Journal of Chromatography A, 1135 (2006) 85–90

Analysis of 5-hydroxymethylfurfural in foods by

gas chromatography–mass spectrometry
E. Teixidó, F.J. Santos, L. Puignou, M.T. Galceran ∗
Departament de Quı́mica Analı́tica, Facultat de Quı́mica, Universitat de Barcelona, Martı́ i Franquès 1-11, 08028 Barcelona, Spain
Received 28 June 2006; received in revised form 5 September 2006; accepted 12 September 2006
Available online 28 September 2006

Abstract
A new, simple and selective method for the analysis of 5-hydroxymethylfurfural (HMF) in foods by gas chromatography coupled to mass
spectrometry (GC–MS) is proposed. Several derivatising procedures based on the formation of an HMF silylated derivative using different reagents
were studied. Among the derivatising reagents examined, N,O-bis-trimethylsilyltrifluoroacetamide (BSTFA) provided the best derivatisation yield.
Sample clean-up was also optimised, using either liquid–liquid extraction with dichloromethane or solid-phase extraction (SPE) with several
commercially available cartridges, and the best results were obtained using ENV+ cartridges. Quality parameters such as day-to-day and run-to-run
precision (RSD < 10%), linearity (between 25 and 700 ng g−1 ) and detection limit (6 ng g−1 ) were established. This method was successfully applied
to the analysis of HMF content in several Spanish food samples from a local market, such as jam, honey, orange juice and bakery products.
© 2006 Elsevier B.V. All rights reserved.

Keywords: Food analysis; 5-Hydroxymethylfurfural; GC–MS; Derivatisation; BSTFA; Solid phase extraction

1. Introduction reports have shown HMF to be an in vitro mutagen promoter

5-Hydroxymethylfurfural or 5-hydroxymethyl-2-furalde-
hyde (HMF) is a heat-induced common product of the
well-known Maillard reaction [1–3] that occurs in many
carbohydrate-rich foods such as biscuits, bread, marmalade,
breakfast cereals and honey. In addition, HMF is also pro-
duced during the acid-catalysed dehydratation of hexoses via
1,2-enolisation [4,5] followed by glucosamine hydrolysis [6,7]
and appears naturally in products where water coexists with
monosaccharides in acidic medium, such as balsamic vinegar
and fruit juice.
For years, the study of HMF in foodstuffs has received spe-
cial attention because it has been found to exhibit mutagenic
and DNA strand-breaking activity [8]. Moreover, the presence
of HMF in food has raised toxicological concern because this
compound and its derivatives, 5-chloromethylfurfural and 5-
sulfooxymethylfurfural, have been found to be citotoxic [9],
genotoxic [10], mutagenic and carcinogenic [11–14], induc-
ing colon-rectum, hepatic and skin cancers. Furthermore, recent
∗ Corresponding author. Tel.: +34 93 402 12 75; fax: +34 93 402 12 33.
E-mail address: mtgalceran@ub.edu (M.T. Galceran).
and initiator of colonic aberrant crypt foci (ACF), which is a
biomarker of genotoxicity and carcinogenicity in rat cell-lines
[15]. However, the toxicological relevance of its exposure has
not yet been clarified and the mechanisms by which HMF exerts
its genotoxicity remain unclear [16].
Although HMF is practically absent in fresh and untreated
food, its concentration tends to rise as a result of heating pro-
cesses, and its concentration is a useful tool to evaluate browning
reaction extension [17]. Moreover, it is a recognised parameter
related to the freshness and quality of some foods, and therefore
the analytical control of HMF has been used in food surveil-
lance to evaluate both the quality of the processing method and
the organoleptic characteristics of the final product. For instance,
the presence of HMF has been used in different foodstuffs, such
as jam and infant foods, as a good indicator of inadequate storage
time or temperature [18]. In fact, the Codex Alimentarius of the
World Health Organisation and the European Union (EU Direc-
tive 110/2001 [19]), have established a maximum HMF quality
level in honey (40 mg kg−1 ) and in apple juice (50 mg kg−1 ) as
deterioration and heat-treatment indicator.
Most of the methodologies currently applied for the analysis
of HMF in food are based on classical spectrophotometric tech-
niques [20,21] and liquid chromatography with UV detection

0021-9673/$ – see front matter © 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.chroma.2006.09.023
86 E. Teixidó et al. / J. Chromatogr. A 1135 (2006) 85–90
(LC–UV) [22–28], which is currently used as reference method
(AOAC method 980.23) [29]. Recently, a method using LC–MS
with APCI source for the analysis of HMF in baby foods has been
published [30]. GC–MS can also be a good option to ensure
the unequivocal identification and quantification of this com-
pound in food matrices. To our knowledge, only two references
are reported in the literature about the use of GC–MS for the
determination of HMF in foodstuffs [31,32]. However, limited
information about the performance of the method is available
and data of HMF levels in food are not provided.
In this paper, a new and reliable method for the analysis
of HMF in food using a derivatisation step and GC–MS is
proposed. For this purpose, several reagents were tested and
derivatising conditions such as time and temperature were opti-
mised in order to maximize the silylation of HMF. In addition, in
order to improve the clean-up and remove matrix interferences,
two methods based on liquid–liquid extraction and solid-phase
extraction were evaluated. Quality parameters were established
and the proposed method was applied to the determination of
HMF in several food samples.
2. Experimental
2.1. Reagents and materials
5-Hydroxymethylfurfural (HMF), 3-acetyl-2,5-dimethyl-
furane, 2-furaldehyde diethylacetal, furan-3-etinil-trimethyl-
silane, pyridine, trifluoroacetic acid and hydroxylamine
chlorhydrate (NH2 OH·HCl), all of them of high purity
(>99%) were purchased from Sigma–Aldrich (St. Louis,
MO, USA). The derivatisation reagents hexamethyldisilazane
(HMDS), N,O-bis(trimethylsilyl)trifluoroacetamide (BSTFA),
trimethylchlorosilane (TMCS), and pyridine were obtained from
Fluka (Bucks, Switzerland). Ethyl acetate and dichloromethane
of residues analysis grade, and methanol, n-hexane and isooc-
tane of HPLC grade, as well as sodium hydroxide, sodium
chloride, sodium sulphate, ammonia solution (25%), acetic acid
(99%), hydrochloric acid (25%) and sodium hydroxide (analyt-
ical reagent grade), were all purchased from Merck (Darmstad,
Germany). Water was purified by means of an Elix-Milli-Q sys-
tem (Millipore, Bedford, MA, USA).
Molecular sieves were purchased from Fluka (Buchs). Before
use, this material was activated at 200 ◦ C and added to the vials
that contained the reagents and solvents used in the analysis in
order to prevent the presence of moisture.
For solid-phase extraction studies, six commercially avail-
able cartridges were used: Oasis, HLB (200 mg, 6 ml) SPE car-
tridges were obtained from Waters (Milford, MA, USA), Strata-
X (200 mg, 6 ml) cartridges were from Phenomenex, Bond Elut
C18 (100 mg, 3 ml) were purchased from Varian (Harbor City,
USA), ENV+ cartridges (200 mg, 3 ml) were from IST (Hen-
goed, Mid-Glamorgan, UK), and both ENVI-18 (200 mg, 3 ml)
and Discovery DSC-18 (100 mg, 1 ml) from Supelco (Gland,
Switzerland). Coupling pieces and stopcocks were also pur-
chased from Varian.
A stock standard solution of HMF (500 ␮g g−1 ) was pre-
pared by weight in methanol. An intermediate standard solution
(7 ␮g g−1 ) was prepared weekly from the primary standard solu-
tion by appropriate dilution in methanol. Calibration standard
solutions were prepared in water by dilution of the secondary
solution in order to obtain HMF concentrations ranging from
25 to 700 ng g−1 . For the optimisation of the derivatisation pro-
cedure, several standard solutions of 100 ng g−1 were used and
prepared in the same way as the calibration solutions. Furan-
3-etinil-trimethylsilane standard solution (340 ng g−1 ) was pre-
pared in hexane and used as internal standard for GC–MS quan-
tification.
All food samples (orange juice, multifloral honey, breakfast
cereals, plum jam, biscuits and oranges) were purchased at a
local market in Barcelona (Spain).
2.2. GC–MS conditions
The determination of HMF by GC–MS was carried out using
a Trace GC 2000 gas chromatograph coupled to a GCQ/Polaris
ion-trap mass spectrometer (ThermoElectron, Austin, TX,
USA) equipped with an AS2000 autosampler (ThermoElec-
tron, Milan, Italy). The chromatographic separation was per-
formed using a DB-5MS (5% phenyl, 95% methylpolysiloxane),
30 m × 0.25 mm i.d., fused-silica capillary column (J&W Scien-
tific, Folsom, CA, USA) of 0.25 ␮m film thickness. The carrier
gas was helium at a constant flow-rate of 1.3 ml/min held by elec-
tronic pressure control. Injector temperature was 280 ◦ C, and the
splitless injection mode (1 min) was used. The oven temperature
programme was 70 ◦ C (held for 1 min) to 110 ◦ C at 7 ◦ C/min
and to 300 ◦ C at 20 ◦ C/min (held for 10 min). The MS operat-
ing conditions were the following: positive electron ionisation
mode (EI+) using automatic gain control (AGC) with an elec-
tron energy of 70 eV and an emission current of 250 ␮A. The ion
source and transfer line temperatures were 200 ◦ C and 280 ◦ C,
respectively. The instrument was tuned using perfluorotributy-
lamine (FC-43) according to manufacturer’s recommendations
in order to achieve the best sensitivity. Electron multiplier volt-
age was set to 1500 V (105 gain) by automatic tuning. EI full-
scan data acquisition was registered over the range m/z 40–200
at 0.66 s per scan (7 ␮scans per scan). Xcalibur version 1.2 soft-
ware was used for control, general operation and data acquisition
of the results.
2.3. Analytical method
Prior to the extraction stage, food samples were ground and
homogenized using a supermixer blender system (Moulinex,
Lyon, France) and an Ultraturrax T25 basic (Ika, Staufen, Ger-
many). Subsamples of 0.1 g were weighted into 40 ml screw-
capped vials and 5–20 ml of acidic water, previously adjusted at
pH 1 with HCl, were added. Then each vial was stirred for 1 min
in a vortex mixer Stuart (Barloworld Scientific Ltd., Stafford-
shire, UK).
A Supelco Visiprep and a Visidry SPE vacuum manifold
(Supelco, Gland, Switzerland) were used for manipulations
of the SPE cartridges and solvent evaporation, respectively.
Initially, the SPE cartridges were conditioned using 5 ml of
methanol and 1 ml of acetic acid aqueous solution 0.1% (v/v).
 E. Teixidó et al. / J. Chromatogr. A 1135 (2006) 85–90
Then the cartridge was loaded with 1 ml aliquot of the aqueous
solution followed by 1 ml of water, in order to wash the car-
tridge and eliminate interferences. The quantitative elution of
the compound was achieved by using 2 ml of methanol and the
resulting extract was evaporated to dryness under a stream of
nitrogen. Then the derivatisation step was performed at room
temperature, adding 1 ml of BSTFA to the dry extract and stir-
ring for 1 min. Finally, an appropriate amount of the internal
standard solution (furan-3-etinil-trimethylsilane) was added to
the final extract before the GC–MS analysis in order to obtain a
concentration of 260 ng g−1 .
2.4. Quantification
Quantification of hydroxymethylfurfural was performed by
internal standard, using a linear working range from 25 to
700 ng g−1 . For this purpose, 10 ml of Milli-Q water spiked with
appropriate amounts of a HMF methanolic standard solution
were prepared following the SPE clean-up and derivatisation
procedures previously described in Section 2.3.
Several compounds were studied in order to select an ade-
quate internal standard for the quantification of HMF. Taking
into account that HMF isotopically labelled standard is not
commercially available, an internal standard with similar chro-
matographic behaviour and chemical properties and not cur-
rently present in food was required. For this purpose, three
furanic compounds such as furan-3-etinil-trimethylsilane, 3-
acetyl-2,5-dimethylfurane and 2-furaldehyde diethylacetal were
tested. Among them, furan-3-etinil-trimethylsilane was selected
because it provided a higher response in GC–MS and a bet-
ter linearity of the calibration curve. An additional advan-
tage of this silylated compound is that it is not present in
food samples and, moreover, it does not consume derivatising
reagent.
3. Results and discussion
3.1. Optimisation of the derivatisation procedure
Preliminary studies were carried out in order to obtain the
trylmethylsilyl ester of HMF. In a first attempt, hexamethyldis-
ilazane (HMDS) in pyridine with trifluoroacetic acetic acid as
suggested by Horvath and Molnar-Perl [31,32] was used. Nev-
ertheless, the derivatising yield was very low and a precipitate
was formed in the reaction vial. Thus, a mixture of hexamethyl-
disilazane and trimethylchlorosilane (TMCS), which are also
proposed for the derivatisation of HMF [33,34], was used. At
these conditions low response was obtained so some additional
experiments were carried out in order to improve the silylating
reaction. Mixtures of HMDS and TMCS at different proportions
(HMDS:TMCS ratios ranging from 1:3 to 3:1) and different tem-
peratures (25 ◦ C, 60 ◦ C and 100 ◦ C) were tested. In all cases, a
reaction time of 30 min was fixed, and a total volume of 200 ␮l of
silylating reagent was added to HMF standard solutions (150 ng)
in MeOH, previously evaporated to dryness under a stream of
nitrogen. At these conditions, the best results were obtained with
the 1:3 HMDS:TMCS ratio. In addition, heating did not enhance
87
Fig. 1. Influence of the derivatising procedure using: (a) HMDS + TMCS (1:3)
and (b) BSTFA on the HMF response at different temperatures (room tempera-
ture, 60 ◦ C and 100 ◦ C).
the derivatisation yield (Fig. 1a), probably because at high tem-
perature a degradation of the reagent or partial evaporation of the
derivative occurs, so room temperature was selected for subse-
quent studies. Finally, the reaction time was also optimised, from
15 to 90 min, and maximum derivatisation yield was obtained
after 60 min.
Although the use of the proposed HMDS/TMCS mixture pro-
vided a relatively high response, other silylating reagents such
as BSTFA, which provides stronger silylating capability and
avoids the use of toxic solvents such as pyridine, were studied.
Thus, derivatisation conditions for BSTFA derivatisation, such
as amount of reagent, temperature and reaction time were opti-
mised. Different amounts of the reagent, ranging from 250 to
1000 ␮l, were added to HMF standard solutions of 150 ng. The
reaction temperature (25 ◦ C, 60 ◦ C and 100 ◦ C) and the reac-
tion time from 15 to 90 min were tested. It was also observed
that heating did not enhance the silylating reaction, so the best
conditions were found to be 500 ␮l of BSTFA at 25 ◦ C (room
temperature) for 15 min (Fig. 1b).
Finally, the response obtained with both silylating proce-
dures was compared and it was found that the derivatising
yield using BSTFA was four-fold higher than that obtained with
HMDS/TMCS (Fig. 1). Moreover, BSTFA made it possible to
reduce the reaction time significantly, so this was selected for
subsequent studies.
3.2. Optimisation of the extraction and clean-up procedure
After optimisation of the derivatising conditions, several
studies were carried out in order to develop a suitable extraction
and clean-up method to obtain clean food sample extracts prior
to the derivatisation step. Liquid–liquid extraction was tested,
using ethyl acetate, dichloromethane and a sequential extraction
with both solvents (ethyl acetate followed by dichloromethane).
In addition, the effect of pH in the extraction yield was studied,
varying it from 4 (pH of honey in water) to 7 by adding NaOH
2 M to the sample. In all cases, three consecutive liquid–liquid
extractions with portions of 15 ml of solvent were performed.
Using this technique, the best extraction yield was obtained
using dichloromethane at pH 7.
88 E. Teixidó et al. / J. Chromatogr. A 1135 (2006) 85–90
In order to quantify these samples, standards for the cal-
ibration curve were prepared by spiking water with suitable
amounts of HMF. Surprisingly, after liquid–liquid extraction,
HMF was only detected in those calibration solutions with very
high concentrations, which means that the analyte was poorly
extracted. In contrast, honey sample extracts with similar HMF
content gave higher responses than the spiked water standards.
This behaviour may be due to the fact that the honey sam-
ples have higher ionic strength, which enhances the extraction
of the analyte. In order to overcome this problem, a saturated
NaCl solution was added to the spiked water standards, but no
improvement in the results was observed. Therefore, standard
addition is required for quantitation. Triplicate analysis of two
honey samples were carried out by spiking the samples at dif-
ferent concentration levels, 0%, 50%, 100%, 150% and 200%
of the estimated concentration of HMF in samples. The results
obtained for two honey samples using this method were 4.1
and 36.7 mg kg−1 . The relative standard deviation (RSD) of the
results obtained was relatively high (RSD 15–20%), although a
good limit of detection (12 ␮g kg−1 ) was obtained. This value
was estimated using a honey sample with a very low content of
HMF.
In order to reduce the volume of solvent required for the anal-
ysis, to improve the precision and to decrease extraction time,
a clean-up method based on solid phase extraction was opti-
mised. SPE commercial cartridges, such as three C-18 cartridges
(ENVI-18, DSC-18 and Bond Elut C18) and three polymeric car-
tridges (Oasis HLB, Strata-X and ENV+) were tested, and the
results were compared. All the cartridges were preconditioned
with methanol and acetic acid aqueous solution (0.1% v/v) and
loaded with 1 ml of HMF aqueous solution following the ana-
lytical method described in Section 2.3. The selection of acidic
solution instead of water to activate the cartridge was due to the
fact that acidic compounds such as HMF are not quantitatively
retained when using only water. Thus, all the cartridges were
eluted with 2 ml of methanol, which was found to be enough
to remove the HMF adsorbed in the cartridge, and evaporated
to dryness. Finally, the extract was derivatised using BSTFA,
following the optimised procedure (see Section 3.1) before the
GC–MS analysis. As can be seen in Fig. 2, Oasis HLB and ENV+
cartridges provided the best recoveries of HMF but the last one
Fig. 2. Extraction efficiency of the HMF derivative using different SPE car-
tridges.
gave a slightly higher response, with the additional advantage of
giving a linear response at low concentration levels. Therefore,
the use of SPE ENV+ cartridges is proposed for the extraction
and clean-up of the samples prior to the derivatisation of HMF.
3.3. Quality parameters
The EI spectrum of the HMF derivative showed a very small
molecular ion (m/z 198), as happens for TMS derivatives of
cyclic alcohols [35]. Therefore, m/z 183 corresponding to the
ion [M-CH3 ]+ was selected for quantification whereas for con-
firmation ions m/z 169 corresponding to [M-COH]+ and m/z 109
[M-OTMS]+ , coming from the loss of 89 units characteristic of
silylated compounds, were used. For the internal standard, two
ions, m/z 164 (molecular ion) and m/z 149, corresponding to
[M-CH3 ]+ , were used for quantification in order to prevent any
interference from the matrix.
Quality parameters such as run-to-run and day-to-day preci-
sion, limits of detection (LODs) and linearity were established
for the proposed method. Linearity ranged between 25 and
700 ng g−1 (r2 = 0.999). LODs and LOQs based on a signal-to-
noise ratio (S/N) of 3:1 and 10:1, respectively, were determined
using both a handmade fresh orange juice and standard solutions.
Fresh orange juice was used as blank sample because that was
the only sample analysed without detectable quantities of HMF,
and the LOD was determined by spiking this sample at very
low concentration levels. The LOD and LOQ for food samples
were 12 ng g−1 (corresponding to 8 pg injected) and 39 ng g−1
(26 pg injected), respectively. These values were similar to those
obtained for standard solutions (LOD: 6 pg injected, LOQ: 20 pg
injected) and it can be explained because SPE makes it possible
to reduce the matrix interferences so standard solutions and sam-
ple behaviour is similar. For run-to-run and day-to-day precision
three replicates of standard solutions at two different levels were
analysed on 1 day and on 3 different days, respectively. Good
precision was achieved with a relative standard deviation (RSD)
of 2.5% for run-to-run and 5% for day-to-day.
3.4. Analysis of HMF in food
To examine the applicability of the proposed SPE-
derivatisation method, food samples of different natures were
studied: solid (biscuits and breakfast cereals), liquid (orange
juice) and semi-solid (honey and jam). All samples were
analysed in triplicate to determine HMF by the optimised
method. The samples were prepared following the methodology
described in the experimental section (see Section 2.3), dissolv-
ing them in a suitable amount of acidic water, depending on both
the type of foodstuff analysed and the HMF levels expected.
As an example, GC–MS chromatograms and mass spectra of
a breakfast cereal sample are given in Fig. 3. No interferences
from other compounds that may have been present in the sam-
ple matrix were detected at these conditions since the abundance
ratio between the quantitation and confirmation ions remain con-
stant (±10%).
The results obtained for the food samples analysed using
the developed method are summarised in Table 1. HMF was
 E. Teixidó et al. / J. Chromatogr. A 1135 (2006) 85–90 89

Fig. 3. GC–MS chromatograms of a breakfast cereal sample and the corresponding spectra of: (a) the HMF derivative (TMS-HMF) and (b) the internal standard
(furan-3-etinil-trimethylsilane) obtained by the proposed SPE-derivatisation method.

Table 1
Amounts of HMF in the foodstuffs analysed
Food Sample name Concentration (␮g/g) Published data (␮g/g) [ref.]

Average (n = 3) RSD %

Honey Multifloral A 38.3 3

0.8–138 [23,36,37]
Multifloral B 4.6 10
Breakfast cereals Honey rings 24.7 3
4–193 [38]
Corn Flakes 46.8 4
Orange juice Juice A 8.5 4
4–22 [39]
Juice B 6.3 7
Biscuits Honey biscuits 7.0 5 –
Jam Plum 12.7 2 –

found at concentration levels between 5 and 47 ␮g g−1 , the 4. Conclusions

precision achieved being lower than 10% RSD. These concen-
trations are in agreement with those reported in the literature.
For instance, for honeys, we found levels of 5 and 38 ␮g g−1 ,
similar to those reported by Costa et al. [36], whose values are
between 1.7 and 39 ␮g g−1 and lower than those found by Spano
et al. [37], which range from 5.1 to 138 ␮g g−1 . Moreover, for
breakfast cereal samples, HMF concentration ranged from 25 to
47 ␮g g−1 , while in the literature HMF levels found are between
4 and 193 ␮g g−1 , depending on the type of cereal. Rice-based
cereals have the lowest HMF levels (from 3.7 to 16.1 ␮g g−1 )
while higher levels correspond to corn (from 14.9 to 53.0 ␮g
g−1 ) and wheat-based cereals (from 61.0 to 193.3 ␮g g−1 )
[38].
The applicability of an SPE-derivatisation method in con-
junction with GC–MS determination for the analysis of HMF in
food samples is demonstrated. This procedure is simple, rapid,
and shows good repeatability, linearity and sensitivity. Sam-
ple preparation with an efficient clean-up followed by an easy
derivatisation and a fast chromatographic determination allows
completion of the whole analysis in about 30 min. For sample
clean-up the best results were obtained using ENV+ SPE poly-
meric cartridges. BSTFA was found to be the most effective
reagent to carry out the derivatisation of HMF, making the per-
formance of this reaction easier, since neither solvent nor catal-
yser were required. Furthermore, this method provides a low
90 E. Teixidó et al. / J. Chromatogr. A 1135 (2006) 85–90
limit of detection, about 100-fold lower than the LC–UV meth-
ods reported in the literature [36–39]. The proposed GC–MS
method was successfully applied to the determination of 5-
hydroxymethylfurfural in several food samples containing a
wide range of HMF levels and can be proposed as an alternative
to the LC–UV Official Method (AOAC 90.183) [29] currently
applied for the determination of hydroxymethylfurfural in foods.
Acknowledgements
This work was carried out with support from the European
Commission, Priority 5 on Food Quality and Safety (Contract
no. FOOD-CT-2003-506820 Specific Targeted Project), “Heat-
generated food toxicants-identification, characterisation and risk
minimisation”. This publication reflects the author’s views and
not necessarily those of the EC. The information in this docu-
ment is provided as is and no guarantee or warranty is given that
the information is fit for any particular purpose. The user thereof
uses the information at its sole risk and liability. E.T. acknowl-
edges the University of Barcelona for a Ph.D. grant (Beca de
Formació en la Recerca i la Docència).
References
[1] M. Friedman, J. Agric. Food Chem. 44 (1996) 631.
[2] A.J. Tomlinson, J.P. Landers, I.A.S. Lewis, S. Naylor, J. Chromatogr. A
652 (1993) 171.
[3] H.E. Berg, M.A.J.S. Boekel, Neth. Milk Dairy J. 48 (1994) 157.
[4] A. Gottschalk, Biochem. J. 52 (1952) 455.
[5] L.W. Kroh, Food Chem. 51 (1994) 373.
[6] L.F. Leloir, C.E. Cardini, Biochim. Biophys. Acta 12 (1953) 15.
[7] M. Jun, Y. Shao, C.T. Ho, U. Koetter, S. Lech, J. Agric. Food Chem. 51
(2003) 6340.
[8] H. Omura, N. Jahan, K. Shinohara, Y. Murakami, The Maillard Reaction in
Foods and Nutrition, American Chemical Society, Washington, DC, 1983,
p. 537.
[9] L. Naessberger, Hum. Exp. Toxicol. 9 (1990) 211.
[10] W.R. Bruce, M.C. Archer, D.E. Corpet, A. Medline, S. Minkin, D. Stamp,
Y. Yin, X.M. Zhang, Mutat. Res. 290 (1993) 111.
[11] Y.C. Lee, M. Shlyankevich, H.K. Jeong, J.S. Douglas, Y.J. Surh, Biochem.
Biophys. Res. Commun. 209 (1995) 996.
[12] Y.J. Surh, A. Liem, J.A. Miller, S.R. Tannenbaum, Carcinogenesis 15
(1994) 2375.
[13] Y.J. Surh, S.R. Tannenbaum, Chem. Res. Toxicol. 7 (1994) 313.
[14] X.M. Zhang, C.C. Chan, D. Stamp, S. Minkin, M.C. Archer, W.R. Bruce,
Carcinogenesis 14 (1993) 773.
[15] H. Glatt, H. Schneider, Y. Liu, Mutat. Res. 580 (2005) 41.
[16] C. Janzowski, V. Glaab, E. Samimi, J. Schlatter, G. Eisenbrand, Food Chem.
Toxicol. 38 (2000) 801.
[17] B. Fallico, E. Arena, M. Zappala, Food Chem. 81 (2003) 569.
[18] M. Rada-Mendoza, M.L. Sanz, A. Olano, M. Villamiel, Food Chem. 85
(2004) 605.
[19] Directive 2001/110/EC of 20 December 2001. Off. J. Eur. Commun.
[20] O. Winkler, Z. Lebensm, Unters. Forsch. 102 (1955) 161.
[21] J.W.J. White, J. AOAC Int. 62 (1979) 509.
[22] M. Castellari, E. Sartini, U. Spinabelli, C. Riponi, S. Galassi, J. Chromatogr.
Sci. 39 (2001) 235.
[23] M.J. Nozal, J.L. Bernal, L. Toribio, J.J. Jimenez, M.T. Martin, J. Chro-
matogr. A 917 (2001) 95.
[24] F. Lo Coco, C. Valentini, V. Novelli, L. Ceccon, Anal. Chim. Acta 306
(1995) 57.
[25] A. Ramirez-Jimenez, E. Guerra-Hernandez, B. Garcia-Villanova, Food
Chem. 83 (2003) 219.
[26] E. Ferrer, A. Alegria, R. Farre, P. Abellan, F. Romero, J. Chromatogr. A
947 (2002) 85.
[27] J.A. Rufian-Henares, C. Delgado-Andrade, F.J. Morales, J. AOAC Int. 89
(2006) 161.
[28] M. Zappala, B. Fallico, E. Arena, A. Verzera, Food Contr. 16 (2005) 273.
[29] Official Methods of Analysis, vol. 44 Association of Official Analytical
Chemists (AOAC) International, MD, USA, 1996. p. 26.
[30] V. Gokmen, H.Z. Senyuva, J. Agric. Food Chem. 5554 (2006) 2845.
[31] K. Horvath, I. Molnar-Perl, Chromatographia 48 (1998) 120.
[32] I. Molnar-Perl, A. Vasanits, K. Horvath, Chromatographia 48 (1998) 111.
[33] D. Fabbri, G. Chiavari, Anal. Chim. Acta 449 (2001) 271.
[34] F. Bartolozzi, G. Bertazza, D. Bassi, G. Cristoferi, J. Chromatogr. A 758
(1997) 99.
[35] F.G. Kitson, B.S. Larsen, C.N. McEwen, Gas Chromatography and Mass
Spectrometry—A Practical Guide, Academic Press, San Diego, CA, USA,
1998.
[36] L.S.M. Costa, M.L.S. Albuquerque, L.C. Trugo, L.M.C. Quinteiro, O.M.
Barth, M. Ribeiro, C.A.B. De Maria, Food Chem. 65 (1999) 347.
[37] N. Spano, L. Casula, A. Panzanelli, M.I. Pilo, P.C. Piu, R. Scanu, A. Tap-
paro, G. Sanna, Talanta 68 (2006) 1390.
[38] B. Garcia-Villanova, E. Guerra-Hernandez, E. Martinez-Gomez, J. Mon-
tilla, J. Agric. Food Chem. 41 (1993) 1254.
[39] J.P. Yuan, F. Chen, J. Agric. Food Chem. 46 (1998) 1286.