Isolation and determination of α-dicarbonyl compounds by RP-HPLC-DAD in green and roasted coffee

Article abstract of DOI:10.1021/jf071917l

Glyoxal, methylglyoxal, and diacetyl formed as Maillard reaction products in heat-treated food were determined in coffee extracts (coffee brews) obtained from green beans and beans with different degrees of roast. The compounds have been reported to be mutagenic in vitro and genotoxic in experimental animals in a number of papers. More recently, α-dicarbonyl compounds have been implicated in the glycation process. Our data show that small amounts of glyoxal and methylglyoxal occur naturally in green coffee beans. Their concentrations increase in the early phases of the roasting process and then decline. Conversely, diacetyl is not found in green beans and forms later in the roasting process. Therefore, light and medium roasted coffees had the highest glyoxal and methylglyoxal content, whereas dark roasted coffee contained smaller amounts of glyoxal, methylglyoxal, and diacetyl. For the determination of coffee α-dicarbonyl compounds, a reversed-phase high performance liquid chromatography with a diode array detector (RP-HPLC-DAD) method was devised that involved the elimination of interfering compounds, such as chlorogenic acids, by solid phase extraction (SPE) and their derivatization with 1,2-diaminobenzene to give quinoxaline derivatives. Checks of SPE and derivatization conditions to verify recovery and yield, respectively, resulted in rates of 100%. The results of the validation procedure showed that the proposed method is selective, precise, accurate, and sensitive.

Full text of DOI:10.1021/jf071917l

J. Agric. Food Chem. 2007, 55, 8877–8882 8877
Isolation and Determination of r-Dicarbonyl
Compounds by RP-HPLC-DAD in Green and Roasted
Coffee
MARIA DAGLIA, ADELE PAPETTI, CAMILLA ACETI, BARBARA SORDELLI,
VALENTINA SPINI, AND GABRIELLA GAZZANI*
Department of Pharmaceutical Chemistry, Faculty of Pharmacy, University of Pavia, Via Taramelli 12,
2
7100 Pavia, Italy
Glyoxal, methylglyoxal, and diacetyl formed as Maillard reaction products in heat-treated food were
determined in coffee extracts (coffee brews) obtained from green beans and beans with different
degrees of roast. The compounds have been reported to be mutagenic in vitro and genotoxic in
experimental animals in a number of papers. More recently, R-dicarbonyl compounds have been
implicated in the glycation process. Our data show that small amounts of glyoxal and methylglyoxal
occur naturally in green coffee beans. Their concentrations increase in the early phases of the roasting
process and then decline. Conversely, diacetyl is not found in green beans and forms later in the
roasting process. Therefore, light and medium roasted coffees had the highest glyoxal and
methylglyoxal content, whereas dark roasted coffee contained smaller amounts of glyoxal, methyl-
glyoxal, and diacetyl. For the determination of coffee R-dicarbonyl compounds, a reversed-phase
high performance liquid chromatography with a diode array detector (RP-HPLC-DAD) method was
devised that involved the elimination of interfering compounds, such as chlorogenic acids, by solid
phase extraction (SPE) and their derivatization with 1,2-diaminobenzene to give quinoxaline
derivatives. Checks of SPE and derivatization conditions to verify recovery and yield, respectively,
resulted in rates of 100%. The results of the validation procedure showed that the proposed method
is selective, precise, accurate, and sensitive.
KEYWORDS: Coffee; glyoxal; methylglyoxal; diacetyl; HPLC-DAD; roasting process
INTRODUCTION
the methyl R-dicarbonyl intermediate, which undergoes trans-
formation to aldehydes, ketones, furans, pyrroles, quinolines,
and indoles. In foods heated at temperatures much higher than
Dicarbonyl compounds are reactive intermediates formed
during food heat treatment, such as roasting, baking, broiling,
and frying. Caramelization and the Maillard reaction are the
two main types of reactions that occur in heated foods (1). At
the chemical level, caramelization involves ring-opening, eno-
lization of reducing sugars, and sugar dehydration, which
consists of a sequence of water ꢀ-eliminations from sugar
molecules and formation of R-dicarbonyl compounds. This
sequence of reactions is normally a slow process. It becomes
significant at high sugar concentrations, in the presence of a
catalytic base, and at high temperature. The Maillard reaction
consists of three main stages. In the first, an amino acid
condenses with a reducing sugar to form the Schiff base, which
undergoes cyclization to N-substituted 1-amino-1-deoxy-2-
ketose by Amadori rearrangement. Recent studies suggest that
R-dicarbonyl compounds are formed from Schiff base. A major
pathway of the Maillard reaction leads from the 1,2-eneaminol
form of the Amadori compound to a deoxyhexosone intermedi-
ate (2). A minor pathway involves conversion of 2,3-enediol to
1
00 °C in the absence of free water, the process of thermal
degradation, called pyrolysis, involves sugars, amino acids, and
other organic constituents.
The most extensively studied R-dicarbonyl compounds are
methylglyoxal, found in foodstuffs such as bread, soy sauce,
instant tea, and roasted coffee, and glyoxal, which has attracted
considerable attention as a lipid peroxidation product. Glyoxal
can also derive from sugar fragmentation and has been isolated
in many foodstuffs such as roasted coffee and cocoa (3).
The toxicological profile of R-dicarbonyl compounds, par-
ticularly glyoxal and methylglyoxal, is not completely clear. In
the 1980s, they came under scrutiny for their possible carcino-
genic potential due to their in vitro mutagenic activity (4) and
genotoxicity in experimental animals (5–7). In recent years, the
interest has focused on the implications of the glycation process.
The formation of R-dicarbonyl compounds is a key step in the
Maillard reaction because these intermediates are up to 20000
times more reactive than glucose in glycation reactions (8).
Advanced glycation end products (AGEs) formed by nonenzy-
*
Corresponding author. Tel. +39 0382 987 373. Fax +39 0382
4
22 975. E-mail: gabriella.gazzani@unipv.it.
1
0.1021/jf071917l CCC: $37.00  2007 American Chemical Society
Published on Web 10/10/2007

8878 J. Agric. Food Chem., Vol. 55, No. 22, 2007
Daglia et al.
matic glycations and glycoxidations are implicated in the
pathophysiological changes associated with a number of dis-
eases, such as diabetes, atherosclerosis, and neurodegenerative
disorders (9–12). AGEs form in physiological conditions in the
presence of R-dicarbonyl compounds derived from a Schiff base
without the delay involved in Amadori rearrangements. Meth-
ylglyoxal, as an AGE precursor, has been shown to be highly
active in glycation even at the low concentrations occurring in
physiological systems. The role of dietary glycation products
is debated. As considerable amounts of Maillard reaction
products are formed in foods due to heat processing, recent
papers discuss food-derived AGEs as possible glycotoxins
two SPE fractions were concentrated to dryness by a Rotavapor R-210
rotary evaporator (BUCHI Italia S.r.l., Milan, Italy), and the residues
¨
were dissolved in 2 mL of Millipore-grade water. All experiments were
performed in triplicate.
Quinoxaline Derivative Preparation. The method described by de
Revel et al., with some modifications (18), was used for the preparation
of derivatized quinoxaline. 1,2-Diaminobenzene (1 mg) was added to
2
mL of R-dicarbonyl standard aqueous solutions of glyoxal (75.2 µM),
methylglyoxal (75.2 µM) or diacetyl (79.3 µM) separately. The
derivatization reaction was carried out also on coffee extract (CR14)
as well as the SPE-F1 fraction obtained either from the R-dicarbonyl
standard mixture (each compound at the same concentrations reported
above) or from coffee extracts. The pH was adjusted to 8.0 with NaOH
(
0.5 M). The reaction mixture was kept at 60 °C for 3 h; after cooling,
(
13–15).
the solutions were directly injected into the HPLC system.
HPLC Analysis. All experiments were performed using an Agilent
Various methods have been developed to determine R-dicar-
bonyl compounds. Methylglyoxal has been determined by
spectrometry after chemical derivatization with 2,4-dinitrophe-
nylhydrazine or 4-amino-5-hydrazino-3-mercapto-1,2,4-triazole
1
100 HPLC system (Agilent, Waldbronn, Germany) equipped with a
gradient quaternary pump, a thermostatted column compartment, and
a DAD. The Agilent Chemstation software was used for control of the
HPLC system and data processing. Quinoxaline derivatives were
separated by a C18 Hypersilcolumn 250 × 4.6 mm, 5 µm (CPS
Analitica, Milan, Italy) with matching Lichrospher 100 RP-18, 5 mm
guard column (Merck, Darmstadt, Germany).
(
16). R-Dicarbonyl compounds have also been analyzed by gas
chromatography (GC) after derivatization with cysteamine to
give 2-acetyl-thiazolidine derivatives (17). The best current
assays involve derivatization of R-dicarbonyl substances with
Chromatographic conditions for gradient elution were as follows:
1
,2-diaminobenzene followed by quantification of the resulting
quinoxaline by GC or high performance liquid chromatography
HPLC). Methods for the determination of R-dicarbonyl com-
pounds have been applied to different matrices, such as wine
18) and cultured animal cells (19). However, only GC has been
-1
flow rate, 0.6 mL min ; volume injected, 20 µL; column temperature,
0 °C; UV spectra were recorded in the 190–600 nm range, and
2
(
chromatograms were acquired at 314 nm. Separations were performed
using a gradient of increasing methanol concentrations in water acidified
(
(
pH 3.00 ( 0.01) with 0.5% acetic acid (v/v) as follows: 40 min linear
gradient from 10 to 50% methanol, 10 min linear gradient from 50 to
5% methanol, 4 min increasing gradient segment to 100% methanol
applied to coffee, probably due to the complexity of its
matrix (20–22).
7
The aims of this study were (1) to devise a method of sample
preparation allowing simultaneous and accurate determination
of glyoxal, methylglyoxal, and diacetyl in coffee brew by
reversed-phase high performance liquid chromatography with
a diode array detector (RP-HPCL-DAD) (2) to investigate their
formation kinetics during the roasting process.
followed by a 4 min isocratic with 100% methanol. The mobile phase
composition was taken to the initial condition in 4 min, and the column
was equilibrated 10 min before the next injection.
Identification and Quantification of Quinoxaline Derivatives.
Retention times and UV spectra were used to identify the quinoxaline
derivative peaks. Stock standard solutions of quinoxaline (Q), 2-methyl-
quinoxaline (2MQ), and 2,3-dimethylquinoxaline (2,3DMQ) were
prepared by dissolving carefully weighed amounts of each standard
compound in 50% (v/v) methanol–Millipore-grade water. Solutions
were analyzed by RP-HPLC-DAD. Each standard solution was diluted
with mobile phase to six final concentrations ranging from 7.5 to 150
µM for Q and 2MQ, and 5–100 µM for 2,3DMQ. Each concentration
was analyzed in triplicate. Quantification of individual compounds was
performed by the external standard method using a six-point regression
curve. The peak area was plotted against concentration, and least-
squares regression analysis was used to fit lines to the data. As the
molar ratio between R-dicarbonyl compounds and the respective
quinoxaline derivatives is 1:1, the amount of each R-dicarbonyl
MATERIALS AND METHODS
Chemicals and Standards. HPLC-grade solvents (methanol and
acetic acid), glyoxal (MW 58.04 Da), methylglyoxal (MW 72.06 Da),
diacetyl (MW 89.09 Da), quinoxaline (MW 130.15 Da), 2-methylqui-
noxaline (MW 144.18 Da), 2,3-dimethylquinoxaline (MW 158.20 Da),
5
-methylquinoxaline, and 1,2-diaminobenzene were purchased from
Sigma-Aldrich (St. Louis, MO).
Coffee Sample and Coffee Extract Preparation. A 200 g aliquot
of Coffea robusta (CR) beans from Java was roasted up to 210 °C in
a pilot roaster apparatus (STA Impianti S.r.l., Bologna, Italy) that
allowed the collection of small aliquots of coffee beans every 2 min.
During roasting, 20 g aliquots of coffee beans were collected every 2
min for 20 min to obtain 10 batches with 10 different degrees of roast
-
1
compound was expressed as µmol L
(µM). In Figure 2, the
concentration of R-dicarbonyl compounds is expressed as mg/100 g of
coffee powder.
(
CR2-CR20). On the basis of the weight lost during the roasting process,
due to vapor formation and cell fragment loss, CR samples roasted for
0 min (CR10, weight loss 11%), 14 min (CR14, weight loss 14%),
RESULTS AND DISCUSSION
1
Isolation and Derivatization of Coffee r-Dicarbonyl
Compounds. Derivatization of R-dicarbonyl compounds oc-
curring in coffee solutions was performed using 1,2-diamino-
benzene (pH 8.0) for 3 h at 60 °C, as described by de Revel et
al. (18) for the determination of R-dicarbonyl compounds in
red and white wines. To verify the yield of the reaction, aqueous
solutions of standard glyoxal, methylglyoxal, and diacetyl were
derivatized as described and analyzed by RP-HPLC-DAD after
addition of 5-methylquinoxaline (5MQ), used as internal
standard. Standard Q, 2MQ, 2,3DMQ, and 5MQ solutions at
the same concentrations were then analyzed. The ratios of the
peak areas of standard quinoxaline derivatives (Q, 2MQ, and
and 20 min (CR20, weight loss 20%) are generally considered as light,
medium, and dark roasted coffee, respectively. All batches were ground
in a laboratory scale mill and passed through a no. 30 sieve. Green
and roasted coffee extracts were prepared by boiling 10 g aliquots of
coffee powder in 100 mL of Millipore grade water for 10 min; each
100 mL solution was filtered through a cellulose acetate/cellulose nitrate
mixed esters membrane (0.45 µm; Millipore Corporation, Billerica,
MA).
Solid-Phase Extraction (SPE). A C18 Sep-Pak cartridge (Waters,
Milford, MA) was conditioned with methanol (10 mL) and distilled
water (2 × 10 mL). Aliquots (2 mL) of glyoxal, methylglyoxal, and
diacetyl water solution mixture (30 µM) or of coffee extracts obtained
from beans with different degrees of roast were passed through the
-
1
2
,3DMQ) or R-dicarbonyl compounds derivatized with 1,2-
cartridge at a flow rate of e2 mL min . Polar substances were eluted
first, with 2 mL of Millipore-grade water (SPE-F1). Less polar
substances were then eluted of with 4 mL of methanol (SPE-F2). The
diaminobenzene to the 5MQ peak area showed that under our
experimental conditions, the yield of the reaction was 99.8–

Coffee R-Dicarbonyl Compounds Determination
J. Agric. Food Chem., Vol. 55, No. 22, 2007 8879
Figure 1. (a) RP-HPLC-DAD chromatograms: standard quinoxaline derivatives detected at 314 nm; (b) SPE-F1 obtained from CR14 roasted coffee
before derivatization with 1,2-diaminobenzene; (c) SPE-F1 obtained from CR14 roasted coffee after derivatization with 1,2-diaminobenzene.
a
Table 1. Yields of Reaction of R-Dicarbonyl Compounds Derivatized with 1,2-Diaminobenzene
standard
derivatized
c
d
c
e
compound
quinoxaline
concentration [µM]
peak area
ratio
peak area
ratio
75.2
75.2
79.3
75.2
928.861
941.621
1044.12
947.219
0.981
0.994
1.102
926.735
942.421
1043.281
946.898
0.979
0.995
1.102
2
2
5
-methylquinoxaline
,3-dimethylquinoxaline
-methylquinoxaline
b
a
b
c
d
e
All experiments were performed in triplicate. Internal standard. Expressed as arbitrary units. Ratio of standard quinoxaline derivatives to 5MQ peak area. Ratio
of derivatized R-dicarbonyl compounds to 5MQ peak area.
1
00.1% for the considered R-dicarbonyl compounds, in line with
injection of the reaction mixture using gradient elution. Various
compositions of the acetic acid/water–methanol eluting mixture
were tested to resolve the derivatized R-dicarbonyl compounds
from the other coffee components. Satisfactory resolution was
achieved for derivatized 2MQ and 2,3DMQ. These quinoxaline
derivatives were identified by comparing their retention times
the results reported by de Revel et al. (18) (Table 1).
For the initial phase of the investigation, a coffee solution
obtained from beans roasted for 14 min (CR14; medium degree
of roast) was subjected to derivatization. R-Dicarbonyl com-
pounds were then analyzed by RP-HPLC-DAD by direct

8880 J. Agric. Food Chem., Vol. 55, No. 22, 2007
Daglia et al.
a
Table 2. Recovery Rates of Standard R-Dicarbonyl Compounds after SPE
standard
derivatized SPE-F1
c
d
c
e
compound
quinoxaline
concentration [µM]
peak area
ratio
peak area
ratio
75.2
75.2
79.3
75.2
928.861
941.621
1044.12
947.219
0.980
0.994
1.100
924.735
938.421
1053.281
948.898
0.974
0.989
1.110
2
2
5
-methylquinoxaline
,3-dimethylquinoxaline
-methylquinoxaline
b
a
b
c
d
e
All experiments were performed in triplicate. Internal standard. Expressed as arbitrary units. Ratio of standard quinoxaline derivatives to 5MQ peak area. Ratio
of derivatized R-dicarbonyl compounds (obtained from SPE-F1) to 5MQ peak area.
Table 3. Calibration Data and Limit of Quantification (LOQ) and Detection (LOD) of R-Dicarbonyl Compounds
compound
quinoxaline
linear range (µmol/L)
regression equation
correlation coefficient
LOQ (µmol/L)
LOD (µmol/L)
a
b
7.0–150.0
7.0–150.0
5.0–100.0
y ) 12.63x - 44.58
y ) 12.49x - 34.96
y ) 13.38x - 16.46
0.9989
0.9974
0.9999
0.18
0.20
0.06
0.05
0.05
0.02
2-methylquinoxaline
2,3-dimethylquinoxaline
a
b
y is the peak area in arbitrary units. x is the concentration of quinoxaline derivatives (µM).
a
Table 4. Method Precision for the Determination of R-Dicarbonyl Compounds in Coffee Extract (CR14) SPE-F1 Derivatized with 1,2-Diaminobenzene
mean content (µM) ( SD
compound
glyoxal
-methylglyoxal
diacetyl
day 1
day 2
day 3
RSD (%) intra-assay (day 1)
RSD (%) interassay
55.80 ( 2.02
80.66 ( 1.68
17.55 ( 0.47
54.28 ( 1.99
78.66 ( 1.54
17.06 ( 0.58
52.80 ( 1.56
80.39 ( 1.67
17.52 ( 0.46
3.62
2.08
2.67
3.41
2.04
2.89
2
a
All experiments were performed nine times.
Table 5. Recovery Rates of the Method for R-Dicarbonyl Compound Determination in Coffee Extracts SPE-F1
a
compound
quinoxaline
amount added (µg/L)
amount found (µg/L)
% recovery (n ) 3)
100.2
2.10
1.50
1.50
10.00
10.00
7.50
18.00
20.50
15.00
2.16
1.46
1.51
10.02
10.21
7.68
18.32
20.77
15.21
102.9
97.33
100.7
101.8
101.3
101.4
2-methylquinoxaline
102.1
102.4
2,3-dimethylquinoxaline
a
Values represent the mean of three determinations.
and UV spectra with those of the standard compounds. The
determination of derivatized glyoxal (Q) failed due to overlap-
ping peaks. Q was coeluted with a coffee compound whose
spectrum showed it to belong to the chlorogenic acid family.
Changes in chromatographic conditions never resulted in peak
separation.
To eliminate the interfering compounds, coffee extracts were
subjected to SPE before derivatization using a C18-SPE
cartridge. Glyoxal, methylglyoxal, and diacetyl were recovered
with water (SPE-F1), whereas elution of the less water soluble
chlorogenic acids (SPE-F2) required methanol.
centration of standard quinoxaline derivatives (µM) as a function
of peak area detected at 314 nm, corresponding to their UV
absorption maxima. Regression equations and correlation coef-
ficients, reported in Table 3, were calculated by the least squares
method. The limit of detection (LOD), calculated as the amount
of analyte required to obtain a signal-to-noise ratio of 3, was
0.05 µmol L- , for Q and 2MQ and 0.02 µmol L , for
2,3DMQ. The limit of quantification (LOQ), that is, the lowest
concentration required to yield a signal-to-noise ratio of 10, was
0.20, 0.18, and 0.06 µmol L- for Q, 2MQ, and 2,3DMQ,
respectively (Table 3).
1
-1
1
Freed of interfering components (Figure 1b), R-dicarbonyl
compounds could then be derivatized and analyzed simulta-
neously by RP-HPLC-DAD. In fact, the derivatized SPE-F1
Method repeatability (intra- and interassay) was evaluated
by analyzing the CR14 derivatized coffee SPE-F1 solution nine
times in three days (Table 4). The inter- and intraassay relative
standard deviation (RSD) of the R-dicarbonyl compounds was
consistently <3.6%.
The accuracy of the HPLC method was determined by
recovery tests performed by adding three different concentrations
of standard quinoxaline derivative solutions to coffee SPE-F1
before derivatization. Spiked samples were then subjected to
the entire procedure. The results showed recovery rates between
97 and 103% (Table 5).
Formation Kinetics of r-Dicarbonyl Compounds during
the Roasting Process. Green and roasted coffee extracts
obtained from beans with different degrees of roasting were
analyzed by RP-HPLC-DAD to verify R-dicarbonyl compounds
content in relation to degree of coffee bean roasting. The results
(Figure 2) showed that very small amounts (about 1 mg/100 g
coffee powder) of glyoxal and methylglyoxal occurred in green
(Figure 1c) exhibited three peaks with the same retention times
and UV spectra as the standard Q, 2MQ, and 2,3DMQ
chromatograms (Figure 1a). The standard glyoxal, methylgly-
oxal, and diacetyl mixture was subjected to SPE to check that
all R-dicarbonyl compounds had been recovered by the SPE
procedure. SPE-F1 given by the standard mixture was deriva-
tized with 1,2-diaminobenzene after adding 5MQ, and analyzed
by HPLC. The ratios between the peak areas (Table 2) of the
standard Q, 2MQ, and 2,3DMQ or R-dicarbonyl derivatized with
1
,2-diaminobenzene obtained from SPE-F1 to 5MQ peak area
at the same concentrations showed that recovery after SPE was
9
9.4–100.9% for the considered R-dicarbonyl compounds.
Method Validation. Coffee R-dicarbonyl compounds were
quantified by RP-HPLC-DAD using the external standard
method with six-point regression curve, by plotting the con-

Coffee R-Dicarbonyl Compounds Determination
J. Agric. Food Chem., Vol. 55, No. 22, 2007 8881
food, were determined in coffee extracts (coffee brews) obtained
from green and roasted coffee beans with different degrees of
roast.
Our data show that small amounts of glyoxal and methylg-
lyoxal occur naturally in green coffee beans, probably as
secondary lipid peroxidation products. Their concentrations
increase in the early phases of the roasting process (after 6–10
min) and then decline. Conversely, diacetyl is not found in green
coffee beans and forms later in the roasting process (14 min).
Therefore, light and medium roasted coffees had the highest
glyoxal and methylglyoxal content, whereas dark roasted coffee
contained smaller amounts of glyoxal, methylglyoxal, and
diacetyl. The reduction in their concentrations in the final phase
of the roasting process may be ascribed to their involvement in
the advanced stages of Maillard reaction, which result in brown
pigments (melanoidins) formed by self- and cross-condensation
of the aldehydes and ketones forming as fragmentation products
(1). Because of the high reactivity of R-dicarbonyl compounds,
which have been implicated in the glycation process, it would
be interesting to monitor their formation during the roasting
process, even though host defense mechanisms probably protect
humans efficiently against their action, since no adverse effect
of regular coffee drinking has been observed in several
epidemiological investigations of healthy individuals (23, 24).
In this study, R-dicarbonyl compounds were simultaneously
detected as quinoxaline derivatives using RP-HPLC-DAD.
The applied method previously used to determine dicarbonyl
compounds in wine, required some modifications and a
specific sample preparation. Because of the very complex
matrix of coffee extract (coffee brew), interfering compounds
such as chlorogenic derivatives had to be eliminated before
the derivatization reaction and elution conditions had to be
optimized. Checks of SPE and derivatization conditions to
verify recovery and yield, respectively, resulted in rates close
to 100%. The results of the validation procedure showed that
the proposed method is selective, precise, accurate, and
sensitive.
The findings reported here demonstrate that coffee R-dicar-
bonyl compound content depends on roasting conditions as well
as ex vivo antiperoxyl radical activity (25) and the antiadhesive
effects of Streptococcus mutans on hydroxyapatite beads (21),
previously investigated by our group. On the whole, these results
should encourage roasters to select roasting conditions that while
maintaining brew taste and aroma can at the same time enhance
their safety and healthiness.
Figure 2. Quantification of (a) glyoxal, (b) methylglyoxal, and (c) diacetyl
in green coffee beans and in beans with different degrees of roast (2–20
min).
coffee. Their content increased in the early phases of roasting;
diacetyl was not found in green beans and was detected only
after 14 min of roasting. Glyoxal content increased in the first
ABBREVIATIONS USED
Q, quinoxaline; 2MQ, 2-methylquinoxaline; 2,3DMQ, 2,3-
dimethylquinoxaline; 5MQ, 5-methylquinoxaline; CR, C. ro-
busta coffee sample; CR14, C. robusta coffee sample roasted
for 14 min.
6
min of roasting, peaking at 13.07 ( 0.39 mg/100 g, then
slowly decreased to 1.93 ( 0.05 at 20 min. Methylglyoxal
exhibited a similar behavior; its concentration (21.19 ( 0.42
mg/100 g) peaked at 10 min and subsequently declined until
the end of the roasting process. The maximum concentration
of methylglyoxal was about twice that of glyoxal. The formation
kinetics of diacetyl was different; the compound was found at
a very low concentrations in roasted coffee, peaking between
ACKNOWLEDGMENT
The work was supported by a grant from MIUR 2005 to G.G.
and took place within activities of the Interdepartmental Center
of Food Quality and Safety (CISQUA). We thank MOKA SIR’S
s.r.l. (Pavia, Italy) for providing the coffee beans and the pilot
roasting apparatus.
1
4 and 16 min then remaining stable until the end of roasting
(
peak concentration 2.28 ( 0.07 mg/100 g).
Recovery tests were carried out by adding the standard
solutions to coffee SPE-F1 before derivatization. Spiked samples
were then subjected to the entire procedure.
LITERATURE CITED
(
1) Powrie, W. D.; Wu, C. H.; Molund, V. P. Browing reaction
systems as sources of mutagens and antimutagens. EnViron. Health
Perspect. 1986, 67, 47–54.
In conclusion, glyoxal, methylglyoxal, and diacetyl, bioactive
compounds formed as Maillard reaction products in heat-treated

8882 J. Agric. Food Chem., Vol. 55, No. 22, 2007
Daglia et al.
(
2) Thornalley, P. J. Dicarbonyl intermediates in the Maillard reaction.
(15) Forster, A.; Kühne, Y.; Henle, T. Studies on absorption and
elimination of dietary Maillard reaction products. Ann. N. Y. Acad.
Sci. 2005, 1043, 474–481.
Ann. N. Y. Acad. Sci. 2005, 1043, 1–7.
(
(
(
3) Shibamoto, T. Analytical methods for trace levels of reactive
carbonyl compounds formed in lipid peroxidation system.
J. Pharm. Biomed. Anal. 2006, 41, 12–25.
4) Ariza, R. R.; Dorado, G.; Barbanco, M.; Pueyo, C. Study of the
cause of direct-acting mutagenicity in coffee and tea using Ara
test in Salmonella typhimurium. Mutat. Res. 1988, 201, 89–96.
5) Kalapos, M. P. Methylglyoxal in living organisms: chemistry,
biochemistry, toxicology and biological implications. Toxicol. Lett.
(
16) Li, R.; Kenyon, G. L. A spectrophotometric determination of
R-dicarbonyl compounds and its application to the enzymatic
formation of a-ketobutyrate. Anal. Biochem. 1995, 230, 37–40.
17) Hayashi, T.; Shibamoto, T. Analysis of methylglyoxal in Foods
and beverages. J. Agric. Food Chem. 1985, 33, 1090–1093.
18) de Revel, G.; Pripis-Nicolau, L.; Berbe, J. C.; Bertrand, A. The
detection of R-dicarbonyl compounds in wine by formation of
quinoxaline derivatives. J. Sci. Food Agric. 2000, 80, 102–108.
19) Chaplen, F. W. R.; Fahl, W. E.; Cameron, D. C. Method for
determination of free intracellular and extracellular methylglyoxal
in animal cells grown in culture. Anal. Biochem. 1996, 238, 171–
(
(
1
999, 110, 145–175.
(
(
6) Bird, R. P.; Draper, H. H.; Valli, V. E. O. Toxicological evaluation
of malonaldehyde: a 12-month study of mice. J. Toxicol. EnViron.
Health 1982, 10, 897–905.
1
78.
(
(
(
7) Baskaran, S.; Balasubramanian, K. H. Toxicity of methylglyoxal
(
20) Hayashi, T.; Reece, C. A.; Shibamoto, J. In Characterization and
Measurement of FlaVor Compounds; Bill, D. D., Mussinam, C. J.,
Eds.; ACS Symposium Series 289; ACS: Washington, DC, 1985;
pp 61, 78.
towards rat enterocytes and colonocytes. Biochem. Intern. 1990,
2
1, 165–174.
8) Nguyen, C. V. Toxicity of AGEs generated from the Maillard
reaction: on the relationship of food-AGEs and biological-AGEs.
Mol. Nutr. Food Res. 2006, 50, 1140–1149.
(21) Daglia, M.; Tarsi, R.; Papetti, A.; Grisoli, P.; Dacarro, C.; Pruzzo,
C.; Gazzani, G. Anti-adhesive effect of green and roasted coffee
on Streptococcus mutans adhesive properties on saliva coated
hydroxyapatite beads. J. Agric. Food Chem. 2002, 50, 1225–1229.
(22) Daglia, M.; Racchi, M.; Papetti, A.; Lanni, C.; Govoni, S.;
Gazzani, G. In vitro and ex vivo antihydroxyl radical activity of
green and roasted coffee. J. Agric. Food Chem. 2004, 52, 1700–
9) Henle, T.; Miyata, T. Advanced glycation end products in uremia.
AdV. Renal Replacement Ther. 2003, 10, 321–331.
(
10) Reddy, V. P.; Obrenovich, M. E.; Atwood, C. S.; Perry, G.; Smith,
M. A. Involvement of Maillard reactions in Alzheimer disease.
Neurotoxicol. Res. 2002, 4, 191–209.
1
704.
(
11) Mukhopadhyay, S.; Sen, S.; Majhi, B.; Das, K. P.; Kar, M.
Methylglyoxal elevation is associated with oxidative stress in
rheumatoid arthritis. Free Radical Res. 2007, 41, 507–514.
12) Chang, T.; Wu, L. Methylglioxal, oxidative stress, and hyperten-
sion. Can. J. Physiol. Pharmacol. 2006, 84, 1229–1238.
13) Shangari, N.; Mehta, R.; O’ Brien, P. J. Hepatocyte susceptibility
to glyoxal is dependent on cell thiamin content. Chem. Biol.
Interact. 2007, 165, 146–154.
(
(
(
23) Higdon, J. V.; Frei, B. Coffee and health: a review of recent human
research. Crit. ReV. Food Sci. Nutr. 2006, 46, 101–123.
24) Dorea, J. G.; da Costa, T. H. M. Is coffee a functional food? Br. J.
Nutr. 2005, 93, 773–782.
25) Daglia, M.; Papetti, A.; Gregotti, C.; Bertè, F.; Gazzani, G. In
vitro antioxidant and ex vivo protective activities of green and
roasted coffee. J. Agric. Food Chem. 2000, 48, 1449–1454.
(
(
(
14) Goldberg, T.; Cai, W.; Melpomeni, P.; Dardaine, V.; Baliga, B. S.;
Uribarri, J.; Vlassara, H. Advanced glycoxidation end products
in commonly consumed foods. J Am. Diet. Assoc. 2004, 104,
Received for review June 28, 2007. Revised manuscript received August
0, 2007. Accepted August 15, 2007.
1
1
287–1291.
JF071917L