Process-induced formation of imidazoles in selected foods

Process-induced formation of imidazoles in selected foods

Food Chemistry 228 (2017) 381–387 Contents lists available at ScienceDirect Food Chemistry journal homepage: Proce...

1MB Sizes 77 Downloads 118 Views

Food Chemistry 228 (2017) 381–387

Contents lists available at ScienceDirect

Food Chemistry journal homepage:

Process-induced formation of imidazoles in selected foods Pascal Mottier ⇑, Claudia Mujahid, Adrienne Tarres, Thomas Bessaire, Richard H. Stadler Nestlé Research Center, Vers-chez-les-Blanc, 1000 Lausanne 26, Switzerland

a r t i c l e

i n f o

Article history: Received 27 October 2016 Received in revised form 3 February 2017 Accepted 6 February 2017 Available online 7 February 2017 Chemicals studied in this article: 4-methylimidazole (PubChem CID: 13195) 2-methylimidazole (PubChem CID: 12749) 2-acetyl-4-tetrahydroxybutylimidazole (PubChem CID: 108037)

a b s t r a c t The presence of 4-methylimidazole (4-MEI), 2-methylimidazole (2-MEI) and 2-acetyl-4-tetrahydroxybu tylimidazole (THI) in some foods may result from the usage of caramel colorants E150c and E150d as food additives. This study demonstrates that alkylimidazoles are also byproducts formed from natural constituents in foods during thermal processes. A range of heat-processed foods that are known not to contain caramel colorants were analyzed by isotope dilution LC-MS/MS to determine the contamination levels. Highest 4-MEI concentrations (up to 466 mg/kg) were observed in roasted barley, roasted malt and cocoa powders, with the concomitant presence of 2-MEI and/or THI in some cases, albeit at significantly lower levels. Low amounts of 4-MEI (<20 mg/kg) were also detected in cereal-based foods such as breakfast cereals and bread toasted to a brown color (medium toasted). The occurrence of 4-MEI in certain processed foods is therefore not a reliable indicator of the presence of the additives E150c or E150d. Ó 2017 Elsevier Ltd. All rights reserved.

Keywords: 4-methylimidazole (4-MEI) 2-methylimidazole (2-MEI) 2-acetyl-4-tetrahydroxybutylimidazole (THI) Thermal treatment Food LC-MS/MS

1. Introduction The presence of substituted imidazoles 4-methylimidazole (4MEI), 2-methylimidazole (2-MEI) and 2-acetyl-4-tetrahydroxybuty limidazole (THI) in foods may result from addition of ammonia caramel colorants E150c (Class III) and E150d (Class IV) (EFSA, 2011). Class III ammonia caramels are commonly added to a wide range of foods, such as bakery products, soy and brown sauces, gravies, soups, vinegars and beers, whilst Class IV sulphite ammonia caramels are added to soft drinks, pet foods and soups (IARC, 2012a). In the manufacturing process of such caramel colors, imidazoles (and others chemicals such as furan and 5-hydroxymethyl-2-furfural) are formed as byproducts, typically when ammonia or ammonium salts react with reducing sugars (EFSA, 2011). The California EPA’s Office of Environmental Health Hazard Assessment added 4-MEI to the list of probable carcinogens which require a warning label when exceeding the ‘‘No Significant Risk Level” (NSRL) of ⇑ Corresponding author. E-mail addresses: [email protected] (P. Mottier), [email protected] (C. Mujahid), [email protected] (A. Tarres), [email protected] (T. Bessaire), [email protected] (R.H. Stadler). 0308-8146/Ó 2017 Elsevier Ltd. All rights reserved.

29 mg/day, which corresponds to 0.4 mg/kg bw/day for a 70-kg adult (OEHHA, 2011). In 2012, the International Agency for Research on Cancer (IARC) classified both 4-MEI (IARC, 2012a) and 2-MEI (IARC, 2012b) as ‘‘possibly carcinogenic to humans (group 2B)”. Additionally, THI is suspected to be a strong immunosuppressant (EFSA, 2011). The European Union has set Maximum Residue Limits (MRLs) only for 4-MEI and THI in E150c caramel colorings, at 200 mg/kg and 10 mg/kg, respectively, and for 4-MEI in E150d caramel colorings at 250 mg/kg. These limits are expressed on equivalent color basis, i.e. expressed in terms of a product having a color intensity of 0.1 absorbance units (EU, 2012). Recently, the Belgian Federal Public Service for Health launched a study whose main objective was the determination of THI and 4-MEI intake by the Belgian population through caramel colorants on the basis of real concentrations in food and in relation to food consumption data. In total, 522 food samples were analyzed for the occurrence of 4-MEI, 2-MEI, and THI (Vanermen et al., 2015). The imidazoles were detected in all foods containing caramel colors as additives but also in some foods not mentioning Class III or Class IV caramel in the ingredients list. This led to the assumption that the labelling may not always be in correspondence with the food composition, which could be interpreted as mislabeling by the food business operator.


P. Mottier et al. / Food Chemistry 228 (2017) 381–387

Alkylimidazoles can be synthesized by condensation of a dicarbonyl compound with ammonia and an aldehyde (Radzisewski, 1882). As pointed out by Hengel and Shibamoto (2013), the DebusRadziszewski reaction pathway suggests that degradation products of sugars, but also lipids, amino acids, or proteins are ideal precursors of imidazoles. Upon thermal degradation sugars and lipids can provide alkyl dicarbonyls and alkyl ketones, whilst amino acids release ammonia and alkyl carbonyls via Strecker degradation (Strecker, 1862). The formation of reactive dicarbonyls (glyoxal, methylglyoxal) and diacetyl after heat treatment of lipids was shown by Jiang, Hengel, Pan,, Seiber, and Shibamoto (2013). Model system studies comprised of dicarbonyls and aldehydes in the presence of ammonium sulfate afford 2-MEI and 4-MEI (Wu, Huang, Kong, & Yu, 2015). THI has been found in class I, III and IV caramel colorants (Elsinghorst, Raters, Dingel, Fischer, & Matissek, 2013; Wang et al., 2016). Generally, formation of THI is via iminofructosamine, an intermediate of sugar breakdown products that have reacted with ammonia. Condensation with concomitantly formed methylglyoxal provides the imidazole heterocycle (Kröplien, Rosdorfer, van der Greef, Long, & Goldstein, 1985). Substituted imidazoles can also be generated during the manufacture of certain heat-processed foods or beverages that do not contain caramel colorings in the recipe. Roasted coffee (Casal, Fernandes, Oliveira, & Ferreira, 2002; Klejdus, Moravcová, Lojková, Vacek, & Kubán, 2006; Lojková, Klejdus, Moravcová, & Kubán, 2006), coffee substitutes (Cunha, Senra, Cruz, Casal, & Fernandes, 2016), licorice (Raters, Elsinghorst, Goetze, Dingel, & Matissek, 2015) and naturally brewed soy sauce (Yamaguchi & Masuda, 2011) are examples of foods and beverages that contain alkylimidazoles but not caramel colorants. This research describes an isotope dilution liquid chromatography tandem mass spectrometry (LC-MS/MS) procedure to determine 4-MEI, 2-MEI and THI in a wide range of foods, validated according to the EU guidance document SANTE/11945/2015 (SANTE, 2016). We also demonstrate that imidazoles are byproducts of processing in several foods, with the consequence that 4MEI is not a reliable indicator of the presence of E150c/E150d in a recipe. 2. Material and methods 2.1. Chemicals and reagents 4(5)-Methylimidazole (4-MEI), 2-Methylimidazole (2-MEI) and 2-Acetyl-4(5)-tetrahydroxybutylimidazole (THI) were obtained from Sigma Aldrich (Fluka, Buchs, Germany). The isotopic labelled internal standards 4-MEI-2H6 (chemical purity 98.1%, isotopic purity 98.7%) and 2-MEI-2H6 (chemical purity 99%, isotopic purity 98.6%) were from C/D/N Isotopes Inc. (Pointe-Claire, Quebec, Canada) whilst THI-13C6 (chemical purity > 99%, isotopic purity > 98.6%) was purchased from ELFI Analytik GbR (Neufahrn, Germany). LC-MS grade solvents (water, methanol, 2-propanol) and ammonium hydroxide solution (25%) were from Merck (Darmstadt, Germany). Sigma Aldrich provided ammonium formate and concentrated formic acid. Solid phase extraction (SPE) cartridges Oasis MCX (6 cc/500 mg) were purchased from Waters (Milford, Il, USA). 2.2. Standard solutions Stock standard solutions of both unlabelled and labelled analytes (1 mg/mL) were prepared in water. Complete solubility was ensured by vortexing and sonication (i.e. up to 15 min for THI). Working standard solutions at 4.0, 1.0 and 0.4 mg/mL in water were then obtained by successive dilutions. All solutions were stored at 20 °C and attained room temperature before use.

2.3. Samples Infant cereals, breakfast cereals, and chocolate were either manufacturing line samples or collected from supermarkets in Switzerland. Sandwich breads, brown sugar (from sugarcane) and honey were purchased from local supermarkets. Roasted malt and roasted barley samples (for beer production) were obtained from a malting company in Belgium, and their respective roasting intensity was evaluated via color measurement (see chapter 2.9). Cocoa products (raw and roasted beans, powder, liquor and butter), roasted nuts, grilled almonds and toasted coconut chips were from various suppliers. Sandwich breads were toasted to different toasting degrees using a kitchen toaster. Non-powdered samples (breakfast cereals, sandwich breads, nuts, almonds, coconut chips and cocoa products) were all ground using a cryogenic grinder (6875D Freezer/Mill, SpexSamplePrep, Stanmore, UK) before analysis. 2.4. Sample preparation Ground or already powdered sample (4 g) was weighed into a 50-mL polypropylene tube (Becton Dickinson, Le Pont de Claix, France) and supplemented with labelled internal standards (25 mL of 4 mg/ mL solution, corresponding to 25 mg/kg equivalent-in-sample concentration). The tube was allowed to stand for equilibration for 10 min. Methanol containing 2% formic acid (40 mL) and a ceramic homogenizer were then added. The slurry was shaken with an automated shaker (GenoGrinder 2010, SPEX SamplePrep) at 1500 rpm for 3 min and subsequently centrifuged at 4000g at room temperature for 10 min (centrifuge Heraeus Multifuge X3R, Thermo Scientific, Reinach, Switzerland). An aliquot of the supernatant (2.5 mL) was added to water containing 2% formic acid (2.5 mL). SPE cleanup was conducted by conditioning an Oasis MCX cartridge with methanol (6 mL) then with methanol: water containing 2% formic acid (1:1 v/v, 6 mL). The extract solution containing the analytes (5 mL) was then loaded onto the cartridge. After penetration, the cartridge was first washed with methanol: water containing 2% formic acid (1:1 v/v, 6 mL) then with methanol (6 mL) and excess solvent was removed by vacuum suction. The target analytes were eluted with methanol containing 1.25% ammonium hydroxide (6 mL) and collected in a 15-mL polypropylene tube. The solution was evaporated to dryness under a stream of nitrogen at 40 °C, then reconstituted in water containing 0.1% formic acid (0.5 mL) and transferred into a 2-mL Eppendorf tube. The resulting solution was centrifuged (17,000g, 5 min, 4 °C, centrifuge Heraeus Frisco 17, Thermo Scientific) prior to LC-MS/MS analysis. 2.5. LC-MS/MS analysis HPLC analysis was performed using a Hypercarb Porous Graphitic Carbon HPLC column (2.1  100 mm, 3 mm, Thermo Scientific) thermostated at 50 °C using an Agilent 1290 Infinity binary pump system (Agilent, Geneva, Switzerland). The mobile phase was comprised of water containing 20 mM of ammonium formate and 0.1% formic acid (solvent A) and methanol containing 20 mM of ammonium formate and 0.1% formic acid (solvent B). A gradient program was set up as follows: 0–1 min with 100% A; 1–2.25 min linear gradient down to 20% A; hold at 20% A for 0.75 min; return to 100% A in 0.1 min and hold at 100% A for 4.9 min (total run time 8 min). The flow rate was 0.5 mL/min and 10 lL was injected onto the column. The HPLC flow was directed into the MS detector between 0.1 and 5 min. MS detection was performed using a Sciex 5500 detector (Foster City, USA) equipped with a Turbo VTM ionization source. MS parameters were set in positive (4-MEI, 2-MEI, 4-MEI-2H6, 2MEI-2H6) or negative (THI, THI-13C6) electrospray ionization (ESI)


P. Mottier et al. / Food Chemistry 228 (2017) 381–387

mode by separately syringe-infusing standard solutions at 1 mg/mL (syringe flow rate of 10 mL/min) along with the HPLC flow at 0.5 mL/min using a T connector. The HPLC flow was comprised of 50% A and 50% B during MS parameter settings. Two acquisition periods were used: Period 1 (0.1–2.7 min) for 4-MEI, 2-MEI and their related internal standard in positive ionization mode and Period 2 (2.7–5 min) for THI and THI-13C6 in negative ionization mode. The block source temperature was maintained at 550 °C and gas values were set as follows: curtain gas 30 psi; nebulizer gas 50 psi; turbo gas 50 psi; collision gas on medium. Other parameters were (positive or negative ionization): ion spray voltage (5 kV or 3.5 kV), entrance potential (10 V or 10 V), and collision exit potential (13 V or 15 V). Quantitative analysis was performed using tandem MS in selected reaction monitoring (SRM) mode alternating two transition reactions for each analyte and its related internal standard with a dwell time of 30 ms. Transition reactions monitored were: 4-MEI (declustering potential DP (V) and collision energy CE (eV) are indicated in brackets): m/z 83.0 ? 56.0 (DP 40, CE 25, target); m/z 83 ? 28 (CE 35, qualifier); 4-MEI-2H6: m/z 88 ? 60 (DP 50, CE 25, target); m/z 88 ? 61 (CE 25, qualifier); 2-MEI: m/z 83 ? 42 (DP 40, CE 25, target); m/z 83 ? 68 (CE 35, qualifier); 2-MEI-2H6: m/z 88 ? 44 (DP 50, CE 29, target); m/z 88 ? 61 (CE 30, qualifier); THI: m/z 229 ? 151 (DP - 40, CE - 20, target); m/z 229 ? 123 (CE - 32, qualifier); THI-13C6: m/z 235 ? 155 (DP - 30, CE - 20, target); m/z 235 ? 126 (CE - 33, qualifier). Data acquisition was carried out using Analyst software 1.5.2 and subsequent data processing done using Multiquant software 3.0 (both from Sciex). 2.6. Quantification The analytes were quantified by isotopic dilution (analyte/IS area ratio (=y) versus analyte/IS concentration ratio (=x)) using seven calibration levels ranging from 0 to 100 ng/mL in water, each level containing ISs at 12.5 ng/mL. This concentration range corresponded to 0 to 200 mg/kg equivalent-in-sample concentrations (ISs at 25 mg/kg). The final equation used to express imidazoles (in mg/kg) was as follows (Bessaire et al., 2015):


Aa Ais



mis  ma

where Aa is the peak area of the analyte in the sample; Ais is the peak area of the IS in the sample; I and S are the intercept and slope of the regression line, respectively; mis is the mass of IS added to the test portion, in ng; ma is the mass of the test portion, in g. Linearity of responses was checked by calculating the relative standard deviation (RSD) of response factors (RF, RF = y/x) which should be RSDRF < 15% (Rodriguez & Orescan, 1998). When the imidazoles were found at amounts out of the calibration range, the related samples were reanalyzed by reducing the test portion size accordingly. 2.7. Confirmation criteria Analytes were considered to be positively identified when the following criteria were met simultaneously (SANTE, 2016): (a) the ratio of the chromatographic retention time of the analyte to that of its IS, i.e. the relative retention time, corresponded to that of the average relative retention time of the calibration solutions within a ±0.1 min tolerance; and (b) the variability of ion ratios for the different transition reactions recorded for both analyte and its corresponding IS were within ±30% (relative) of average of calibration standards from the same sequence.

2.8. Method validation Validation was conducted according to an approach widely used for pesticide residues analysis (SANTE, 2016). In summary, a cereal-based sample (breakfast cereals) was spiked at 10 mg/kg (target LOQ) and 50 mg/kg (5  LOQ). Precision data were obtained by two operators, each performing six replicates at the mentioned fortification levels over two different days (leading to a total of 12 separate experiments for each fortification level). Both nonfortified samples and reagent blanks were analyzed in duplicate as well to verify the absence or presence of the analytes at low levels before fortification trials. A malted chocolate powder was spiked at 20 mg/kg and 100 mg/kg and analyzed similarly to generate performance data on cocoa-based samples. Recovery, repeatability SD(r) and intermediate reproducibility SD(iR) precision data were calculated according to ISO 5725–2 document (ISO, 1994). Recovery values at the fortified concentrations were calculated from the median of spiked experiments performed under iR conditions. The overall uncertainty at each fortification level was obtained by combining precision and recovery contributions as proposed by Barwick and Ellison (Barwick & Ellison, 2000):

Uð%Þ ¼ 2  RSDðuÞ ¼ 2 

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2 RSDðiRÞ þ RSDðRecÞ2

Where: U(%) is the expanded uncertainty at the 95% confidence interval; RSD(u) is the relative standard deviation of uncertainty; RSD(iR) is the relative standard deviation of intermediate reproducibility and RSD(Rec) is the relative standard deviation of recovery. 2.9. Color measurement Color measurement of malt roasting intensity was assessed using a Color Flex CX1051 instrument from Hunterlab (Reston, VA, USA) as described by Bessaire et al. (2016). The single light parameter CbL was used to assess the influence of L * (level of light or dark) and b⁄ (from blue to yellow) by combining the two individual parameters together using the following equation:

C bL

sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi   2   2 b L ¼ þ   Lmax bmax

3. Results and discussion 3.1. Analytical procedure The imidazoles were first analyzed in positive and negative ESIMS modes to optimize the MS conditions. Both 2-MEI and 4-MEI showed prominent protonated molecular ions [M+H]+. Under our conditions, THI exhibited a predominant deprotonated molecular ion [MH] although other authors have conducted analysis of THI in the positive ESI mode (Lojková et al., 2006; Raters et al., 2015; Schlee et al., 2013; Vanermen et al., 2015; Wang et al., 2016). Collision-induced dissociation (CID) mass spectra of these precursors were then recorded at various collision energies before selecting the optimal fragment ions, MS/MS transition reactions and related instrument parameters. Thus, the product ion spectrum of 2-MEI exhibited three main peaks at m/z 42 (loss of CH3CN), 56 (loss of HCN) and m/z 68 (loss of CH3). The transition reactions m/z 83 ? 42 and m/z 83 ? 68 were chosen for 2-MEI quantification and confirmation, respectively. Similarly, the 4MEI product ion spectrum was dominated by ions at m/z values of 28 ([C2H4]+), 42 and 56. The transitions m/z 83 ? 56 and m/z 83 ? 28 were applied for quantification and confirmation, respectively. As already noticed by Ratnayake, Halldorson, Bestvater, & Tomy, 2016, ESI + spectra of 2-MEI-2H6 and 4-MEI 2H6 labelled


P. Mottier et al. / Food Chemistry 228 (2017) 381–387

compounds exhibited a parent ion at m/z 88 (rather than at m/z 89), evidencing a 2H/H exchange and leading to a down-mass shift of 1. The transitions reactions selected were therefore m/z 88 ? 44 and m/z 88 ? 61 for 2H6-2-MEI and m/z 88 ? 60 and m/z 88 ? 61 for 2H6-4-MEI, with fragments at m/z 60 and m/z 61 representing [C22H4N2]+ and [C23H5HN]+, respectively. In negative ESI, the product ion spectrum of THI showed two main peaks at m/z 151 (combined loss of water and glycolaldehyde) and m/z 123 (subsequent loss of CO from m/z 151) which were used to set the quantification and confirmation transition reactions. Attribution to these fragment losses was consistent with the main product ions observed when tuning 13C6-THI . Since both 2-MEI and 4-MEI (and their related IS) share similar transition reactions, an efficient chromatographic separation was required for correct quantification. Optimal conditions were achieved with a Hypercarb column (flat sheets of hexagonally arranged carbon atoms) employing a 8-min LC gradient. The imidazole 2-MEI eluted first at retention time 1.45 min, comfortably longer than the void volume of the column (equivalent to 0.48 min), thereby ensuring efficient separation of 2-MEI and early-eluting impurities that may contribute to signal suppression. Typical chromatograms of imidazoles under the described conditions are illustrated in Fig. 1A (see Material and Methods part). Similar to Wang and Schnute (2012), but in contrast to Vanermen et al. (2015) and Wang et al. (2016), a high background noise was observed when using acetonitrile as organic solvent for

the LC gradient. Interference from the acetonitrile cluster ion [2 M +H]+, that incidentally has the same m/z 83 as the 2-MEI and 4-MEI precursor ions, may lead to this background noise. Therefore, we chose methanol as LC solvent, which enabled detection of the target analytes down to low mg/kg concentrations. The extraction procedure described in this study is comparable to that of Vanermen et al. (2015), encompassing extraction with acidic methanol followed by cleanup on cation-exchange SPE. An organic/aqueous solvent exchange is usually recommended before the SPE step and requires a dryness step with the concomitant risk to lose the analytes upon evaporation. This risk was avoided by simply diluting the methanol extract with acidified water prior to SPE, enabling sufficient retention of the analytes. 3.2. Method performance characteristics The efficiency of the quantification by the isotope dilution approach was first verified over the 0–200 lg/kg range by comparing solvent-based and matrix-matched calibration curves (Fig. 1B). Only minor differences in the slopes were noticed when compared to the solvent-based calibration, demonstrating the importance of the use of labelled isotopomers as internal standards that allow compensation for matrix effects. All matrix-matched or solventbased calibration curves followed a linear model with r2 > 0.99 and residual deviations below 10%, which fulfils SANTE/11945/2015 requirements (SANTE, 2016). Analytical per-

Fig. 1. A) LC-MS/MS chromatograms of 4-MEI, 2-MEI and THI and relative internal standard in a breakfast cereals sample spiked at the LOQ level (10 mg/kg–IS at 25 mg/kg). Target SRM and qualifier SRM are depicted in blue and in pink, respectively .B) Solvent-based and matrix-matched calibration curves for 4-MEI, 2-MEI and THI in a breakfast cereals and a malted chocolate powder sample. For clarity reasons, individual data points are not shown.


P. Mottier et al. / Food Chemistry 228 (2017) 381–387 Table 1 Method Performance Data. Fortification at Low Level (LL) Level (HL) f

a b c d e f


and High

Rec (%)


RSD(r) (%)


RSD(iR) (%)


U (%)











Breakfast Cereals Malted Chocolate Powder

106 104

108 102

16 6

5 4

17 7

6 4

49 21

16 13


Breakfast Cereals Malted Chocolate Powder

96 85

91 89

6 6

4 2

5 5

5 3

15 14

15 10


Breakfast Cereals Malted Chocolate Powder

114 99

110 99

4 2

4 1

15 2

8 3

45 12

24 10

Recovery values were calculated from the median of spiked experiments performed under intermediate reproducibility (iR) conditions (n = 12); Relative standard deviation of the median of data obtained under repeatability conditions (n = 12) Relative standard deviation of the median of data obtained under iR conditions (n = 12); Expanded uncertainty limit given at the 95% confidence interval level; LL (Low Level) corresponds to 10 mg/kg for breakfast cereals and 20 mg/kg for malted chocolate powder; HL (High Level) corresponds to 50 mg/kg for breakfast cereals and 100 mg/kg for malted chocolate powder.

formance data at two fortification levels (10 and 50 mg/kg for breakfast cereals, and 20 and 100 mg/kg for malted chocolate powder) were RSD(r)  16%, RSD(iR)  17%; recovery ranged from 85 to 114% (Table 1). Additional recovery figures were generated by systematically analysing the surveyed foods as unspiked and spiked samples. Bread, nuts & almonds, coconut chips, sugar and cone biscuit were fortified at 10 mg/kg and median recoveries were 85%, 89% and 102% for 4-MEI, 2-MEI and THI respectively (n = 19). Similarly, breakfast cereals, infant cereals, cocoa-based products and honey were fortified at the 50 mg/kg level and median recovery data were 112%, 88% and 102% for 4-MEI, 2-MEI and THI respectively (n = 48). Median recovery data for roasted malt and roasted barley at the 100 mg/kg fortification level were 93%, 96% and 96% for 4-MEI, 2-MEI and THI respectively (n = 12). Laboratory LOQ was set at 10 mg/kg for each imidazole in all matrices except for roasted malt and roasted barley (LOQ 20 mg/kg). 3.3. Survey of food products Cocoa-based foods: Cocoa powder is a main ingredient in chocolate products and in cocoa/chocolate powder for beverages. The manufacture of cocoa powder entails heat steps (ICCO, 2013). To assess if thermal treatments may generate imidazoles, cocoa powders (n = 16) from different suppliers but also roasted and

non-roasted cocoa beans, cocoa butter, cocoa liquor and chocolate (>65% cocoa content) were included in the study. Table 2 and Fig. 2 (box-whisker plots) summarize the results obtained on the cocoa powders (alkalized or non treated, i.e. natural), demonstrating systematic presence of 4-MEI (range 68 to 320 mg/kg), as well as the presence of 2-MEI (range < 10 to 64 mg/kg) and THI (range < 10 to 80 mg/kg) in 90% of samples. No significant difference was found between the different powders, indicating that the alkalization process is most likely not the determining factor for imidazole formation. Imidazoles were not detected in cocoa beans (either nonroasted or roasted) and cocoa butter, whilst roasted cocoa nibs and cocoa liquor contained 4-MEI at detectable amounts. Related finished products including chocolate and cocoa/chocolate powders for beverages showed 4-MEI at much lower levels (<10 to 42 mg/ kg) as compared to cocoa powders. THI was found only in one chocolate powder (13 mg/kg) and 2-MEI was not detected at all. In this study, the one sample of roasted cocoa nibs that was analyzed for the presence of alkylimidazoles revealed quantifiable amounts of 4-MEI. The roasting step therefore contributes to the formation of imidazoles, but the amount seems low when compared to the levels of 4-MEI detected in cocoa powders. During milling, refining, and conching, temperatures of 80–90 °C are achieved (Stadler, Hughes, & Guillaume-Gentil, 2014), and these may be sufficient to generate additional alkylimidazoles, warrant-

Table 2 Survey of Alkylimidazoles in non-caramel containing foods. Sample Description

Nb. of Samples

Sandwich breads, non-toasted Sandwich breads, low-toasted (golden color) Sandwich breads, medium-toasted (brown) Sandwich breads, highly-toasted (dark brown) Malt, roasted (for beer production, 1 supplier) a) Barley, roasted (for beer production, 1 supplier) Infant cereals (2 suppliers) Cone biscuit for ice cream (1 supplier) Breakfast cereals (2 suppliers) Roasted peanuts, grilled almonds, roasted hazelnuts, toasted coconut chips (6 suppliers) Sugarcane sugar (2 suppliers) Honey (1 supplier) Cocoa powder (7 suppliers) b) Cocoa beans, raw and roasted (1 supplier) Cocoa nibs, roasted (1 supplier) Cocoa liquor (1 supplier) Cocoa butter (1 supplier) Chocolate (1 supplier) Cocoa and chocolate powders for beverage (1 supplier)

3 3 3 3 11 1 4 1 11 6 2 1 16 2 1 1 2 2 3

a), b)

Spread of data is shown in Figs. 4 and 2, respectively.

Concentrations of analytes (Min–Max, mg/kg) 4-MEI



<10 <10 <10–18 12–61 <20–323 466 <10 <10 <10–15 <10 <10 <10 68–320 <10 19 17 <10 15; 24 <10–42

<10 <10 <10 <10–13 <20–135 135 <10 <10 <10 <10 <10 <10 <10–64 <10 <10 <10 <10 <10 <10

<10 <10 <10 <10 <20–55 <20 <10 <10 <10 <10 <10, 18 <10 <10–80 <10 <10 <10 <10 <10 <10–13


P. Mottier et al. / Food Chemistry 228 (2017) 381–387

Fig. 2. Box-whisker plots of samples of cocoa powder (n = 16). Boxes represent quartiles 1 and 3; the median is indicated by the horizontal line inside the box; whiskers represent the upper and lower 25% of the distribution.

Fig. 3. Evidence of 2- & 4-MEI formation in sandwich bread upon toasting. Values are expressed in mg/kg.

ing further study across the cocoa processing line. In addition, more representative data is required for roasted beans and cocoa nibs, due to the very limited data currently available. 3.4. Cereal-based foods Ìt can be expected that toasting bread leads to the formation of imidazoles via Maillard driven pathways. Therefore, three breads

of different cereal composition, i.e. Bread 1 (wheat), Bread 2 (wheat, barley) and Bread 3 (wheat, barley, corn, spelt, rye and emmer) were analyzed prior and after a toasting step. As shown in Table 2 and Fig. 3, all highly-toasted breads (toasted to a very dark color) contain 4-MEI (range 12 to 61 mg/kg). Those toasted breads with the highest concentrations of 4-MEI also contained 2-MEI (12 and 13 mg/kg). THI was not detected in any of the toasted bread samples. These results on cereal-based foods prompted us to extend our study and encompass malt grains, that were roasted at different temperatures (between 110 to 230 °C) for subsequent beer production. Previous work in our laboratory demonstrated the formation of mepiquat (known as a growth regulator used in agriculture) in malt upon roasting (Bessaire et al., 2016). As anticipated and shown in Fig. 4, color measurements (CbL parameter) of the roasted malts revealed a positive correlation between color (correlated to intensity of thermal load) and 4-MEI content (<20– 323 mg/kg). The trend was comparable to that of mepiquat and acrylamide formation (Bessaire et al., 2016; Mizukami, Yoshida, Isagawa, Yamazaki, & Ono, 2014), i.e. the lighter the color (i.e. the lower the thermal load) the lower the concentration of process contaminant(s). As in the case of toasted breads, 2-MEI (90 and 135 mg/kg) was only found in malt samples with the highest amounts of 4-MEI (283 and 323 mg/kg). A general trend across the study is a systematically lower amount of 2-MEI compared to 4-MEI, which is corroborated as a general trend in foods analyzed for their imidazoles content. Possibly 2-MEI is furnished via one single chemical reaction pathway compared to two pathways that can operate for 4-MEI (Wu et al., 2015), leading to a higher abundance of 4-MEI. Of the eleven roasted malt samples analyzed, THI was found in only three samples at concentrations ranging from 25 to 51 mg/kg. No evidence of a positive correlation with 4-MEI could be determined, suggesting an alternative formation pathway for THI. In agreement with the malt extract data, analysis of a roasted barley sample (for beer production) confirmed that the cereal roasting step determines the formation of imidazoles (Table 2). A limited sampling of infant cereals and cone biscuit for ice cream failed to show the presence of imidazoles at the method LOQ, most likely due to relatively milder thermal treatment. Trace amounts of 4-MEI were detected in breakfast cereals (up to 15 mg/kg), in four brands containing cocoa and/or almonds. Grilled almonds but also roasted peanuts & hazelnuts, toasted coconut chips, sugarcane sugar and honey are frequent ingredients in breakfast cereals recipes, and could be expected to furnish precursor molecules upon thermal processing. However, none of the targeted imidazoles were found in the samples under study, with the exception of THI that could be quantified (18 mg/kg) in a sugar (from sugarcane) sample.

Fig. 4. Correlation between 4-MEI levels and malt roasting intensity (color measurement CbL, n = 11).

P. Mottier et al. / Food Chemistry 228 (2017) 381–387

4. Conclusion This study demonstrates that 4-MEI, 2-MEI and THI can be ‘‘naturally” generated through thermal processing of foods. Because alkylimidazoles are thermally-induced byproducts, they can be considered unsuitable as markers or indicators that demonstrate the addition of the colorants E150c and E150d. Along with coffee products, licorice and soy sauces, evidence is provided for the first time that processed cereal- and cocoa-based ingredients contain 4MEI, 2-MEI and THI. On the other hand, concentrations found in finished products, i.e. infant cereals, breakfast cereals, bread (when not toasted excessively to a very dark color) and chocolate, were all below 50 mg/kg. Further research is needed to assess the possible wider occurrence of alkylimidazoles, albeit at very low levels, on exposure via the diet, and to better understand the precise role of natural food constituents in the chemical formation pathways.

References Barwick, V. J., & Ellison, S. L. R. (2000). Valid analytical measurement (VAM) project: Development and harmonization of measurement uncertainty principles. Part d: Protocol for uncertainty evaluation from validation data. LGC/VAM/1998/088. Teddington: LGC Limited. Bessaire, T., Tarres, A., Goyon, A., Mottier, P., Dubois, M., Tan, W. P., & Delatour, T. (2015). Quantitative determination of sodium monofluoroacetate ‘‘1080” in infant formulas and dairy products by isotope dilution liquid chromatography tandem mass spectrometry. Food Additives and Contaminants, 32, 1885–1892. Bessaire, T., Tarres, A., Stadler, R. H., Wermann, S., Hofmann, J., Theurillat, V., ... Delatour, T. (2016). Mepiquat: A process-induced byproduct in roasted cerealbased foodstuffs. Journal of Agricultural and Food Chemistry, 64, 1185–1190. Casal, S., Fernandes, J. O., Oliveira, M. B., & Ferreira, M. A. (2002). Gas chromatographic-mass spectrometric quantification of 4-(5-)methylimidazole in roasted coffee after ion-pair extraction. Journal of Chromatography A, 976, 285–291. Cunha, S., Senra, L., Cruz, R., Casal, S., & Fernandes, J. O. (2016). 4-Methylimidazole in soluble coffee and coffee substitutes. Food Control, 63, 15–20. EFSA (2011). Scientific opinion on the re-evaluation of caramel colours (E 150 a,b,c, d) as food additive EFSA Panel on Food Additives and Nutrient Sources added to Food (ANS). EFSA Journal, 9(3). scientific_output/files/main_documents/2004.pdf/Accessed October 2016. Elsinghorst, P. W., Raters, M., Dingel, A., Fischer, J., & Matissek, R. (2013). Synthesis and application of (13)C-labeled 2-acetyl-4-((1R,2S,3R)-1,2,3,4tetrahydroxybutyl)imidazole (THI), an immunosuppressant observed in caramel food colorings. Journal of Agricultural and Food Chemistry, 61, 7494–7499. EU (2012). Commission Regulation No 231/2012 of 9 March 2012 laying down specifications for food additives listed in Annexes II and III to Regulation (EC) No 1333/2008 of the European Parliament and of the Council. Official Journal of the European Union, L83/1 – 295. PDF/?uri=CELEX:32012R0231&from=EN/Accessed October 2016. Hengel, M., & Shibamoto, T. (2013). Carcinogenic 4(5)-methylimidazole found in beverages, sauces, and caramel colors: Chemical properties, analysis, and biological activities. Journal of Agricultural and Food Chemistry, 61, 780–789. IARC (2012a). International Agency for Research on Cancer. Monographs on the Evaluation of Carcinogenic Risks to Humans (2012). Volume 101: Some chemicals present in industrial and consumer products, food and drinking water - 4-methylimidazole. vol101/mono101-015.pdf/Accessed October 2016 IARC (2012b). International Agency for Research on Cancer. Monographs on the Evaluation of Carcinogenic Risks to Humans. Volume 101: Some chemicals present in industrial and consumer products, food and drinking water - 2methylimidazole. mono101-014.pdf/Accessed October 2016. ICCO (2013). International Cocoa Organization. Processing Cocoa http://www.icco. org/about-cocoa/processing-cocoa.html/Accessed October 2016. ISO (1994). ISO 5725-2:1994. Accuracy (Trueness and Precision) of measurement methods and results; Part 2: basic method for the determination of repeatability and reproducibility of a standard measurement method. Jiang, Y., Hengel, M., Pan, C., Seiber, J. N., & Shibamoto, T. (2013). Determination of toxic a-dicarbonyl compounds, glyoxal, methylglyoxal, and diacetyl released to the headspace of lipid commodities. upon heat treatment. Journal of Agricultural and Food Chemistry, 61, 1067–1071.


Klejdus, B., Moravcová, J., Lojková, L., Vacek, J., & Kubán, V. (2006). Solid-phase extraction of 4(5)-methylimidazole (4MeI) and 2-acetyl-4(5)-(1,2,3,4tetrahydroxybutyl)-imidazole (THI) from foods and beverages with subsequent liquid chromatographic-electrospray mass spectrometric quantification. Journal of Separation Science, 29, 378–384. Kröplien, U., Rosdorfer, J., van der Greef, J., Long, R. C., & Goldstein, J. H. (1985). 2Acetyl-4(5)-(1,2,3,4-tetrahydroxybutyl)imidazole detection in commercial caramel color III and preparation by a model. browning reaction. Journal of Organic Chemistry, 50, 1131–1133. Lojková, L., Klejdus, B., Moravcová, J., & Kubán, V. (2006). Supercritical fluid extraction (SFE) of 4(5)-methylimidazole (4-MeI) and 2-acetyl-4(5)-(1,2,3,4)tetrahydroxybutyl-imidazole (THI) from ground-coffee with high-performance liquid chromatographic-electrospray mass spectrometric quantification (HPLC/ ESI-MS). Food Additives & Contaminants Part A. Chemistry, Analysis, Control, Exposure & Risk Assessment, 23, 963–973. Mizukami, Y., Yoshida, M., Isagawa, S., Yamazaki, K., & Ono, H. (2014). Acrylamide in roasted barley grains: Presence, correlation with colour and decrease during storage. Food Additives & Contaminants Part A. Chemistry, Analysis, Control, Exposure & Risk Assessment, 31, 995–1000. OEHHA (2011). Office of Environmental Health Hazard Assessment. No significant risk level (NSRL) for the proposition 65 carcinogen 4 methylimidazole. California Environmental Protection Agency. prop65/law/pdf_zip/100711_4MEIfindings.pdf/Accessed October 2016 Radzisewski, B. (1882). Ueber glyoxalin und seine homologe. Berichte der Deutschen chemischen Gesellschaft, 15, 2706. Raters, M., Elsinghorst, P. W., Goetze, S., Dingel, A., & Matissek, R. (2015). Determination of 2-methylimidazole, 4-methylimidazole, and 2-acetyl-4(1,2,3,4-tetrahydroxybutyl)imidazole in licorice using high-performance liquid chromatographytandem mass spectrometry stable-isotope dilution analysis. Journal of Agricultural and Food Chemistry, 63, 5930–5934. Ratnayake, G., Halldorson, T., Bestvater, L., & Tomy, G. T. (2016). Determination of 4 (5)-methylimidazole in carbonated beverages by isotope-dilution liquid chromatography-tandem mass spectrometry. Food Additives & Contaminants Part A. Chemistry, Analysis, Control, Exposure & Risk Assessment, 32, 1075–1082. Rodriguez, M., & Orescan, D. B. (1998). Confirmation and quantitation of selected sulfonylurea, imidazolinone, and sulfonamide herbicides in surface water using electrospray LC/MS. Analytical Chemistry, 70, 2710–2717. SANTE/11945/2015. Guidance document on analytical quality control and method validation procedures for pesticides residues analysis in food and feed. http:// en.pdf/Accessed October 2016. Schlee, C., Markova, M., Schrank, J., Laplagne, F., Schneider, R., & Lachenmeier, D. W. (2013). Determination of 2-methylimidazole, 4-methylimidazole and 2-acetyl4-(1,2,3,4-tetrahydroxybutyl)imidazole in caramel colours and cola using LC/ MS/MS. Journal of Chromatography B. Analytical Technologies in the Biomedical and Life Sciences, 927, 223–226. Stadler, R. H., Hughes, G., & Guillaume-Gentil, O. (2014). Safety of food and beverages: Coffee, tea and herbals, cocoa and derived products. In Y. Motarjemi (Ed.), Encyclopedia of Food Safety (Vol. 3, pp. 371–383). Waltham, MA: Academic Press. Elsevier Inc.. Strecker, A. (1862). Notiz über eine eigenthümliche Oxydation durch Alloxan [Notice of a curious oxidation by alloxan]. Annalen der Chemie und Pharmacie, 123, 363–365. Vanermen, G., Jacobs, G., Vloemans, P., Fierens, T., Van Holderbeke, M., Cornelis, C., De Maeyer, M., Sioen, I. (2015). Study of the intake of the contaminants THI and 4-MEI by the Belgian population through the food colours ammonia caramel and sulphite ammonia caramel. Study accomplished under the authority of Belgian FPS Health, Food Chain Safety and Environment and presented at the ‘‘Mass Spectrometry in Food and Feed II” conference (15.09.2015 - Ghent – Belgium Faculty of Pharmaceutical Sciences). October 2016. Wang, L., Ren, B., Liu, Y., Lu, Y., Chang, F., & Yang, L. (2016). 2-Acetyl-4tetrahydroxybutylimidazole and 4-methylimidazole in caramel colours, vinegar and beverages in China. Food Additives & Contaminants Part B, 8, 163–168. Wang, J., & Schnute, W. C. (2012). Simultaneous quantitation of 2-acetyl-4tetrahydroxybutylimidazole, 2and 4-methylimidazoles, and 5hydroxymethylfurfural in beverages by ultrahigh-performance liquid chromatography-tandem mass spectrometry. Journal of Agricultural and Food Chemistry, 60, 917–921. Wu, X., Huang, M., Kong, F., & Yu, S. (2015). Short communication: Study on the formation of 2-methylimidazole and 4-methylimidazole in the Maillard reaction. Journal of Dairy Sciences, 98, 8565–8971. Yamaguchi, H., & Masuda, T. (2011). Determination of 4(5)-methylimidazole in soy sauce and other. foods by LC-MS/MS after solid-phase extraction. Journal of Agricultural and Food Chemistry, 59, 9770–9775.