbrand G. 2011. Identification
of gaps in knowledge concerning toxicology of 3-MCPD
and glycidol esters. Eur J Lipid Sci Technol. 113:314–318.Haines TD, Adlaf KJ, Pierceall RM, Lee I,
Venkitasubramanian P, Collison MW. 2011. Direct deter-
mination of MCPD fatty acid esters and glycidyl fatty acid
esters in vegetable oils by LC-TOFMS. J Am Oil Chem
Soc. 88:1–14.
Hamlet CG, Asuncion L, Velısek J, Dolezal M, Zelinkova Z,
Crews C. 2011. Formation and occurrence of esters of 3-
chloropropane-1,2-diol (3-CPD) in foods: what we know
and what we assume. Eur J Lipid Sci Technol.
113:279–303.Hamlet CG, Sadd PA. 2004. Chloropropanols and their
esters in cereal products. Czech J Food Sci. 22:259–262.
Hamlet CG, Sadd PA, Gray DA. 2004. Chloropropanols and
their esters in baked cereal products. Paper presented at:
American Chemical Society Division of Agricultural and
Food Chemistry 227th annual meeting; 28 March–1 April
2004; Anaheim, CA, USA.Hinrichsen N. 2011. 3-MCPD and glycidyl esters: a perspec-
tive. Paper presented at: OVID/BLL conference; 18
January 2011; Berlin, Germany.Honda H, Onishi M, Fujii K, Ikeda N, Yamaguchi T,
Fujimori T, Nishiyama N, Kasamatsu T. 2011.
Food Additives & Contaminants: Part A 33
Measurement of glycidol hemoglobin adducts in humans
who ingest edible oil containing small amounts of
glycidol fatty acid esters. Food Chem Toxicol.
49(10):2536–2540.Hori K, Koriyama N, Omori H, Kuriyama M, Arishima T,
Tsumura K. 2012. Simultaneous determination of
3-MCPD fatty acid esters and glycidol fatty acid esters in
edible oils using liquid chromatography time-of-flight
mass spectrometry. LWT – Food Sci Technol. 48(2):
204–208.Hrncirık K, Zelinkova Z, Ermacora A. 2011. Critical factors
of indirect determination of 3-chloropropane-1,2-diol
esters. Eur J Lipid Sci Technol. 113:361–367.International Agency for Research on Cancer (IARC). 2000.
International Agency for Research on Cancer
Monographs on the Evaluation of Carcinogenic
Risks to Humans. Vol. 77: some industrial chemicals.
Summary of data reported and evaluation. Available
from: http://monographs.iarc.fr/ENG/Monographs/vol77/
volume77.pdf/Jones AR. 1973. The metabolism of biological alkylating
agents. Drug Metab Rev. 2:71–100.Jones AR. 1975. The metabolism of 3-chloro-, 3-bromo- and
3-iodopropan-1,2-diol in rats and mice. Xenobiotica.
5:155–165.Jones AR. 1983. Antifertility actions of alpha-chlorohydrin
in the male. Aust J Biol Sci. 36:333–350.Jones AR, Fakhouri G. 1979. Epoxides as obligatory
intermediates in the metabolism of alpha-halohydrins.
Xenobiotica. 9:595–599.
Jones AR, Milton DH, Murcott C. 1978. The oxidative
metabolism of alpha-chlorohydrin in the male rat and the
formation of spermatocoeles. Xenobiotica. 8:573–582.Karasek L, Wenzl T, Ulberth F. 2010. Proficiency test
on the determination of 3-MCPD esters in edible oil
[Internet]. Geel (Belgium): Publications Office of the
European Union. EUR 24356 EN (DOI 10.2787/2587)
Available from: http://irmm.jrc.ec.europa.eu/interlaboratory_
comparisons/3_MCPD/Documents/eur_24356_en_3-mpcd_
esters_in_edible_oil.pdf/Karsulınova L, Folprechtova B, Dolezal M, Dostalova J,
Velısek J. 2007. Analysis of the lipid fractions of coffee
creamers, cream aerosols, and bouillon cubes for their
health risk associated constituents. Czech J Food Sci.
25:257–264.Kuhlmann J. 2008. Uberbefunde bei der Bestimmung von 3-
MCPD-Estern in Olen & Fetten? Mogliche Ursachen und
Konsequenzen. Paper presented at: 2nd Workshop on
Analysis of 3-MCPD-esters in Edible Oils of Federal
Institute for Risk Assessment; November 2008; BfR,
Berlin (Germany).Kuhlmann J. 2011a. Determination of bound 2,3-epoxy-1-
propanol (glycidol) and bound monochloropropanediol
(MCPD) in refined oils. Eur J Lipid Sci Technol.
113:335–344.Kuhlmann J. 2011b. Indirect determination of bound
glycidol and MCPD in refined oils. Paper presented at:
102nd AOCS Annual Meeting & Expo; May 1–4 2011;
Cincinnati, OH, USA.
Kuksis A, Marai L, Myher JJ, Cerbulis J, Farrell Jr HM.
1986. Comparative study of the molecular species of
chloropropanediol diesters and triacylglycerols in milk fat.
Lipids. 21:183–190.Kusters M, Bimber U, Reeser S, Gallitzendorfer R,
Gerhartz M. 2011. Simultaneous determination and dif-
ferentiation of glycidyl esters and 3-monochloropropane-
1,2-diol (MCPD) esters in different foodstuffs by GC-MS.
J Agric Food Chem. 59(11):6263–6270.
Landin HH, Osterman-Golkar S, Zorcec V, Tornquist M.
1996. Biomonitoring of epichlorohydrin by hemoglobin
adducts. Anal Biochem. 240:1–6.Lebensmittelchemisches Institut des Bundesverbandes der
Deutschen Sußwarenindustrie (LCI). 2008. Wie viele
3-MCPD-Ester gibt es? – Ein Rechenmodell. Cologne
(Germany): Lebensmittel Chemisches Institute des
Bundesverbandes der Deutschen Sußwarenindustrie e. V.
Available from: http://www.lci-koeln.de/deutsch/veroef-
fentlichungen/lci-focus/wieviele-3-mcpd-ester-gibt-es-ein-
rechenmodell/Lynch BS, Bryant DW, Hook GJ, Nestmann ER, Munro IC.
1998. Carcinogenicity of monochloro-1,2-propanediol
(a-chlorohydrin, 3-MCPD). Int J Tox. 17:47–76.MacMahon S, Begley T, Diachenko G. 2011. LC-MS/MS
detection of glycidyl esters and 3-MCPD esters in edible
oils. Paper presented at: 102nd AOCS Annual Meeting &
Expo; 1–4 May 2011; Cincinnati, OH, USA.
Masukawa Y, Shiro H, Kondo N, Kudo N. 2011.
Generalized method to quantify glycidol fatty acid esters
in edible oils. J Am Oil Chem Soc. 88:15–21.Masukawa Y, Shiro H, Nakamura S, Kondo N, Jin N,
Suzuki N, Ooi N, Kudo N. 2010. A new analytical method
for the quantification of glycidol fatty acid esters in edible
oils. J Oleo Sci. 59:81–88.
Miyazaki K, Koyama K, Sasako H, Hirao T. 2012. Indirect
method for simultaneous determinations of 3-chloro-1,2-
propanediol fatty acid esters and glycidyl fatty acid esters.
J Am Oil Chem Soc. 89:1403–1407.Moravcova E, Vaclavik L, Lacina O, Hrbek V,
Riddellova K, Hajslova J. 2012. Novel approaches to
analysis of 3-chloropropane-1,2-diol esters in vegetable
oils. Anal Bioanal Chem. 402:2871–2888.National Toxicology Program (NTP). 1990. Toxicology and
carcinogenesis studies of glycidol (CAS No. 556-52-5) in
F344/N rats and B6C3F1 mice (gavage studies). Technical
Report Series No. 374. Available from: http://ntp.niehs.
nih.gov/ntp/htdocs/LT_rpts/tr374.pdf/National Toxicology Program (NTP). 2007. Toxicology and
carcinogenesis study of glycidol (CAS No. 556-52-5) in
genetically modified haploinsufficient p16(Ink4a)/p19(Arf)
mice (gavage study). Natl Toxicol Program Genet Modif
Model Rep. 13:1–81.Nomeir AA, Silveira DM, Ferrala NF, Markham PM,
McComish MF, Ghanayem BI, Chadwick M. 1995.
Comparative disposition of 2,3-epoxy-1-propanol (gly-
cidol) in rats following oral and intravenous administra-
tion. J Toxicol Environ Health. 44:203–217.Pinkston JD, Stoffolano PJ. 2011. Update on the develop-
ment of a sensitive, accurate, and user-friendly method for
the direct determination of 3-MCPD esters. Paper
presented at: 102nd AOCS Annual Meeting & Expo;
May 1–4 2011; Cincinnati, OH, USA.Reece P. 2005. The origin and formation of 3-MCPD in
foods and food ingredients. Available from: http://
34 C. Crews et al.
www.foodbase.org.uk//admintools/reportdocuments/43_84_FINAL_REPORT.pdf/
Robert MC, Oberson JM, Stadler RH. 2004. Model studieson the formation of monochloropropanediols in thepresence of lipase. J Agric Food Chem. 52:5102–5108.
Schilter B, Scholz G, Seefelder W. 2011. Fatty acid esters of
chloropropanols and related compounds in food: toxico-logical aspects. Eur J Lipid Sci Techol. 113:309–313.
Seefelder W, Varga N, Studer A, Williamson G, Scanlan FP,
Stadler RH. 2008. Esters of 3-chloro-1,2-propanediol(3-MCPD) in vegetable oils: significance in the formationof 3-MCPD. Food Addit Contam A. 25(4):391–400.
Shimizu M, Kudo N, Shiro H, Yasunaga K, Masukawa Y,Katsuragi Y, Yasumasu T. 2011. Comparison of indirectand direct quantification of glycidol fatty acid ester inedible oils. J Oleo Sci. 59(10):535–539.
Shiro H, Kondo N, Kibune N, Masukawa Y. 2011. Directmethod for quantification of glycidol fatty acid esters inedible oils. Eur J Lipid Sci Technol. 113:356–360.
Shiro H, Kondo N, Masukawa Y. 2011. Validation ofquantitative method for glycidol fatty acid esters (GEs) inedible oils. Paper presented at: 102nd AOCS Annual
Meeting & Expo; 1–4 May 2011; Cincinnati, OH, USA.Svejkovska B, Novotny O, Divinova M, Reblova Z,Dolezal M, Velısek J. 2004. Esters of 3-chloropropane-
1,2-diol in foodstuffs. Czech J Food Sci. 22:190–196.US Food and Drug Administration (USFDA). 2008.Guidance levels for 3-MCPD(3-chloro-1,2-propanediol) inacid-hydrolyzed protein and Asian-style sauces. Available
from: http://www.fda.gov/ICECI/ComplianceManuals/CompliancePolicyGuidanceManual/ucm074419.htm/
Van den Wijngaard AJ, Janssen DB, Witholt B. 1989.
Degradation of epichlorohydrin and halohydrins by
bacterial cultures isolated from freshwater sediment. JGen Microbiol. 135:2199–2208.
Velısek J, Davıdek J, Kubelka V, Bartosova J, Tuekova A,Hajslova J, Janıeek G. 1979. Formation of volatilechlorohydrins from glycerol (triacetin, tributyrin) andhydrochloric acid. Z Lebens Wissen Technol. 12:234–236.
Weisshaar R. 2008. Determination of total 3-chloropropane-1,2-diol (3-MCPD) in edible oils by cleavage of MCPDesters with sodium methoxide. Eur J Lipid Sci Technol.
110:183–186.Weisshaar R. 2011. Fatty acid esters of 3-MCPD: overviewof occurrence and exposure estimates. Eur J Lipid Sci
Technol. 113:304–308.Weisshaar R, Perz R. 2010. Fatty acid esters of glycidol inrefined fats and oils. Eur J Lipid Sci Technol. 112:158–165.
Zelinkova Z, Dolezal M, Velısek J. 2009a. 3-Chloropropane-
1,2-diol fatty acid esters in potato products. Czech J FoodSci. 27:421–424.
Zelinkova Z, Dolezal M, Velısek J. 2009b. Occurrence of 3-
chloropropane-1,2-diol fatty acid esters in infant and babyfoods. Eur. Food Res. Tech. 228:571–578.
Zelinkova Z, Novotny O, Schurek J, Velisek J, Hajslova J,
Dolezal M. 2008. Occurrence of 3-MCPD fatty acidesters in human breast milk. Food Addit Contam A.25:669–676.
Zelinkova Z, Novotny O, Velısek J, Dolezal M. 2007.Determination of 3-chloropropane-1,2-diol fatty acidesters. Paper presented at: EURO FOOD CHEM XIV,Food Quality and Issue of Molecule Based Science; 29–31
August 2007; Paris, France.Zelinkova Z, Svejkovska B, Dolezal M, Velisek J. 2006.Fatty acid esters of 3-chloropropane-1,2-diol in edible oils.
Food Addit Contam. 23:1290–1298.
Food Additives & Contaminants: Part A 35
top related