Scs

Qa

b

a

ARRAA

KCG(ES(H

1

p1(T(maofoe

1

w

0h

Journal of Chromatography A, 1314 (2013) 208– 215

Contents lists available at ScienceDirect

Journal of Chromatography A

j our nal homep age: www.elsev ier .com/ locate /chroma

imultaneous determination of total fatty acid esters ofhloropropanols in edible oils by gas chromatography–masspectrometry with solid-supported liquid–liquid extraction

ing Liua,1, Feng Hana,b,1, Ke Xiea,b, Hong Miaoa,∗, Yongning Wua,b,∗

Key Laboratory of Food Safety Risk Assessment, Ministry of Health; China National Center of Food Safety Risk Assessment, Beijing 100021, ChinaDepartment of Food Science and Engineering, Wuhan Polytechnic University, Wuhan 430023, China

r t i c l e i n f o

rticle history:eceived 5 March 2013eceived in revised form 1 August 2013ccepted 20 August 2013vailable online 7 September 2013

eywords:hloropropanols estersas chromatography–mass spectrometry

GC–MS)dible oilsolid-supported liquid–liquid extractionSLE)eptafluorobutyrylim idazole (HFBI)

a b s t r a c t

This study aimed to establish a novel robust method for the simultaneous determination of total fattyacid esters of 4 chloropropanols including fatty acid esters of 3-monochloropropane-1,2-diol (3-MCPDesters), 2-monochloropropane-1,3-diol (2-MCPD esters), 1,3-dichloropropan-2-ol (1,3-DCP esters) and2,3-dichloropropan-1-ol (2,3-DCP esters) in edible oils. In this method, sodium methylate in methanolwas used as the reagent for the ester cleavage reaction of chloropropanols esters. The reaction prod-ucts were extracted by a sodium sulfate solution, and then purified by solid-supported liquid–liquidextraction (SLE) using diatomaceous earth (HydromatrixTM) as the sorbent. Finally, the extracts werederivatized with heptafluorobutyrylim idazole (HFBI) and analyzed by gas chromatography–mass spec-trometry (GC–MS). Quantification was achieved using deuterated chloropropanols as their respectiveinternal standards, including 3-MCPD-d5, 2-MCPD-d5, 1,3-DCP-d5 and 2,3-DCP-d5. A good linear relation-ship between peak area and concentrations was obtained within the range of 0.025–2.000 mg L−1 witha correlation coefficients not less than 0.999 for all the chloropropanols esters. The limits of detection(LODs) of esters of 3-MCPD, 2-MCPD, 1,3-DCP and 2,3-DCP (calculated as corresponding chloropropanols)

−1

were 30, 30, 100 and 30 �g kg , respectively. The average recoveries of the 3-MCPD esters and the4 chloropropanols spiked at 0.1, 0.5 and 2 mg kg−1 into blank oil matrix were typically in a range from70.7% to 113.3%. The robust method validation data including calibration, LOD/LOQ, accuracy and repeata-bility and proficiency test results (Z-score: −0.5) of the official Food Analysis Performance Assessment

d thalorop

Scheme (FAPAS) indicatedetermination of total ch

. Introduction

3-Monochloropropane-1,2-diol (3-MCPD) and other chloro-ropanols such as 2-monochloropropane-1,3-diol (2-MCPD),,3-dichloropropan-2-ol (1,3-DCP) and 2,3-dichloropropan-1-ol2,3-DCP) have been recognized as contaminants in various food.hey were first detected in acid-hydrolyzed vegetable proteinsacid-HVP) [1], and then were also detected in bakery products,

eat and fish products, and soups [2,3]. Recently, 3-MCPD fattycid esters have been identified and could be found in a varietyf processed foods and food ingredients. They are believed to be

ormed at high temperatures during the deodorization step of theil refining process [4]. In refined oils, the concentrations of 3-MCPDsters may reach a thousand times more than the levels of free

∗ Corresponding authors at: No. 7 Panjiayuannanli, Chaoyang District, Beijing00021, China. Tel.: +86 10 67779118; fax: +86 10 67779118.

E-mail addresses: miaoh@cfsa.net.cn, miaohong0827@163.com (H. Miao),uyongning@cfsa.net.cn (Y. Wu).1 These authors contributed equally to this work.

021-9673/$ – see front matter. Crown Copyright © 2013 Published by Elsevier B.V. All rittp://dx.doi.org/10.1016/j.chroma.2013.08.074

t the present quantitative method could be successfully applied to theropanols esters in various edible oils.

Crown Copyright © 2013 Published by Elsevier B.V. All rights reserved.

3-MCPD [3,5]. Additionally, 3-MCPD esters have even beendetected in human breast milk and other foods, such as liquidseasoning or bakery foods recently [3,6]. These fatty acid estersof chloropropanols aforementioned represent the bound form of3-MCPD, from which free 3-MCPD could be released by a lipase-catalysed hydrolysis reaction [7,8]. To date, the adverse effects of3-MCPD have been concerned due to its kidney and reproductiontoxicity, as well as immune-suppression and potential carcinogeniceffects [5,9–13]. The Germany Federal Institute for Risk Assessment(BfR) made the health-risk assessment of 3-MCPD esters, based onavailable analytical method of 3-MCPD esters and by assuming that100% of the 3-MCPD is released from the 3-MCPD esters exposure[10]. It is estimated that dietary exposure of infants and adults to3-MCPD esters (calculated as 3-MCPD) was 5–20 times as much asTDI of 2 �g kg−1 body weight per day adopted by Joint FAO/WHOExpert Group on Food Additives in 2001 [14,15].

In recent years, different analytical methods were reportedfor the determination of 3-MCPD fatty acid easters. The methodsinclude either the conversion of the esters into a free forms (e.g.free 3-MCPD) for analysis (indirect methods) [16–18], or the

ghts reserved.

ogr. A 1314 (2013) 208– 215 209

dmoiofedecWmnaafttpfaIclpGm[(cftbmdmd

cp2fefp

ptstbl(coFmeo

2

2

m

Q. Liu et al. / J. Chromat

irect determination of the individual esters separately (directethods) [19–21]. The direct method, mainly with LC–MS, can

nly detect some of the individual chloropropanols esters sincet is not possible to commercially obtain the standard substancef all chloropropanol esters. However, there are many kinds ofatty acids in oil, which leading to different chloropropanolssters including mono-esters and di-easters. Therefore, it is veryifficult for the direct method to detect all the chloropropanolssters. The measurements of total chloropropanols esters areritical when estimating the health-risk of these contaminants.

ith the direct method, the health-risk of chloropropanols estersay be underestimated because some chloropropanls esters may

ot be analyzed. Compared to the direct methods, the indirectnalytical methods, which the present study applied, have obviousdvantages, such as higher sensitivity due to the detection of theree forms of chloropropanols, the need of reference standards andhe internal standards of the 4 corresponding chloropropanols forhe quantification and comprehensive analysis of all the chloro-ropanols esters and so on. More significantly, the results obtainedrom the indirect methods can be first-hand used for health-riskssessment. Therefore, the indirect methods are more practical.n most of the indirect methods, 3-MCPD esters were commonlyleavaged into free 3-MCPD by transesterification in acid or alka-ine medium [12,13,16,22,23], or hydrolyases [17], followed byurification, derivatization and quantification of free 3-MCPD byC–MS. The most prevailing indirect methods derive from the Ger-an Society for Fat Science (DGF) standard methods C-III 18 (09)

6,13,17,24–28], derivatizing 3-MCPD with phenylboronic acidPBA). However, the DGF-based methods for the determination ofhloropropanols esters may not be able to avoid the interferencerom glycidyl esters. Besides, these methods need a programmed-emperature vaporization (PTV) inlet. Otherwise it could conductaseline shift seriously, which considerably influence chro-atographic measurement efficiency. Furthermore, another

isadvantage of all these methods based on DGF C-III 18 (09) treat-ent is that the fatty acid esters of 1,3-DCP and 2,3-DCP cannot be

etected because their conversion products cannot react with PBA.Although 3-MCPD was known as the most important food

ontaminant among the four chloropropanols, the other chloro-ropanols (free and conjugated forms of 2-MCPD, 1,3-DCP and,3-DCP) should also be measured, especially as the conjugatedorms of these three chloropropanols may be present in foods. How-ver, to our best knowledge, no well-validated method is availableor simultaneously detecting all the fatty acid esters of chloro-ropanols in foods.

The aim of the present study is to develop a novel analyticalrocedure for the simultaneous determination and differentia-ion of fatty acid esters of four chloropropanols (their chemicaltructures were shown in Fig. 1a) in edible oils. This methodransforms conjugated-chloropropanols into free chloropropanolsy sodium methoxide, and efficiently purified by solid-supported

iquid–liquid extraction (SLE). N-HeptafluorobutyrylimidazoleHFBI), as the derivatization reagent, reacts with these fourhloropropanols to form detectable compounds for simultane-us GC–MS analysis. A scheme of the reaction is shown inig. 1b using 3-MCPD esters as an example. Within the scope ofethod development, critical pretreatment parameters such as

ster cleavage time and purification steps were also explored andptimized.

. Experimental

.1. Chemicals and materials

Anhydrous sodium sulfate, sodium chloride and sodiumethoxide were obtained from the Jinke Institute of Fine

Fig. 1. The chemical strictures of fatty acid of 4 chloropropanols (a) and the schemeof the reaction in this method (b).

Chemicals (Tianjin, China). Anhydrous sodium sulfate was heatedfor 8 h at 200 ◦C and then allowed to cool down and kept in a des-iccator until use. Methyl-tert-butyl ether (MTBE) was purchasedfrom Sigma–Aldrich (St. Louis, MO). High performance liquid chro-matography (HPLC) grade hexane, methanol, ethyl acetate andglacial acetic acid were purchased from J. T. Baker (Phillipsburg,NJ, USA). Ultrapure water obtained from a Milli-Q filter system(Millipore, Bedford, MA, USA) was used throughout the analy-sis. N-Heptafluorobutanoylimidazole (HFBI) was purchased fromPierce Biotechnology (Rockford, IL, USA). Sodium methoxide wasdissolved in methanol (0.5 mol L−1) for transesterification. Theaqueous solutions of sodium chloride (NaCl) and sodium sulfate(Na2SO4) were prepared with the concentration of 200 g L−1. ChemElutTM columns using diatomaceous earth (HydromatrixTM) as thesorbent were obtained from Agilent (Palo Alto, CA, USA) and wereused for the SLE cleanup.

2.2. Standards

3-Monochloro-1,2-propandiol (3-MCPD, 98%), 1,3-dichloropropan-2-ol (1,3-DCP, 97%) and 2,3-dichloropropan-1-ol(2,3-DCP, 97%) were purchased from Fluka (Weinheim, Germany)while 2-monochloro-1,3-propandiol (2-MCPD, 98%) and 3-MCPDesters of rac-oleoyl, rac-linoleoyl, rac-linolenoyl, rac-1,2-dilinolenoyl, rac-1,2-dilinoleoyl, rac-palmitoyl, rac-1,2-dioleoyl,

rac-1,2-distearoyl and glycidyl esters of oleate, linoleate, linole-nate, stearate, palmitate (with chemical purities of ≥98%) werepurchased from Toronto Research Chemicals (Toronto, Canada).

2 togr. A

fd(df1rnTagopewD1it

2

lSnpS(n

2

tt2a3c

sbtMem

aadgrri

2

opbo

10 Q. Liu et al. / J. Chroma

1,3-Dichloropropan-2-ol-d5 (1,3-DCP-d5, 98%) was purchasedrom Isotec (Weinheim, Germany). 3-Monochloro-1, 2-propandiol-5 (3-MCPD-d5, 99%) was purchased from Dr. EhrenstorferAugsburg, Germany). 2-Monochloro-1,3-propandiol-d5 (2-MCPD-5), 2,3-dichloropropan-1-ol-d5 (2,3-DCP-d5), 3-MCPD-d5 estersor rac-1-oleoyl, rac-1,2-dilinolenoyl, rac-1,2-dilinoleoyl, rac--linoleoyl, rac-1-linolenoyl, rac-1,2-dioleoyl, rac-1,2-distearoyl,ac-1-palmitoyl, and glycidyl-d5 esters of oleate, linoleate, linole-ate, all with chemical purities of ≥98%, were purchased fromoronto Research Chemicals (Toronto, Canada). The deuteriumtoms of the above isotope standards are labeled in the skeleton oflycerol or glycidyl other than in the corresponding carbon chainf fatty acid. Individual stock solutions (1000 mg L−1) were pre-ared by dissolving an appropriate amount of each standard inthyl acetate and stored at −20 ◦C in amber glass vessels. Mixedorking solutions of standards (3-MCPD, 2-MCPD, 1,3-DCP and 2,3-CP) and deuterium-labeled standards (3-MCPD-d5, 2-MCPD-d5,,3-DCP-d5 and 2,3-DCP-d5) at 10 mg L−1 were prepared by merg-

ng all of the above stock solutions and diluting with hexane, andhen stored at −20 ◦C.

.3. Apparatus

Detection and quantification were performed with an Agi-ent Technologies 7890A gas chromatograph equipped with aeries 5975C quadrupole mass selective detector (Agilent Tech-ologies, Palo Alto, CA, USA). Data acquisition and processing wereerformed using an Agilent workstation (GCMSD For 1701EA).eparation was performed on a capillary column of DB-5MS30 m × 0.25 mm I.D., 0.25 �m film thickness) from Agilent Tech-ologies (Palo Alto, CA, USA) for all the analysis.

.4. GC–MS analysis

Splitless injection mode was used and the temperature of injec-or was set at 250 ◦C. The injection volume was 1 �L. The GCemperature programming was 50 ◦C for 1 min, followed by a◦C min−1 ramp to 90 ◦C and a final ramp of 40 ◦C min−1 to 250 ◦C,nd hold at 250 ◦C for 5 min. The total GC run time of analysis was0 min per sample. The ultrapure grade helium was employed asarrier gas at a flow rate of 1 mL min−1.

The tandem quadrupole instrument was operated in theelected ion-monitoring mode (SIM). The analytes were ionizedy electron impact (EI), 70 eV of ion energy, with a solvent delayime of 8 min. The temperatures of transfer line, MS source, and

S quadrupole were 280 ◦C, 230 ◦C, and 150 ◦C, respectively. Thelectron multiplier voltage was 1000 V, and the dwell time for eacheasured ion was set to 25 ms.Table 1 lists the studied compounds and the internal standards

long with their retention times, selected ions, limits of detectionnd quantification. The quantitation and qualifier abundances wereetermined by injection of standards under the same chromato-raphic conditions using full-scan with the mass-to-charge ratioanging from 60 to 400 m/z. The analytes were identified by theiretention times, the identification of quantitation ion and qualifierons, and the ratios of the qualifier ions to the quantitation ion.

.5. Samples

The edible oil samples, including a crude palm oil and vari-

us refined vegetable oils such as camellia-seed oil, sunflower oil,eanut sesame blend oil, phytosterols oil, tea-seed oil and sesamelend oil, were purchased from local markets. Samples of virginlive oils were used as matrix blanks and for the preparation of

1314 (2013) 208– 215

spiked samples. They were analyzed in advance and fatty acid esterslevels of four chloropropanols were found to be below the LODs.

2.6. Sample pretreatment

About 0.1 g oil sample was accurately weighted into a 12 mL-glass test tube with screwed cap, and 20 �L working internalstandard solution as well as 0.5 mL MTBE/ethyl acetate mixed solu-tion (8:2, v/v) were added. After vortex, 1 mL of NaOCH3 solutionwas added into the tube with a full vortex, and followed by placingfor 4 min at room temperature. Subsequently, 3 mL hexane fol-lowed by 3 mL of glacial acetic acid in Na2SO4 solution (1:29, v/v)were added. The upper phase was discarded after full vortex andliquid–liquid phase separation.

The aqueous extracts were further purified by SLE with ChemElutTM column. After loading the extracts and holding for 3 min,the column was washed with 20 mL hexane to remove the non-polar components, and then, the target compounds were elutedwith 80 mL of diethyl ether. The eluents were collected into 150 mL-flasks filled with 15 g anhydrous Na2SO4, then evaporated to almostdryness in a 35 ◦C water bath by a vacuum rotary evaporator.

The residue was dissolved in 2 mL hexane and derivatized with40 �L HFBI at 70 ◦C for 20 min. After cool down, 1 mL NaCl solutionwas added into the derivatives, and the mixture was shaken for 30 s.After dried with a little amount of anhydrous Na2SO4, the upperlayer was transferred into a sampler vial for GC–MS determination.

2.7. Validation

The quantification of the chloropropanols esters was achievedby using 4 deuterium isotope standards as internal standards.The series of calibration standards, including the levels of 25, 50,100, 200, 400, 800 and 2000 �g L−1 containing 4 deuterated inter-nal standards at 100 �g L−1, were prepared and derivatized, andinjected into GC–MS for three times. The average peak areas of theanalytes (A) and of the internal standards (Ai) were recorded. Allthe results were calculated using internal standard correction.

The limit of detection (LOD) and limit of quantification (LOQ),following International Union of Pure and Applied Chemistry(IUPAC) recommendation [29], were defined and determined asthe minimum detectable amount of analyses from blank oil spikedextract with a signal-to-noise ratio (S/N) of 3:1 and 10:1, respec-tively.

The accuracy and repeatability of the method were assessed bythe quantification of five replicates of analyte-free olive oils, andspiked at three levels of concentration (0.1, 0.5 and 2 mg kg−1) withfatty acid esters of 3-MCPD and glycidol. Additionally, to check theefficiency of ester cleavage, four chloropropanols were individuallyspiked into the blank matrix with the same levels as 3-MCPD esters.

3. Results and discussion

3.1. Selection of internal standards

In this study, the selection of internal standards is criticalfor analyzing 3-MCPD esters in edible oils. The standards ofesters of 3-MCPD and glycidyl and their one-to-one correspond-ing deuterium-labeled 3-MCPD-d5 and glycidyl-d5 esters werehydrolyzed, respectively. Meanwhile, the 3-MCPD esters or glycidylesters respectively spiked by 3-MCPD-d5 were also hydrolyzed. The3-MCPD levels determined were expressed as recovery rates. Theresults shown in Table 2 clearly indicated that the recoveries with

the spiking of internal standard 3-MCPD-d5 ranged from 43.3% to114.0% (RSD, 0.3–26.4%), while the recoveries with the spiking ofcorresponding esters of 3-MCPD-d5 or glycidyl-d5 as internal stan-dards ranged from 69.2% to 310.1% (RSD, 0.7–15.6%).

Q. Liu et al. / J. Chromatogr. A 1314 (2013) 208– 215 211

Table 1Method parameters including retention times, selected ions, regression equation, linear range, and limits of detection and quantification (LOD and LOQ, �g kg−1).

Chloropropanols derivatives Retention times Characteristic fragment ions (m/z)a Regression equation Linear range (mg L−1) LOD LOQ

3-MCPD 15.062 253, 289, 291, 275 y = 1.91x − 0.125 0.025–2.000 30 1003-MCPD-d5 14.857 257, 294, 296, 278 – – – –2-MCPD 15.471 253, 111, 277, 275 y = 1.10x − 0.095 0.025–2.000 30 1002-MCPD-d5 15.279 257, 116, 280, 278 – – – –1,3-DCP 11.296 75, 77, 275, 277 y = 1.52x − 0.144 0.025–2.000 100 2001,3-DCP-d5 11.127 79, 81, 278, 280 – – – –2,3-DCP 12.138 75, 77, 111, 253 y = 2.52x − 0.194 0.025–2.000 30 100

3tlgmlTow3o

oMs

3

ctMcuNocet

TTs

2,3-DCP-d5 11.922 79, 81, 116, 257

a Quantifier ions are underlined.

The results in Table 2 showed that the recoveries with-MCPD-d5 as internal standards were generally better thanhe esters of 3-MCPD-d5 or glycidyl-d5, especially for rac 1-inoleoyl-3-MCPD, rac 1-linolenoyl-3-MCPD, glycidyl-linoleate andlycidyl-linolenate. In other words, the recoveries of unsaturatedono-esters with the one-to-one corresponding deuterium-

abeled standards did not satisfy the recoveries requirements.herefore, 3-MCPD-d5 was adopted as the internal standard. More-ver, different oil having a different profile of fatty acid esters, itas not appropriate to select a single deuterium labeled ester of

-MCPD as the internal standard for the analysis of all the edibleils.

Considering there are no commercial deuterium-labeled estersf 2-MCPD, 1,3-DCP, and 2,3-DCP available, deuterium-labeled 2-CPD, 1,3-DCP, and 2,3-DCP can also be used as one-to-one internal

tandards.

.2. Sample preparation

Some investigations demonstrated that fatty acid esters of gly-idol (oxirane-2-methanol, 2, 3-epoxy-1-propanol) may interferehe determination and lead to the over-estimated values of 3-

CPD esters, which were almost completely transformed intoyclic 1,3,2-dioxaboralane derivative of 3-MCPD by derivatizationsing phenylboronic acid in NaCl solution [30]. Consequently, thea2SO4 solution was used to extract the target compounds instead

f NaCl solution. From Table 2, it is concluded that glycidyl estersannot be cleavaged into free 3-MCPD using Na2SO4 solution asxtraction solution either in blank solvent or blank oil matrix. Addi-ionally, seven oil samples were determined using NaCl solution

able 2he comparison of average recoveriesa of 3-MCPD esters and glycidyl esters with differeolvent and oil matrix using different extraction solvents.c

Compounds Spiking in solvent

IS: 3-MCPD-d5 IS: Correspo

NaCl NaCl

Oleoyl-3-MCPD 114.0 (0.3) 69.2 (6.7)

Linoleoyl-3-MCPD 70.4 (16.1) 222.9 (15.6Linolenoyl-3-MCPD 93.3 (26.4) 259.6 (0.7)

1,2-Dilinolenoyl-3-MCPD 97.5 (1.4) 96.7 (5.8)

1,2-Dilinoleoyl-3-MCPD 96.5(2.7) 98.3 (3.0)

Palmitoyl-3-MCPD 99.8 (4.1) 93.8 (5.3)

1,2-Dioleoyl-3-MCPD 73.5 (14.2) 75.6 (4.5)

1,2-Distearoyl-3-MCPD 100.1 (4.3) 92.8 (2.2)

Glycidyl-oleate 74.0 (4.1) 98.8 (2.1)

Glycidyl-linoleate 43.3 (13.5) 310.1 (2.7)

Glycidyl-linolenate 81.4 (13.7) 232.7 (0.9)

Glycidyl-stearate 111.6 (13.7) d

Glycidyl-palmitate 105.1 (3.5) d

a Results are the mean of five replicates (%), and RSD % are given in parentheses.b The results are recoveries of 3-MCPD esters and glycidyl esters using corresponding 3c The blank solvent and blank oil matrix were all spiked at 0.5 mg kg−1.d Deuterium-labeled glycidyl esters for stearate and palmitate are not available. ND: no

– – – –

and Na2SO4 solution, respectively. Lower values of chloropropanolsesters were obtained using Na2SO4 solution instead of NaCl solu-tion (Table 3). Based on the above facts, the Na2SO4 solution wasselected as the solvent for extracting the target compounds to avoidthe interference for determination of chloropropanols esters byglycidyl esters.

The cleavage time of esters, the diethyl ether amount, and typesof derivatization agents can influence quantitatively the reaction.In this study, the optimum cleavage time was investigated in detail.One mL of sodium methoxide solution (0.5 mol L−1) was added intopalm oil to react for 2, 4, 6, 8, 10 and 12 min, respectively. The estercleavage time of 2 min was shown to provide the highest quan-tity of 3-MCPD (see supplementary Table S1). With the increase ofester cleavage time, the decrease of 3-MCPD level was observed.However, the ratios of the area of 3-MCPD to 3-MCPD-d5 wereclose between each other with different ester cleavage time treat-ments, the RSD of which was 6.4%. If internal standard method wasapplied to the determination, the ester cleavage time only affectthe sensitivity rather than the accuracy. Considering the internalstandard quantification and operation feasibility for a large amountof sample preparation in a batch, a cleavage time of 4 min wasselected.

Supplementary material related to this article can befound, in the online version, at http://dx.doi.org/10.1016/j.chroma.2013.08.074.

Fig. 2 shows the influence of the diethyl ether amount used to

elute chloropropanols from the diatomaceous earth on the peakareas, which affect the sensitivity. The results indicated that thepeak areas of 3-MCPD and 3-MCPD-d5 increased with the increas-ing volume of diethyl ether in the range of 40–70 mL, and then the

nt internal standards, and recoveries of the studied glycidyl esters compounds in

Spiking in blank oil

nding esters-d5b IS: 3-MCPD-d5

Na2SO4 NaCl Na2SO4

– – –) – – –

– – –– – –– – –– – –– – –– – –ND 13.7 (1.3) NDND 10.8 (1.6) NDND 15 (11.3) NDND 14.4 (3.2) NDND 18.7 (10.0) ND

-MCPD-d5 esters and glycidyl-d5 esters as internal standards.

t detected.

212 Q. Liu et al. / J. Chromatogr. A 1314 (2013) 208– 215

Table 3Concentration of chloropropanols esters in oil samples using Na2SO4 solution and NaCl solution as extraction solvent (mg kg−1).

3-MCPD esters 2-MCPD esters 1,3-DCP esters 2,3-DCP esters

Na2SO4 NaCl Na2SO4 NaCl Na2SO4 NaCl Na2SO4 NaCl

Camellia-seed oil 1.452 1.540 1.008 1.060 ND ND ND NDSunflower oil 0.957 1.018 0.675 0.696 ND ND ND NDPeanut sesame blend oil 0.262 0.273 0.215 0.211 ND ND ND NDPhytosterols corn oil 0.428 0.489 0.338 0.366 ND ND ND NDSunflower oil 1.014 1.023 0.726 0.765 ND ND ND NDTea-seed oil 0.277 0.456 0.232 0.228 ND ND ND NDSesame blend oil 0.320 0.434 0.281 0.265 ND ND ND ND

N

m8

u[Mmdtcbheab

ddwlcioaeuTbo

D, not detected.

oderate growth was observed from 70 to 90 mL (Fig. 2). Therefore,0 mL diethyl ether was selected for the elution of chloropropanols.

According to DGF standard methods C-III 18 (09), PBA wassed as the derivatization agent in many previous studies14,16,22,30–32]. The advantage is that 3-MCPD released from 3-

CPD esters in NaCl solution can be derivatized directly, whichakes it to be a relatively easy and convenient method. The major

isadvantage, however, is that 1,3-DCP and 2,3-DCP released fromheir corresponding easters cannot be derivatized with PBA andould not be detected. In addition, if the split/splitless injector hadeen used instead of the PTV injector, baseline noise and shift wouldave seriously affected the response of the target compounds. Afteright consecutive injections of the same sample, a baseline driftppeared as a result and therefore accurate quantification cannote achieved.

It is well known that instrumental detection capability dependsirectly on the cleanliness of the extract. In order to overcome theisadvantage of traditional methods of DGF [24], a novel method,hich based on the cleanup procedure of the solid-supported

iquid–liquid extraction (SLE) by commercial column of diatoma-eous earth were proposed in this work. In late 1997, SLE as annnovative purification method was proposed for the purificationf combinatorial libraries [33]. The technique involves supportingn aqueous buffer on a bed of coarse mesh, calcinated diatomateousarth sold under the product name Hydromatrix. It has already been

sed with success mainly in the biological fluid analysis [34–37].he application of SLE to the analysis of phenolic compounds haseen described in balsamic vinegar, olives and wine [38,39]. Onef the advantages of SLE over extraction is fully compatible with a

Fig. 2. Influence of dieth

wide range of solvents, regardless of density. Extractions are readilyperformed in any solvent that is not miscible with water. In addi-tion, because emulsions are not an issue with SLE, solvent mixturescan be used regardless of density. Therefore, the technique couldbe used for extracting chloropropanols from the aqueous solution.The extracted chloropropanols were transferred from the aqueousphase to the organic phase (ethyl ether) through this purificationstep of absorbing water and removing impurities. Thereafter, thefour chloropropanols released from all the chloropropanols esterscan be derivatized with the HFBI and analyzed by GC–MS. Fig. 3shows the chromatograms and spectrums of standard sample anal-ysis. The baseline noise and drift are inconspicuous.

3.3. Method validation

3.3.1. Calibration curvesThe calibration data and regression equation of the calibration

curves are presented in Table 1. The ANOVA test for the linear-ity of the calibration curves was performed within the range of0.025–2.000 mg L−1, and the linearity was always satisfied withcorrelation coefficients (r) not less than 0.999.

3.3.2. Limits of detection (LODs) and quantification (LOQs)The limits of detection (LODs) and quantification (LOQs) were

considered as the minimum amount of target analyte that pro-

duces a chromatogram peak with a signal-to-noise ratio (S/N) ofthree and ten times as the background chromatographic noise,respectively. The S/N was measured at the spiked levels of 10, 20,30, 40, 50, 100, 200, and 300 �g kg−1, respectively. Results from

yl ether amount.

Q. Liu et al. / J. Chromatogr. A 1314 (2013) 208– 215 213

F sampl −1

d ters dd

T1a

3

rest

ig. 3. The chromatograms and mass spectrums of the processed virgin olive oil

erivative, (2) 1,3-DCP esters derivative, (3) 2,3-DCP-d5 derivative, (4) 2,3-DCP eserivative, (8) 2-MCPD esters derivative.

able 1 show that LODs of fatty acid esters of 3-MCPD, 2-MCPD,,3-DCP and 2,3-DCP (calculated as free forms) were 30, 30, 100nd 30 �g kg−1, respectively.

.3.3. Recovery experimentsThe accuracy and precision of the method were examined by the

eproducibilities using spiked blank sample matrices. The recov-ry estimation of this method was performed on spiked blank oilamples at concentration levels of 0.1, 0.5 and 2 mg kg−1. Sincehe commercial fatty acid esters of 2-MCPD, 1,3-DCP and 2,3-DCP

e with the optimized method at the fortified level of 0.5 mg kg . (1) 1,3-DCP-d5

erivative, (5) 3-MCPD-d5 derivative, (6) 3-MCPD esters derivative, (7) 2-MCPD-d5

are not available, 2-MCPD, 1,3-DCP and 2,3-DCP were spiked intoblank oil samples. Average recoveries of the eight 3-MCPD estersand the four chloropropanols ranged from 70.7% to 113.3%. Thereproducibility of this method represented by the relative standarddeviation (RSD) percentage is less than 15.0% at each spiked level(Table 4).

The recoveries of 3-MCPD esters are comparable with thatof 3-MCPD, meaning the complete cleavage of ester bond inchloropropanols esters. Consequently, the recoveries of the chloro-propanols can represent that of the corresponding chloropropanol

214 Q. Liu et al. / J. Chromatogr. A 1314 (2013) 208– 215

Table 4Average recoveriesa of the chloropropropanol esters, chloropropanols and glycidyl esters at three spiked levels.

Compound Fortification levels (mg kg−1)

0.1 0.5 2

Recovery RSD Recovery RSD Recovery RSD

Oleoyl-3-MCPD 113.3 3.8 109.2 5.6 112.8 3.9Linoleoyl-3-MCPD 99.8 3.3 89.9 3.2 84.8 3.1Linolenoyl-3-MCPD 87.5 10.2 89.4 5.4 77.9 5.91,2-Dilinolenoyl-3-MCPD 89.5 0.8 77.0 1.0 93.1 6.61,2-Dilinoleoyl-3-MCPD 98.7 2.2 70.7 7.0 76.6 4.7Palmitoyl-3-MCPD 95.6 8.8 76.9 2.1 84.0 4.81,2-Dioleoyl-3-MCPD 98.8 4.0 72.5 3.3 82.2 8.71,2-Distearoyl-3-MCPD 105.7 3.8 97.2 1.5 108.2 12.13-MCPD 94.9 12.4 96.9 9.8 95.5 8.52-MCPD 96.6 7.3 86.5 2.6 85.8 10.91,3-DCP 90.0 3.9 97.6 3.9 106.2 10.12,3-DCP 93.9 7.7 104.8 10.1 98.3 3.5Glycidyl-oleate ND – ND – 8.8 2.4Glycidyl-linoleate ND – ND – 2.1 5.0Glycidyl-linolenate ND – ND – 7.3 8.5Glycidyl-stearate ND – ND – 7.0 2.2Glycidyl-palmitate ND – ND – 8.0 12.4

a Results are the mean of five replicates (%), and RSD % are given in parentheses. ND: not detected.

ult (La

elmb

sATw4ttn

3

p

Fig. 4. FAPAS proficiency test res

sters. Additionally, the recoveries of five glycidyl esters were veryow (ND-8.8%), which can draw the same conclusion that the deter-

ination of fatty acid esters of chloropropanols was not interferedy glycidyl esters in the proposed method.

Additionally, this method was applied to the refined palm oilample of proficiency test 2631 of Food Analysis Performancessessment Scheme (FAPAS) [40] for the analysis of 3-MCPD esters.he value submitted (4.53 mg kg−1) by this novel method falls wellithin the scope of specified “satisfactory” range (assigned value,

.99 mg kg−1), with Z-score obtained at −0.5 (Fig. 4). In addition,he level of 2-MCPD esters in the palm oil sample determined withhe novel method (2.32 mg kg−1) was also submitted, which wasot a compulsive analyte in this performance test.

.4. Method application

To further demonstrate the utility and performance of the pro-osed methodology, different edible oils including soybean oil, corn

b No. 21 is the present method).

oil, sesame oil, virgin olive oil, peanut oil, linseed oil, tea-seed oil,tea-seed olive blend oil, sunflower-seed oil and blend oil were ana-lyzed.

The purification process was demonstrated to be efficient, andno baseline shift appeared after several repetitious analyses. The3-MCPD esters and 2-MCPD esters were detected in all samplesexcept virgin olive oil. Total concentrations of 3-MCPD esters in allthe 10 oil samples ranged from ND to 1.54 mg kg−1, and that for 2-MCPD esters were generally in the range of ND-1.06 mg kg−1. Theesters of 2,3-DCP was found in a sunflower oil and a tea-seed oil,with the concentration of 0.300 mg kg−1 and 0.340 mg kg−1, respec-tively, while no esters of 1,3-DCP was detected in any of the testedoil samples.

4. Conclusions

This paper proposed a novel analytical procedure for the rou-tine analysis of total fatty acid esters of chloropropanols in edible

ogr. A

ottPL(qmAMobuo

A

sPgsp

R

[

[

[

[

[[

[

[[[

[

[

[[

[

[

[

[

[

[

[[

[

[

[[

[

[

Q. Liu et al. / J. Chromat

ils based on GC–MS. The proposed method has robust resistanceo the matrix of edible oil, and is suitable for the determination ofotal fatty acid esters of chloropropanols by GC–MS without theTV injectors. The method validation data including calibration,OD/LOQ, accuracy and repeatability and proficiency test resultsZ-score: −0.5) of the official FAPAS indicated that the presentuantitative method could be successfully applied to the deter-ination of total chloropropanols esters in various edible oils.ccording to results of sample determination, fatty acid esters of 3-CPD and 2-MCPD are found at significant levels in various edible

ils, while esters of 1,3-DCP and 2,3-DCP might also present in oils,ut at relatively low concentrations. The results obtained will beseful for future studies focusing on the formation and occurrencef fatty acid esters of chloropropanols.

cknowledgements

We would like to thank all the donors who provided edible oilamples. This study was supported by the National Basic Researchrogram (973, No. 2012CB720804) and the National High-Tech Pro-ram (863, No. 2010AA023003). No funding bodies had any role intudy design, data collection and analysis, decision to publish, orreparation of the manuscript.

eferences

[1] J. Velísek, J. Davidek, J. Hajslová, V. Kubelka, G. Janícek, B. Mánková, Z. Lebensm,Unters. Forsch. A 167 (1978) 241.

[2] I. Racamonde, P. González, R. Lorenzo, A. Carro, J. Chromatogr. A 1218 (2011)6878.

[3] B. Svejkovska, O. Novotny, V. Divinova, Z. Reblova, M. Dolezal, J. Velísek, CzechJ. Food Sci. 22 (2004) 190.

[4] K. Hrncirik, G. van Duijn, Eur. J. Lipid Sci. Technol. 113 (2011) 361.[5] B.S. Lynch, D.W. Bryant, G.J. Hook, E.R. Nestmann, I.C. Munro, Int. J. Toxicol. 17

(1998) 47.[6] Z. Zelinková, O. Novotny, J. Schurek, J. Velísek, J. Hajslová, M. Dolezal, Food

Addit. Contam. 25 (2008) 669.[7] C.G. Hamlet, P.A. Sadd, D.A. Gray, J. Agric. Food. Chem. 52 (2004) 2059.[8] R.H. Stadler, V. Theurillat, A. Studer, F. Scanlan, W. Seefelder, Deutsche

Forschungsgemeinschaft (DFG German Research Foundation), Thermal Pro-cessing of Food: Potential Health Benefits and Risks, Wiley VCH Verlag GmbH,Darmstadt, Germany, 2007, pp. 141.

[9] M. Hwang, E. Yoon, J. Kim, D.D. Jang, T.M. Yoo, Regul. Toxicol. Pharm. 53 (2009)

102.

10] Federal German Institute for Risk Assessment, Stellungnahme Nr. 047/2007 desBfR vom 11, BfR, Berlin. http://www.bfr.bund.de/cm/208/saeuglingsanfangsund folgenahrung kann gesundheitlich bednkliche 3 mcpd fettsaeureesterenthalten.pdf

[[[

1314 (2013) 208– 215 215

11] Safety evaluation of certain food additives and contaminants/prepared by thesixty-seventh meeting of Joint FAO/WHO Expert Committee on Food Addi-tives (JECFA), Geneva 2007, WHO Food Additives Series, No. 58, pp. 209–238.

12] M. Kusters, U. Bimber, A. Ossenbruggen, S. Reeser, R. Gallitzendorfer, M. Ger-hartz, J. Agric. Food. Chem. 58 (2010) 6570.

13] Z. Zelinková, B. Svejkovská, J. Velísek, M. Dolezal, Food Addit. Contam. 23 (2006)1290.

14] J. Kuhlmann, Eur. J. Lipid Sci. Technol. 113 (2011) 335.15] FAO/WHO, Safety evaluation of certain food additives and contami-

nants/prepared by the fifty-seventh meeting of Joint FAO/WHO ExpertCommittee on Food Additives (JECFA), Rome, 2001, pp. 114, WHO TechnicalReport Series, No. 909.

16] V. Divinova, B. Svejkovska, M. Dolezal, J. Velísek, Czech J. Food Sci. 22 (2004)182.

17] C. Hamlet, P. Sadd, Czech J. Food Sci. 22 (2004) 259.18] Z. Zelinková, M. Dolezal, J. Velísek, Eur. Food Res. Technol. 228 (2009) 571.19] K. Yamazaki, M. Ogiso, S. Isagawa, T. Urushiyama, T. Ukena, N. Kibune, Food

Addit. Contam. 30 (2013) 52.20] K. Hori, N. Koriyama, H. Omori, M. Kuriyama, T. Arishima, K. Tsumura, LWT-Food

Sci. Technol. 48 (2012) 204.21] E. Moravcova, L. Vaclavik, O. Lacina, V. Hrbek, K. Riddellova, J. Hajslova, Anal.

Bioanal. Chem. 402 (2012) 2871.22] K. Hrncirik, Z. Zelinkova, A. Ermacora, Eur. J. Lipid Sci. Technol. 113 (2011) 361.23] W. Seefelder, N. Varga, A. Studer, G. Williamson, F. Scanlan, R. Stadler, Food

Additi. Contam. 25 (2008) 391.24] DGF Standard Method C-III 18 (09), Tensiden und verwandten Stoffen Wis-

senschaftliche Verlagsgesellschaft, Stuttgart, Germany, 2009.25] S. Abu-El-Haj, M.J. Bogusz, Z. Ibrahim, H. Hassan, M. Al Tufail, Food Control 18

(2007) 81.26] S.W.C. Chung, K.P. Kwong, J.C.W. Yau, A.M.C. Wong, Y. Xiao, J. Food Comp. Anal.

21 (2008) 569.27] M. Dolezal, M. Chaloupská, V. Divinová, B. Svejkovská, J. Velisek, Eur. Food Res.

Technol. 221 (2005) 221.28] S. Takaichi, K. Matsui, M. Nakamura, M. Muramatsu, S. Hanada, Comp. Biochem.

Physiol. B: Biochem. Mol. Biol. 136 (2003) 317.29] IUPAC Compendium of Chemical Terminology, 2nd ed., 1997, See website:

http://www.chemsoc.org/chembytes/goldbook/30] R. Weißhaar, R. Perz, Eur. J. Lipid Sci. Technol. 112 (2010) 158.31] B. Matthäus, F. Pudel, P. Fehling, K. Vosmann, A. Freudenstein, Eur. J. Lipid Sci.

Technol. 113 (2011) 380.32] T.D. Haines, K.J. Adlaf, R.M. Pierceall, I. Lee, P. Venkitasubramanian, M.W. Col-

lison, J. Am. Oil Chem. Soc. 88 (2011) 1.33] J.G. Breitenbucher, K.L. Arienti, K.J. McClure, J. Comb. Chem. 3 (2001)

528.34] S.X. Peng, T.M. Branch, S.L. King, Anal. Chem. 73 (2001) 708.35] A.Q. Wang, W. Zeng, D. Musson, J. Rogers, A. Fisher, Rapid Commun. Mass

Spectrom. 16 (2002) 975.36] M. Hengel, B. Hung, J. Engebretson, T. Shibamoto, J. Agric. Food. Chem. 51 (2003)

6635.37] H. Wingfors, M. Hansson, O. Päpke, S. Bergek, C.-A. Nilsson, P. Haglund, Chemo-

sphere 58 (2005) 311.38] J. Pazourek, G. González, A. Revilla, J. Havel, J. Chromatogr. A 874 (2000) 111.39] F. Nave, M.J. Cabrita, C.T. da Costa, J. Chromatogr. A 1169 (2007) 23.40] Food Analysis Performance Assessment Scheme (FAPAS). Available from:

http://www.fapas.com/index.cfm