Journal of the American Oil Chemists' Society

, Volume 90, Issue 4, pp 493–500

Collaborative Study for the Analysis of Glycidyl Fatty Acid Esters in Edible Oils using LC–MS

Authors

  • Michael R. Blumhorst
    • Research Division, Archer Daniels Midland CompanyJames R. Randall Research Center
  • Mark W. Collison
    • Research Division, Archer Daniels Midland CompanyJames R. Randall Research Center
  • Richard Cantrill
    • AOCS
  • Hiroki Shiro
    • Tochigi Research LaboratoriesKao Corporation
  • Yoshinori Masukawa
    • Tokyo Research LaboratoriesKao Corporation
  • Shigeru Kawai
    • Tokyo Research LaboratoriesKao Corporation
    • Tokyo Research LaboratoriesKao Corporation
Original Paper

DOI: 10.1007/s11746-012-2187-7

Cite this article as:
Blumhorst, M.R., Collison, M.W., Cantrill, R. et al. J Am Oil Chem Soc (2013) 90: 493. doi:10.1007/s11746-012-2187-7

Abstract

An interlaboratory study was conducted to evaluate a method for determining glycidyl fatty acid esters (GE) in edible oils. Samples were dissolved in tert-butyl methyl ether/ethyl acetate and subjected to two solid-phase extraction (SPE) steps. The first SPE step utilized methanol elution from a C18 cartridge, and the second SPE step utilized n-hexane/ethyl acetate elution from a silica cartridge. The final extract was analyzed using liquid chromatography with a single quadrupole mass spectrometer in selected ion monitoring (SIM) mode. Quantification was performed using external standardization. Eighteen samples (9 oils × 2 blind duplicates) were assayed for glycidyl palmitate, glycidyl stearate, glycidyl oleate, glycidyl linoleate and glycidyl linolenate by 17 collaborating laboratories from seven countries. Sample matrices included palm, olive, corn, soybean and rapeseed oils. Repeatability (RSDr) ranged from 6.85 to 19.88 % and reproducibility (RSDR) ranged from 16.58 to 35.52 % for samples containing greater than 0.5 mg/kg of individual GE. HORRATR values ranged from 0.62 to 14.70 for determination of total GE. The method provides acceptable results for quantification of GE in edible oils.

Keywords

Collaborative studyGlycidyl fatty acid estersEdible oilInterlaboratory study

Introduction

The initial detection of glycidyl fatty acid esters (GE) in vegetable oils was the result of research to investigate the origin of 3-monochloro-1,2-propanediol (3-MCPD) and 3-MCPD esters of fatty acids in these oils [14]. In their work directed at identifying the precursors to these compounds, Weißhaar and Perz [5] reported the existence of relatively high levels of GE. In response to this finding, Masukawa et al. [6] performed a survey of commercial oils sold in Japan and reported that GE were detected in every sample tested. Recent work by Haines et al. [7] cast doubt on the extent of 3-MCPD and 3-MCPD ester formation during oil processing, but these researchers detected GE in a variety of vegetable oils with amounts correlated to the diacylglyceride content of the oil [7].

The toxicological profiles of GE are incomplete, and the level of glycidol released from these esters in the digestive tract is unknown at this time. However, glycidol is classified as a genotoxic carcinogen (IARC group 2A, “probably carcinogenic to humans”) by the International Agency for Research on Cancer [8]. Therefore, methods to provide accurate quantification of GE are required. Blumhorst et al. [9] reported a direct liquid chromatography–mass spectrometry (LC–MS) method which can be used as a screening method to quantify GE in processed oil samples. These researchers reported acceptable method recovery and repeatability, but Masukawa et al. [6] reported a rapid decline in instrument signal/noise following injection of diluted oils without further clean-up. Therefore, this study was conducted to provide a validated method for the quantification of GE in commercial fats and oils. The method is believed to be applicable for all specified matrices and requires only a single quadrupole LC–MS instrument.

Materials and Methods

Qualification Study

The method for the initial qualification study was based on the procedure of Shiro et al. [10]. Three qualification samples were assayed by participating laboratories, but results from this trial were variable. Reasons for this variability included the use of low-purity standards and a method that allowed too much flexibility by the analyst (e.g. the use of plastic pipettes/collection tubes introduced interfering compounds into sample extracts). These issues were addressed in the revised method which was evaluated in another qualification study.

Each laboratory was supplied with three qualifying samples consisting of refined, bleached, deodorized (RBD) palm oil, extra virgin olive oil and spiked extra virgin olive oil. Reference standards were obtained independently by each participating laboratory. Laboratories were required to report successful results on those test samples prior to beginning the collaborative sample analysis. Seventeen of the 24 laboratories that participated in the 2nd qualification study obtained results that were acceptable. These 17 laboratories represented seven countries including France, Japan, Canada, Germany, China, the United Kingdom and the United States (Table 1).
Table 1

List of collaborating laboratories

Principal investigator

Company

Address

Shaun MacMahon

CFSAN-FDA

FDA Center for Food Safety, Applied Nutrition, 5100 Paint Branch Parkway, College Park MD 20740

Florence Lacoste

ITERG

11 rue Monge, Parc Industriel Bersol 2, 33600-Pessac, France

Takatoshi Yamashita

J-Oil Mills, Inc.

7-41, Daikoku-cho, Tsurumi-ku, Yokohama-City, 230-0053, Japan

Adam Becalski

Health Canada, HPFB, FD, BCS, FRD

PL 2203D, RM D350, 251 Banting Drive, Ottawa, ON K1A 0L2

Nobuyuki Kibune

Japan Food Research Laboratories

Saito laboratory, 7-4-41, Saitoasagi, Ibaraki-shi, Osaka 567-0085, Japan

Colin Crews

FERA

The Food and Environment Research Agency,Room 08GA08, Sand Hutton, York YO41 1LZ, UK

Mark Collison

ADM Research

1001 Brush College Rd, Decatur, IL 62521, USA

Hiroki Shiro

Kao Corp

Analytical Science Research Center, 2606 Akabane, Ichikaimachi, Haga-gun, Tochigi 321-3497, Japan

Yasuhiko Shigematsu

Kewpie Corporation R&D Division

Food Safety Research Center, 5-13-1, Sumiyoshi-cho, Fuchu-shi, Tokyo, 183-0034, Japan

Hiroshi Hirai

The Nisshin Oillio Group, Ltd

1 Shinmei-cho, Yokosuka 239-0832, Japan

Hong Yang

Wilmar International

Biotechnology Research & Development Center (Shanghai) Co, Ltd, No 118 Gaodong Road, Pudong New District, Shanghai, China

Hildburg Fry

BfR Berlin

Thielallee 88-92, D-14195 Berlin, Germany

Yingyao Wang

Academy of State Administration of Grain

C/o Ma Rong, No 11 Baiwangzhuang Street, Xicheng District, Beijing, China 100037

Tetsuji Tomita

Showa Sangyo Co., Ltd

Research & Development Center, 20-2, Hinode, 2-Chome, Funabashi, Chiba 273-0015, Japan

Claudia Schulz

Eurofins WEJ Contaminants

Neulander Kamp 1, D-21079 Hamburg, Germany

Toru Fukazawa

Japan Institute of Oils and Fats, Other Foods Inspection

3-27-8, Nihonbashi-Hamacho, Chuo-ku, Tokyo, 103-0007, Japan

Yuko Katagiri

ADEKA Corporation

Chemical Analysis Group, 7-2-34 Higashiogu, Arakawaku Tokyo, 116-8553, Japan

Collaborative Study

Collaborative study samples were acquired by representatives of the American Oil Chemists’ Society in Urbana, IL where they were thoroughly mixed, transferred to individual sample-size vials, flushed with nitrogen, and stored at ambient temperature in the dark. Appropriate samples were spiked in bulk prior to mixing and transferring to vials. Eighteen samples (9 oils × 2 blind duplicates) were provided to participating laboratories for the collaborative study. These samples included extra virgin olive oil, spiked extra virgin olive oil, three RBD palm oils, RBD palm oil spiked with glycidyl stearate, RBD rapeseed oil, RBD soybean oil, and RBD corn oil.

Reference Standards and Calibration Solutions

Analytes for this study included glycidyl palmitate (C16:0), glycidyl stearate (C18:0), glycidyl oleate (C18:1), glycidyl linoleate (C18:2) and glycidyl linolenate (C18:3). Standards may be purchased from Wako Pure Chemical Industry Ltd., Tokyo, Japan; and Tokyo Chemical Industry Co. Ltd., Tokyo, Japan. Glycidyl fatty acid ester standards are available from other manufacturers, but purity must be clearly indicated and/or determined because standards may contain other alkyl glycidyl fatty acid esters as impurities. Glycidyl fatty acid ester standards were stored at –20 °C with a nitrogen headspace to avoid degradation. All calibration solutions were prepared in methanol:isopropanol 1:1 (v:v). The concentration range for individual GE in the calibration solutions was 5–200 ng/mL (5, 10, 50, 100 and 200 ng/mL). Calibration solutions were stable for at least 2 weeks at 4 °C when stored under nitrogen.

Sample Analysis

The analytical method was based on the procedure of Shiro et al. [10] and is described in detail elsewhere [11]. Briefly, 1-g samples of oil were diluted to 10.0 mL with tert-butyl methyl ether:ethyl acetate 4:1 (v:v), and subjected to two SPE steps followed by LC–MS analysis. The first SPE step utilized a C18 cartridge with methanol as the elution solvent. After changing the solvent to n-hexane/ethyl acetate 95:5 (v:v), the second SPE step utilized a silica cartridge with n-hexane/ethyl acetate 95:5 (v:v) as the elution solvent. Upon collection of the appropriate fractions, this solution was evaporated to dryness using nitrogen, and 1.0 mL of methanol:isopropanol 1:1 (v:v) was added. Samples were assayed using reversed phase chromatography with a single quadrupole mass spectrometer and atmospheric pressure chemical ionization (APCI) in positive ion mode. The mass spectrometer used must provide signal:noise >10 upon analysis of a 20-μL injection of a 5-ng/mL GE standard solution. Quantification was performed using selected ion monitoring (SIM) of the [M + H]+ ion with external standardization. With the exception of the SPE cartridges, the use of any plastic equipment (e.g., plastic pipette tips or plastic stop valves for SPE) should be avoided because plastic resins may introduce compounds that interfere with analyte resolution. Also, solid oils (fats) should be heated to not more than 5 °C above their melting point and thoroughly homogenized prior to sampling. Extended exposure to high temperature may cause decomposition or conversion of GE in the sample. Extra virgin olive oil previously determined to contain no detectable level of GE was used as the blank sample and also to prepare a spiked sample containing a total nominal GE concentration of 50 mg/kg.

Statistical Evaluation

Reproducibility and repeatability values were calculated according to the AOAC/IUPAC Harmonized Protocol. Collaborative study data submitted with less than the limit of quantification (LOQ) were not considered in the evaluation of the results, therefore, some reported values consisted of fewer than 17 data points for statistical analysis.

Results and Discussion

Preliminary precision values are given in Tables 2 and 3 and reported elsewhere [11]. If the method is followed exactly with the same LC–MS instrument, it provides a limit of quantitation (LOQ) for each GE of <1 ng/mL (Table 2) which corresponds to a sample LOQ of 0.1 mg/kg. Recovery values for the five GE ranged from 100 to 109 % with RSD values <9 % (Table 3). Based on this data, the method was submitted for evaluation in the collaborative study.
Table 2

Relative standard deviations (RSD %) during consecutive runs, limits of detection (LOD), limits of quantification (LOQ) and calibration line obtained from standard glycidyl fatty acid esters (GE)

GE

RSD %a

LODb

LOQb

Calibration linec

 

Retention time

Area

(ng/mL)

(ng/mL)

Equation

R2

C16:0-GE

0.10

1.9

0.23

0.76

y = 1,600x – 20

0.9999

C18:0-GE

0.18

3.3

0.19

0.63

y = 1,300x – 730

0.9995

C18:1-GE

0.15

4.2

0.26

0.87

y = 1,600x + 1,600

0.9998

C18:2-GE

0.11

4.9

0.19

0.63

y = 1,500x + 3,400

0.9997

C18:3-GE

0.09

3.3

0.25

0.84

y = 1,400x – 100

0.9999

aObtained from six consecutive runs of the standard GE solution at a concentration of 100 ng/mL

bDefined as signal:noise = 3 for LOD and 10 for LOQ based on 20-μL injections of the standard GE solution at a concentration of 5 ng/mL

cCalculated from the equation y = Ax + B, where x is the injected concentration (ng/mL), y is the peak area, A is the slope, and B is the intercept

Table 3

Levels of glycidyl fatty acid esters (GE) in a commercial palm oil with recovery values

GE

Level (mg/kg)a

Recovery %a, b

 

Mean ± std dev

Mean ± std dev

RSD %

C16:0-GE

7.5 ± 0.26

101.5 ± 5.0

4.9

C18:0-GE

1.2 ± 0.03

100.2 ± 5.9

5.9

C18:1-GE

13.5 ± 0.22

106.0 ± 8.7

8.2

C18:2-GE

4.2 ± 0.09

105.4 ± 7.1

6.8

C18:3-GE

<0.5c

109.0 ± 0.6

0.6

aAnalyzed in triplicate

bObtained from the ratio of the difference between quantified levels of the GE in the spiked oil and those in the non-spiked oil to the known levels spiked (10 mg/kg of each GE)

cDetermined by the lowest concentration of the standard solution for calibration and dilution degrees (100-fold dilution) in the oil sample preparation

For extra virgin olive oil, spiked extra virgin olive oil, rapeseed oil, soybean oil and corn oil, repeatability values (RSDr) were <12 % for each analyte when sample levels were >0.5 mg/kg (Table 4). Repeatability values generally increased in these matrices as analyte levels decreased below 0.5 mg/kg. Greater variability at low concentrations is not unusual, but the increased variability here was due, in part, to the fact that these levels were less than the lowest calibration standard specified (5 ng/mL). Differences in method interpretation between collaborating laboratories resulted in several laboratories reporting values between 0.1 and 0.5 mg/kg as <LOQ whereas others reported actual values. Reproducibility values (RSDR) for these same matrices were greater than desired ranging from 16.58 to 23.97 % for individual analytes present at >0.5 mg/kg. Reproducibility may have been affected by GE degradation in the samples. For instance, a spiked sample was prepared at the beginning of this study to contain 10 mg/kg of each GE for a nominal total concentration of 50 mg/kg in extra virgin olive oil. This sample was stored under the identical conditions as all other samples, i.e., in individual nitrogen-flushed vials at room temperature in the dark. In this way, appropriate sample vials were simply gathered and shipped to participating laboratories at the desired time without the need for subsequent exposure to light or air. Results from the first qualification study conducted 3 months after preparation show detected values slightly less than theoretical values (43.20 mg/kg detected vs. 50 mg/kg theoretical, Table 5). At 10 months (2nd qualification study), measured values showed a significant decrease in GE content. Results from this same sample assayed during the full collaborative study showed a 24 % reduction in total GE content from month 10 to 13, and this trend was consistent for all five analytes. The time window from sample receipt to reporting results for the full collaborative study was 1.5 months (13–14.5 months after preparation) so GE degradation likely was a factor in the greater than expected RSDR values. These results indicate that GE may not be stable in certain oil matrices, and analysts should be aware of this potential instability when designing analytical protocols and reporting results.
Table 4

Glycidyl fatty acid ester results determined by 17 laboratories in a collaborative study

Sample ID

Value

Glycidyl fatty acid ester

  

C16:0

C18:0

C18:1

C18:2

C18:3

Total

Extra virgin olive oil

 

N

      
 

Mean (mg/kg)

NQa

NQ

NQ

NQ

NQ

NQ

Spiked extra virgin olive oil

 

N

17

17

17

17

17

17

 

Mean (mg/kg)

3.94

3.91

4.70

4.52

4.31

21.29

 Repeatability

s(r)

0.35

0.34

0.38

0.37

0.49

1.86

 

RSD(r)

8.88

8.62

7.99

8.09

11.28

8.72

 

r

0.98

0.94

1.05

1.02

1.36

5.20

 Reproducibility

s(R)

0.81

0.69

0.78

0.91

1.05

3.96

 

RSD(R)

20.47

17.62

16.58

20.06

24.34

18.61

 

R

2.26

1.93

2.18

2.54

2.94

11.09

Rapeseed oil

 

N

  

15

  

14

 

Mean (mg/kg)

NQ

NQ

0.45

NQ

NQ

0.52

 Repeatability

s(r)

  

0.17

  

0.20

 

RSD(r)

  

36.44

  

37.67

 

r

  

0.46

  

0.55

 Reproducibility

s(R)

  

0.20

  

0.22

 

RSD(R)

  

43.77

  

42.42

 

R

  

0.56

  

0.62

Soybean oil

 

N

   

15

 

14

 

Mean (mg/kg)

NQ

NQ

NQ

0.48

NQ

0.56

 Repeatability

s(r)

   

0.08

 

0.09

 

RSD(r)

   

17.45

 

16.08

 

r

   

0.23

 

0.25

 Reproducibility

s(R)

   

0.21

 

0.23

 

RSD(R)

   

43.51

 

41.39

 

R

   

0.58

 

0.64

Corn oil

 

N

9

 

15

16

 

15

 

Mean (mg/kg)

0.20

NQ

0.82

1.49

NQ

2.51

 Repeatability

s(r)

0.03

 

0.06

0.10

 

0.15

 

RSD(r)

13.71

 

7.57

6.85

 

5.95

 

r

0.08

 

0.15

0.29

 

0.42

 Reproducibility

s(R)

0.11

 

0.15

0.36

 

0.45

 

RSD(R)

53.61

 

18.35

23.97

 

17.95

 

R

0.31

 

0.42

1.00

 

1.26

Palm oil (1)

 

N

17

12

17

17

 

17

 

Mean (mg/kg)

2.34

0.50

5.11

1.33

NQ

9.26

 Repeatability

s(r)

0.46

0.08

0.96

0.26

 

1.92

 

RSD(r)

19.88

16.95

18.83

19.17

 

20.77

 

r

1.30

0.24

2.69

0.72

 

5.38

 Reproducibility

s(R)

0.78

0.15

1.50

0.44

 

2.74

 

RSD(R)

33.16

29.31

29.45

33.32

 

29.58

 

R

2.17

0.41

4.21

1.24

 

7.67

Palm oil (2)

 

N

17

13

17

16

 

17

 

Mean (mg/kg)

2.37

0.48

5.12

1.35

NQ

9.40

 Repeatability

s(r)

0.34

0.10

0.68

0.21

 

1.42

 

RSD(r)

14.49

19.91

13.27

15.90

 

15.14

 

r

0.96

0.27

1.90

0.60

 

3.99

 Reproducibility

s(R)

0.63

0.14

1.13

0.35

 

2.47

 

RSD(R)

26.64

29.76

21.99

26.18

 

26.31

 

R

1.77

0.40

3.15

0.99

 

6.93

Palm oil (3)

 

N

17

14

17

17

 

17

 

Mean (mg/kg)

2.39

0.49

5.37

1.39

NQ

9.44

 Repeatability

s(r)

0.23

0.09

0.51

0.15

 

0.94

 

RSD(r)

9.84

17.42

9.45

10.51

 

10.00

 

r

0.66

0.24

1.42

0.41

 

2.64

 Reproducibility

s(R)

0.54

0.15

0.98

0.32

 

1.85

 

RSD(R)

22.79

30.10

18.30

22.88

 

19.58

 

R

1.52

0.42

2.75

0.89

 

5.17

Palm oil spiked with glycidyl stearate (C18:0)

 

N

17

17

15

17

 

17

 

Mean (mg/kg)

2.68

7.26

5.66

1.39

NQ

16.44

 Repeatability

s(r)

0.36

0.74

0.60

0.18

 

1.61

 

RSD(r)

13.50

10.24

10.63

13.20

 

9.79

 

r

1.01

2.08

1.68

0.51

 

4.50

 Reproducibility

s(R)

0.91

2.40

1.09

0.49

 

5.25

 

RSD(R)

33.97

33.08

19.33

35.52

 

31.93

 

R

2.55

6.72

3.06

1.38

 

14.70

Results obtained in a collaborative study organized by AOCS in 2011. Precision data calculated according to the AOAC/IUPAC Harmonized Protocol. Each laboratory was supplied with blind duplicates of 9 commercial oil samples

aNQ = no laboratory reported results above their individual LOQ

Table 5

Values of individual and total glycidyl fatty acid esters determined by study participants in the “spiked” sample at different times after preparation

Spiked level (mg/kg)

C16:0

C18:0

C18:1

C18:2

C18:3

Total

10

10

10

10

10

50

1st qualification study (month 3)

 No. of laboratories

20

20

20

20

20

20

 Mean

9.07

8.84

8.85

8.53

9.50

43.20

 Std dev

1.83

1.90

2.34

1.87

2.97

10.09

2nd qualification study (month 10)

 No. of laboratories

15

15

15

15

15

15

 Mean

5.32

5.04

5.92

5.77

5.37

27.96

 Std dev

0.57

0.63

0.71

0.75

0.86

3.26

Collaborative study (month 13)

 No. of laboratories

17

17

17

17

17

17

 Mean

3.94

3.91

4.70

4.52

4.31

21.29

 Std dev

0.81

0.69

0.78

0.91

1.05

3.96

Repeatability and reproducibility values generally were greater in palm oil samples than in the other oils. Repeatability (RSDr) ranged from 9.45 to 19.88 % and reproducibility (RSDR) ranged from 18.30 to 35.52 % for individual GE levels >0.5 mg/kg. Several factors may have played a role in the increased variability. Palm oil was the only “hard” oil included in this study, and heating is required to obtain a representative sample. However, extended exposure to high temperature (>5 °C above melt temperature) may cause decomposition or conversion of GE in the sample. Based on results from our laboratory, palm oil samples were free of any interfering compounds that would have caused peaks to be incorrectly identified (Fig. 1). However, matrix-related ion enhancement/suppression remains a possibility in spite of the clean-up procedures utilized.
https://static-content.springer.com/image/art%3A10.1007%2Fs11746-012-2187-7/MediaObjects/11746_2012_2187_Fig1_HTML.gif
Fig. 1

LC–MS selected ion monitoring chromatograms of 5 glycidyl esters. a Mixed standard, b palm oil sample, c palm oil sample spiked with glycidyl stearate

The international collaborative study confirmed the performance characteristics of the method over a wide range of GE concentrations. The method requirement of a single quadrupole mass spectrometer means that instrument costs are attainable for most dedicated, analytical laboratories. We suggest that this method be considered by relevant bodies as the first internationally validated method for GE analysis of edible oils.

Copyright information

© AOCS 2012