Toward a kinetic model for acrylamide formation in a glucose-asparagine reaction system

Article abstract of DOI:10.1021/jf050504m

A kinetic model for the formation of acrylamide in a glucose-asparagine reaction system is proposed. Equimolar solutions (0.2 M) of glucose and asparagine were heated at different temperatures (120-200°C) at pH 6.8. Besides the reactants, acrylamide, fructose, and melanoidins were quantified after predetermined heating times (0-45 min). Multiresponse modeling by use of nonlinear regression with the determinant criterion was used to estimate model parameters. The proposed model resulted in a reasonable estimation for the formation of acrylamide in an aqueous model system, although the behavior of glucose, fructose, and asparagine was slightly underestimated. The formation of acrylamide reached its maximum when the concentration of sugars was reduced to about 0. This supported previous research, showing that a carbonyl source is needed for the formation of acrylamide from asparagine. Furthermore, it is observed that acrylamide is an intermediate of the Maillard reaction rather than an end product, which implies that it is also subject to a degradation reaction.

Full text of DOI:10.1021/jf050504m

J. Agric. Food Chem. 2005, 53, 6133−6139 6133
Toward a Kinetic Model for Acrylamide Formation in a
Glucose Asparagine Reaction System
−
JEROEN J. KNOL,†,§ WIL A. M. VAN LOON,‡,§ JOZEF P. H. LINSSEN,*,†,‡
ANNE-LAURE RUCK,‡ MARTINUS A. J. S. VAN BOEKEL,† AND
‡
ALPHONS G. J. VORAGEN
Product Design and Quality Management Group and Laboratory of Food Chemistry,
Department of Agrotechnology and Food Science, Wageningen University,
Post Office Box 8129, 6700 EV Wageningen, The Netherlands
A kinetic model for the formation of acrylamide in a glucose-asparagine reaction system is pro-
posed. Equimolar solutions (0.2 M) of glucose and asparagine were heated at different tempera-
tures (120-200 °C) at pH 6.8. Besides the reactants, acrylamide, fructose, and melanoidins
were quantified after predetermined heating times (0-45 min). Multiresponse modeling by use of
nonlinear regression with the determinant criterion was used to estimate model parameters. The
proposed model resulted in a reasonable estimation for the formation of acrylamide in an aqueous
model system, although the behavior of glucose, fructose, and asparagine was slightly underestimated.
The formation of acrylamide reached its maximum when the concentration of sugars was reduced to
about 0. This supported previous research, showing that a carbonyl source is needed for the formation
of acrylamide from asparagine. Furthermore, it is observed that acrylamide is an intermediate of the
Maillard reaction rather than an end product, which implies that it is also subject to a degradation
reaction.
KEYWORDS: Acrylamide; multiresponse kinetic modeling; Maillard reaction
INTRODUCTION
review about this subject was published by Wenzl et al. (12).
Evaluation of the health risk resulted in the recommendation
that intake levels of acrylamide should be reduced to protect
health, although there is no direct risk on cancer with the current
intake of about 35 µg day-1 person-1 (13, 14). Possible ways
to reduce the formation of acrylamide include reducing the
temperature, reducing precursors during processing, and influ-
encing the pathway of formation by changing the pH (15-17).
More information on research, analysis, formation, and control
of acrylamide is available in an extensive review that was
published recently (18).
The Maillard reaction is highly complex and cannot be
described with simple reaction kinetics. To better understand
mechanisms behind the formation of acrylamide, an advanced
approach is necessary (19). Kinetic modeling, using multire-
sponse models, has proven to be a useful tool to enable further
unraveling of reaction mechanisms in the Maillard reaction (20,
21). Knowledge of kinetics provides the tool to predict the
effect of certain time-temperature combinations in a quantita-
tive way.
In April 2002, Tareke et al. (1) reported that several heat-
processed foods contain relatively high amounts of acrylamide.
These results were confirmed by other research groups: fried
and baked potato products were found to contain the highest
amounts of acrylamide, followed by cereal products, whereas
only low amounts were detected in meat products (2-4). A
high amount in foods is of major concern as acrylamide is
known to be a neurotoxic, genotoxic, and carcinogenic com-
pound in animals (5), which is classified by the IARC (6) as a
probable human carcinogen.
About half a year later the suggestion was published that
acrylamide is formed in the Maillard reaction and that asparagine
is the precursor (7-9). Further elucidation of the pathway by
use of 13C-labeled isotopes showed that asparagine needs a
carbonyl source to form acrylamide (10, 11). According to
Friedman (5), the proposed mechanisms predict that acrylamide
may result from the general reaction of asparagine with any
aldehyde or ketone. Simultaneously, many researchers focused
on optimization of methods for determination of acrylamide. A
Model systems of asparagine and a carbonyl such as glucose,
both dry and in aqueous solutions, have been used mainly to
study the pathway of acrylamide formation (3, 22). The yield
of acrylamide was about 0.1% of asparagine, and a wide range
of other compounds (e.g., melanoidins) was found. Furthermore,
an effect of time and temperature on both formation and
* To whom correspondence should be addressed: telephone +31-317-
483548; fax +31-317-483669; e-mail jozef.linssen@wur.nl.
† Product Design and Quality Management Group.
‡ Laboratory of Food Chemistry.
§ These authors contributed equally.
10.1021/jf050504m CCC: $30.25
© 2005 American Chemical Society
Published on Web 06/28/2005

6134 J. Agric. Food Chem., Vol. 53, No. 15, 2005
Knol et al.
degradation of acrylamide has been found. Recently, a kinetic
model for the formation of acrylamide in potato, wheat, and
rye model systems was published by Wedzicha et al. (23).
However, using food model systems to gain kinetic insights into
reactions can cause disturbances by the presence of food matrix
components. Therefore, the aim of this study was to attempt to
model the formation of acrylamide in a glucose-asparagine
reaction system at different time-temperature combinations.
Once information about the kinetics of formation of acrylamide
is obtained, one can move on to more realistic systems to
describe the effect of complicating factors as present in real
foods.
MATERIALS AND METHODS
Figure 1. (A) Reaction network for the formation of acrylamide from
asparagine through early Maillard reaction adapted from Stadler et al.
(26). (B) Simplified reaction network for the formation of acrylamide from
asparagine and glucose/fructose through early Maillard reaction. Asn,
asparagine; Decarb. Amadori comp., decarboxylated Amadori compound;
Product X, assumed product(s) formed from acrylamide.
Chemicals. The following compounds were obtained commer-
cially: D-glucose, D-fructose, Na2HPO4, and KH2PO4 (Merck, Darm-
stadt, Germany) and L-asparagine (Fluka, Buchs, Switzerland). All
chemicals used in this study were of analytical grade.
Preparation of Reaction Mixtures. Equimolar solutions of glu-
cose and asparagine (0.2 M) were prepared in phosphate buffer (0.1
M, pH 6.8). Samples (10 mL) were heated in hermetically closed
screw-capped glass tubes (Schott, 16 × 160 mm) at 120, 140, 160,
180, and 200 °C in an oil bath. The tubes were immersed in the oil
up to the cap. At predetermined heating times (0, 1, 2, 4, 9, 15, 30,
and 45 min), samples were taken and immediately cooled in ice and
stored at -20 °C prior to analysis. Experiments were carried out in
triplicate.
extinction coefficient of 282 L/mol‚cm (30), a value derived for
melanoidins formed from glucose and asparagine. The concentration
of melanoidins is thus expressed as moles of sugar incorporated in the
brown polymers. Although melanoidins are of course complex mixtures
of various molecules, several studies have shown that they can be
quantified in this way (20, 31, 32).
Kinetic Modeling. A kinetic model was derived from the proposed
reaction network taken from Stadler et al. (33) (shown in Figure 1).
Our arguments for using the reaction network published by Stadler et
al. (33) and the steps taken to derive a kinetic model will be discussed
further on. For each reaction step, a differential equation was set up,
and these were translated into a mathematical model. The differential
equations were solved by numerical integration. For both numerical
integration and parameter estimation, the software package Athena
Visual Studio v. 10.0 (34) was used. The parameters of the model,
that is, the rate constants and activation energies, were estimated by
nonlinear regression by use of the determinant criterion (20). For
modeling purposes, the individually measured concentrations were used
(measured in triplicate). For the visualization of the experimental results
(Figure 2), the average values with their standard deviation (represented
by the error bars) were used.
Analysis of Acrylamide. The analysis of acrylamide was based on
the HPLC method described by Barber et al. (24). Prior to analysis,
samples were diluted in Millipore water (1:10) and centrifuged for 5
min at 16000g. The HPLC system used for the analysis consisted of a
P4000 pump, an AS3000 autosampler, and a UV3000 detector (Thermo
Separation Products, San Jose, CA). A Synergi 4 µm hydro reversed-
phase C18 column (80 Å, 250 × 2.00 mm) with AJO-4286 guard
column (Phenomenex, Aschaffenburg, Germany) was used for the
analysis (25). Samples (20 µL) were eluted isocratically at 20 °C with
a solution containing 1% methanol and 99% 5 mM heptanesulfonic
acid in Millipore water at a flow rate of 0.2 mL/min. Acrylamide (tr )
6.2 min) was detected by its absorbance at 200 nm with a UV detector
and quantified by the external standard procedure with a calibration
curve. The limit of detection obtained (10 µg kg-1) was similar to what
Barber et al. (24) found.
Analysis of Sugars. The diluted samples (1:10) were analyzed for
sugars by HPLC (see Analysis of Acrylamide) by use of the method
of Martins et al. (26). An ion-exchange column (ION-300, Interaction
Chromatography Inc., San Jose, CA) with guard column was used for
the analysis at 85 °C. Sulfuric acid (2.5 mM) in Millipore water was
used as the eluent with a flow rate of 0.4 mL/min. Sugars were detected
by monitoring the refractive index and quantified by the external
standard procedure with a calibration curve. The formation of sugar
isomers in the Maillard reaction has been confirmed by HPLC and
electrochemical detection recently (27, 28).
Analysis of Asparagine. Samples were diluted 1:1000 with Millipore
water and the EZ:faast amino acid analysis kit (Phenomenex, Aschaffen-
burg, Germany) was used for determination of asparagine (29).
Norvaline was added as an internal standard (20 nmol) before sample
preparation. After sample preparation, samples were analyzed on a Carlo
Erba GC5300 system (Interscience BV, Breda, The Netherlands) on a
Zebron amino acid column (10 m × 0.25 mm) that was included in
the kit. Samples (2.5 µL) were injected onto the column at 250 °C
(split ratio 1:15). The oven temperature started at 110 °C and was
programmed to heat to 250 °C at a rate of 20 °C/min, followed by a 1
min isothermal hold. The external standard procedure was used for
quantification.
RESULTS AND DISCUSSION
Identification and Quantification of Reactants and Main
Products. During heating of the glucose-asparagine reaction
mixtures, the concentrations of the reactants decreased over time.
As expected, the rate of degradation of glucose and asparagine
increased with temperature. At 180 and 200 °C, a complete loss
of glucose was observed after 9 and 5 min, respectively (Figure
2A). The loss of asparagine (Figure 2B) was slower compared
to the loss of glucose. A complete loss of asparagine was mea-
sured at 180 and 200 °C after 30 and 9 min, respectively. The
loss of sugar being faster than the loss of amino acid is in
line with what has been found in a glucose-glycine reaction
system by Martins and Van Boekel (20). The slower loss can
possibly be explained by the regeneration of asparagine from
the initial condensation products such as the Amadori rear-
rangement product and the possible formation of diglucosyl-
amine (20).
As fructose and mannose are known to be formed from
glucose by isomerization (26), some formation of these com-
pounds was expected. Analysis showed no formation of man-
nose, but fructose was indeed formed (Figure 2C). This
corresponds with findings of Brands and Van Boekel (35) and
Martins and Van Boekel (20), where only fructose was found
as an isomerization product of glucose. There was a clear
Analysis of Melanoidins. Quantification of melanoidins was
performed by measuring the absorbance at 470 nm on a spectropho-
tometer (UV-1601, Shimadzu, Kyoto, Japan). When necessary, the
samples were diluted with Millipore water. The concentration of
melanoidins was calculated from the Lambert-Beer equation with an

Kinetic Modeling of Acrylamide Formation
J. Agric. Food Chem., Vol. 53, No. 15, 2005 6135
Figure 2. Experimental data of glucose/asparagine (0.2 mol/L, pH 6.8) aqueous reaction system heated at 120 (
0), 140 (/), 160 (2), 180 (O), and 200
°
C ( ). (A) Loss of glucose; (B) loss of asparagine; (C) formation of fructose; (D) formation of acrylamide; (E) formation of melanoidins.
[
maximum at 160, 180, and 200 °C in the formation of fructose
at 9, 4, and 2 min, respectively. For the reaction mixtures heated
at 180 and 200 °C, these maximum values were followed by a
complete loss after 30 and 9 min, respectively. The decrease in
fructose concentration can be explained by the reversible
isomerization reaction between glucose and fructose so that
fructose re-forms back into glucose, which is then consumed.
However, it is also quite likely that fructose participates in the
formation of acrylamide (36-38).
The initial rate of formation of acrylamide increased with
temperature (Figure 2D). At 140 and 160 °C, the increase in
acrylamide concentration was followed by a steady state in
which the formation of acrylamide was probably in equilibrium
with its degradation. At 180 and 200 °C, the increase was
followed by a fast decrease. The maximal acrylamide formation
at 180 and 200 °C corresponded with the time at which the
concentration of glucose and fructose reached zero (Figure
2A,C). This is in line with the findings of Yaylayan et al. (10),
who showed that asparagine needs a carbonyl source to form
acrylamide. The decrease in acrylamide can be explained by
its reactive nature. The increase and subsequent decrease of the
acrylamide concentration at 180 and 200 °C is a typical behavior
for an intermediate (20). Acrylamide may have polymerized or
reacted further in Michael-type addition reactions (33). Wedz-
icha et al. (23) attributed the loss of acrylamide after prolonged
heating at 180 °C to the reactivity of acrylamide with food
components. These pathways have not been fully elucidated,
and further research in this area is necessary to expand the
reaction network of the Maillard reaction between asparagine
and glucose.
Melanoidins are known as the main end products of the
Maillard reaction. The structure of these brown nitrogenous
polymers and copolymers is largely unknown, which makes
quantification difficult. Previous studies have shown that it is
possible to relate the absorbance to the number of sugar
molecules incorporated in the melanoidins in aqueous model
systems (20, 30, 32). At 160, 180, and 200 °C, we observed
after some time the formation of insoluble particles, which are
probably high molecular weight melanoidins. The absorbance
measurements were hindered by the scattering effect of these
particles. Moreover, the values for absorbance decreased during
formation of these particles, causing an underestimation of the
melanoidins concentration. Because of this, we excluded the
results for the reaction mixtures where these solid compounds
were found. This was the case for the reaction mixtures heated
for 45 min at 160, 180, and 200 °C, for 30 min at 180 and 200
°C, and for 15 min at 200 °C. Considerable browning was
observed in samples during heating. The induction time, during
which no browning was detected, decreased with increasing
temperature (Figure 2E). Furthermore, the formation of acryl-
amide could not be associated directly with the degree of
coloring of the reaction mixtures at 160, 180, and 200 °C after
15, 9, and 4 min, respectively. This is due to their difference in
status: acrylamide is an intermediate, whereas melanoidins are
the main end products of the Maillard reaction. As our proposed
kinetic model applies only to an aqueous model system, this
relationship cannot be used yet directly in food systems. The
relationship between acrylamide and color formation in foods,
for instance the one found in yeast-leavened wheat bread by
Surdyk et al. (37), is induced by effects such as concentration

6136 J. Agric. Food Chem., Vol. 53, No. 15, 2005
Knol et al.
gradients, diffusion and evaporation, and the presence of other
compounds such as other amino acids and sugars and buffering
agents. Therefore, more research is needed with model systems
that can take these effects into consideration.
where
-Ea
RTav
X ) k0 exp
(3)
(4)
(5)
(
)
Proposal of a Kinetic Model Based on a Reaction
Network. By analyzing the reactant degradation and the
formation of the main products in the Maillard reaction between
glucose and asparagine, a kinetic model could be proposed. The
mathematical model derived from the kinetic model was then
confronted with the experimental data to test the hypothesized
model. Several reaction networks for the formation of acryl-
amide in foods have been proposed. These publications have
revealed the Maillard reaction as the major reaction pathway
involved (3, 7-9). The hypothesis proposed by Mottram and
co-workers (7, 23) emphasizes the pathway of Strecker degrada-
tion. The hypothesis proposed by Stadler et al. (8), however,
pointed toward glycoconjugates such as N-glycosides formed
in the early Maillard reaction as key intermediates. A recent
publication by Stadler et al. (33) supports their hypothesis as
well as the work published by Zyzak et al. (11) and Yaylayan
et al. (10). Furthermore, Stadler et al. (33) suggest that the
mechanism proposed by Mottram et al. (7) is not a main one.
We did not pursue the identification and quantification of all
these suggested intermediates in this study and, therefore, we
are not able to shed more light on this matter. However, we
chose the reaction network proposed by Stadler et al. (8), shown
in Figure 1A, as initial basis for the proposal of a kinetic model,
as their proposed reaction network has been more elaborately
described in recent literature (33). To fit the model to the
experimental data, this reaction network was first adapted to
reduce the number of parameters. Because the intermediates
(e.g., N-glycosyl conjugate, azomethine ylide) were not analyzed
and quantified in our experiments, the number of parameters
was relatively high in comparison to the number of measured
responses. Second, the new reaction network proposed in Figure
1B also takes the measured responses for fructose and mel-
anoidins and the observed loss of acrylamide into account. The
formation of fructose was suggested by including the reversible
isomerization reaction between glucose and fructose. Further-
more, the reaction between asparagine and fructose was added,
since fructose is also a reactant in the formation of acrylamide
(36-38). The formation of melanoidins was suggested to result
from further reaction of Schiff base, and the loss of acrylamide
was accounted for by the putative formation of hitherto unknown
product(s).
1 1
1
Y )
-
(
)
R T Tav
ΣT
Tav
)
n
The model was fitted to the data obtained at 120, 140, 160,
180, and 200 °C at pH 6.8 simultaneously by substituting eq 2
for the rate constants.
At the lower reaction temperatures (120 and 140 °C), quite
a good fit was observed for all compounds (Figures 3A-E).
For the acrylamide and melanoidins concentrations, the model
fits the data well at all temperatures (Figure 3D,E). However,
the good fit for the melanoidins concentration may have been
partially caused by the lack of data for the higher temperatures.
For the estimation of the acrylamide concentration, the model
was not restrained by experimental data for the products formed
in the degradation reaction from acrylamide, and therefore, the
model was able to fit the loss of acrylamide (k6) to the
experimental observations for acrylamide. More knowledge
about the reaction products from acrylamide would validate the
estimated rate constants for the formation and loss of acrylamide.
The observed lack of fit for the individual responses was
rather high, especially for fructose and at the higher tempera-
tures. The model underestimated the decrease in glucose
concentration, especially during the first 10 min (Figure 3A).
The increase in fructose was also underestimated, but the model
predicted the tendencies as found in the experimental observa-
tions (Figure 3C). The same goes for asparagine, where there
was an underestimation for the decrease (Figure 3B). These
underestimations are undoubtedly caused by the limitations of
the present model. Reactions of glucose and fructose are limited
to the isomerization reaction and the reaction with asparagine
to form the Schiff base. The model does not take into account
the degradation of sugars to carbohydrate fragments and the
formation of compounds via the Strecker degradation route.
Therefore, this model is currently under further investigation,
but nevertheless the results already show some remarkable
phenomena that are worth discussing.
The estimated parameters, that is, rate constants and activation
energies, are shown in Table 1. The rate constants for the
different reaction temperatures have been derived from the
estimated values found for X. The estimated values of X are
the same as the derived values for the rate constants at 160 °C,
as Tav in this case is 160 °C. The temperature dependence is
consistent for all six reactions, but the precision of estimation
for the parameters is low. The average 95% highest posterior
density interval is about (27% for the rate constants and (11%
for the activation energies. The activation energy for the
formation of the Schiff base from fructose and asparagine is
about 2 times higher than the activation energy for the same
formation reaction with glucose and asparagine as reactants,
suggesting that the contribution from fructose becomes more
important at higher temperatures. The rate of the Schiff base
formation reaction from fructose and asparagine (k3) increases,
therefore, much faster with increasing temperature than the rate
constant of reaction 1 (k1). At 120 °C, k3 is 35% lower than k1,
but at 180 °C k3 is 2-3 times higher than k1. This is consistent
with the observations by Robert et al. (36) that fructose is more
efficient in the formation of acrylamide than glucose by a factor
Test of the Hypothesized Mechanism. The proposed reac-
tion model presented in Figure 1B was translated into a
mathematical model. For each reaction step, a differential
equation was set up by use of the law of mass action, and the
obtained differential equations were solved by numerical
integration. The temperature dependence, which plays a major
role in the Maillard reaction, was also taken into account by
including an Arrhenius relationship between the rate constants
(k) of the different reactions. The Arrhenius equation
-Ea
RT
k ) k exp
(1)
(
)
0
where k0 is the so-called frequency factor, R is the gas constant
[8.314 J/(mol‚K)], Ea is the activation energy (joules/mole), and
T is the absolute temperature (kelvins), was reparametrized to
avoid statistical problems in estimation (39) as follows:
k ) X exp(-YEa)
(2)

Kinetic Modeling of Acrylamide Formation
J. Agric. Food Chem., Vol. 53, No. 15, 2005 6137
Figure 3. Model fit (lines) to experimental data (symbols) of glucose/asparagine (0.2 mol/L, pH 6.8) aqueous reaction system heated at 120 (
140 ( − ‚ −), 160 ( ‚‚‚), 180 ( − − −) and 200 C ( ). (A) Loss of glucose; (B) loss of asparagine; (C) formation of fructose; (D) formation
of acrylamide; (E) formation of melanoidins.
0; − ‚‚ −),
/
;
2
;
O;
°
[; s
Table 1. Estimates of Rate Constantsa and Activation Energiesa as Found by Kinetic Modeling for the Proposed Kinetic Model
-
-1
k (×
10 3 min
)
120
0.131
°
C
140
°
C
160
0.668
°
C
180
°
C
200
°
C
Ea (kJ/mol)
k1
k2
k3
k4
k5
k6
±
0.025
1.2
0.045
0.081
3.5
0.308
±
±
±
±
±
±
0.049
3.0
0.14
0.23
4.6
±
0.13
9.6
0.44
0.51
5.9
1.35
±
±
±
±
±
±
0.36
35
1.6
1.2
8.9
45
2.58
±
±
±
±
±
±
0.89
114
6.4
4.3
16
57.6
81.7
102
±
±
±
±
±
±
8.0
8.6
14
4.98
0.0819
0.176
15.7
±
±
±
±
±
16.7
0.369
0.712
28.4
50.1
1.45
2.53
48.7
88.1
±
±
±
±
±
136
5.04
8.05
79.6
250
341
15.8
23.2
125
94.4
40.1
85.1
11
5.0
14
7.96
5.1
28.1
13
25
650
136
a
±
95% highest posterior density (HPD) interval.
Table 2. Model Discrimination Results for Our Proposed Model with
(A) or without (B) the Reversible Isomerization Reaction from Fructose
to Glucosea
of about 3 at 180 °C. The higher reactivity of fructose with
asparagine in the formation of the Schiff base could make the
reversible isomerization step of fructose to glucose insignificant.
Therefore, model discrimination was performed for our proposed
model with or without the reversible isomerization reaction from
fructose to glucose. Model discrimination can be used to provide
information about the most plausible model (20). In this case
we used the Akaike criterion (the model with the lowest value
is preferred) and the posterior probability (the model with the
highest posterior probability is preferred) to assess the most
plausible model (40, 41). The results from our test, shown in
Table 2, support the model without the reversible isomerization
reaction.
model
p
SS
n
AICc
∆
AICc
PPB
PPS
A
B
16
14
1.32
1.29
×
×
105
105
600
600
6505.2
6477.9
27.3
0.0
−
89.83
0.234
0.757
−89.33
a p, number of parameters; SS, residual sum of squares; n, number of data
points including the replicates; AICc, Akaike criterion; AICc,(AICc difference with
∆
the smallest value taken as reference; PPB, log10 of posterior probability; PPS,
normalized posterior probability share.
of acrylamide in an aqueous model system. The kinetic model
supports the observations (11) that for the formation of
acrylamide from asparagine a carbonyl source is needed, and
our results suggest an important role for fructose, especially at
the higher temperatures. Fructose is invariably formed from
glucose isomerization. Furthermore, the experimental observa-
The aim of this study was to work toward a kinetic model
for the formation of acrylamide in the Maillard reaction of
glucose and asparagine at an initial pH of 6.8. Despite the current
underestimation for the behavior of sugars and asparagine, the
proposed model gives a reasonable estimation for the formation

6138 J. Agric. Food Chem., Vol. 53, No. 15, 2005
Knol et al.
(17) Biedermann-Brem, S.; Noti, A.; Grob, K.; Imhof, D.; Bazzocco,
D.; Pfefferle, A. How much reducing sugar may potatoes contain
to avoid excessive acrylamide formation during roasting and
baking? Eur. Food Res. Technol. 2003, 217, 369-373.
(18) Taeymans, D.; Wood, J.; Ashby, P.; Blank, I.; Studer, A.; Stadler,
R. H.; Gonde´, P.; Van Eijck, P.; Lalljie, S.; Lingnert, H.;
Lindblom, M.; Matissek, R.; Mu¨ller, D.; Tallmadge, D.; O’Brien,
J.; Thompson, S.; Silvani, D.; Whitmore, T. A review of
acrylamide: an industry perspective on research, analysis,
formation, and control. Crit. ReV. Food Sci. Nutr. 2004, 44, 323-
347.
(19) Van Boekel, M. A. J. S. Kinetic aspects of the Maillard
reaction: a critical review. Nahrung 2001, 45, 150-159.
(20) Martins, S. I. F. S.; Van Boekel, M. A. J. S. A kinetic model for
the glucose/glycine Maillard reaction pathways. Food Chem.
2005, 90, 257-269.
(21) Mundt, S.; Wedzicha, B. L. A kinetic model for the glucose-
fructose-glycine browning reaction. J. Agric. Food Chem. 2003,
51, 3651-3655.
(22) Yasuhara, A.; Tanaka, Y.; Hengel, M.; Shibamoto, T. Gas
chromatographic investigation of acrylamide formation in brown-
ing model systems. J. Agric. Food Chem. 2003, 51, 3999-
4003.
tions and the kinetic modeling suggested that acrylamide is not
an end product in the Maillard reaction. The behavior of
acrylamide suggests that it is an intermediate. The multiresponse
model derived in this study is a first step into the realization of
a tool that can be used to predict how acrylamide reduction in
foods containing asparagine and reducing sugars can be ac-
complished. Further research is on the way to determine
intermediate reaction products in order to extend the kinetic
model. In addition, the formation of compounds via the Strecker
degradation route and the degradation of sugars into carbohy-
drate fragments will be investigated to improve the current
kinetic model in estimating the behavior of sugars and aspar-
agine.
LITERATURE CITED
(1) Tareke, E.; Rydberg, P.; Karlsson, P.; Eriksson, S.; To¨rnqvist,
M. Analysis of acrylamide, a carcinogen formed in heated
foodstuffs. J. Agric. Food Chem. 2002, 50, 4998-5006.
(2) Ahn, J. S.; Castle, L.; Clarke, D. B.; Lloyd, A. S.; Philo, M. R.;
Speck, D. R. Verification of the findings of acrylamide in heated
foods. Food Addit. Contam. 2002, 19, 1116-1124.
(3) Becalski, A.; Lau, B. P.-Y.; Lewis, D.; Seaman, S. W. Acryl-
amide in foods: Occurrence, sources, and modeling. J. Agric.
Food Chem. 2003, 51, 802-808.
(4) Konings, E. J. M.; Baars, A. J.; Van Klaveren, J. D.; Spanjer,
M. C.; Rensen, P. M.; Hiemstra, M.; Van Kooij, J. A.; Peters,
P. W. J. Acrylamide exposure from foods of the Dutch population
and an assessment of the consequent risks. Food Chem. Toxicol.
2003, 41, 1569-1579.
(5) Friedman, M. Chemistry, biochemistry, and safety of acrylamide.
A review. J. Agric. Food Chem. 2003, 51, 4504-4526.
(6) IARC. Acrylamide. In IARC Monographs on the eValuation of
carcinogen risk to humans: some industrial chemicals; Inter-
national Agency for Research on Cancer: Lyon, France, 1994;
Vol. 60, pp 389-433.
(7) Mottram, D. S.; Wedzicha, B. L.; Dodson, A. T. Acrylamide is
formed in the Maillard reaction. Nature 2002, 419, 448-449.
(8) Stadler, R. H.; Blank, I.; Varga, N.; Robert, F.; Hau, J.; Guy, P.
A.; Robert, M.-C.; Riediker, S. Acrylamide from Maillard
reaction products. Nature 2002, 419, 449-450.
(9) Weisshaar, R.; Gutsche, B. Formation of acrylamide in heated
potato products. Model experiments pointing to asparagine as
precursor. Deutsch. Lebensm.-Rundsch. 2002, 98, 397-400.
(10) Yaylayan, V. A.; Wnorowski, A.; Perez Locas, C. Why aspar-
agine needs carbohydrates to generate acrylamide. J. Agric. Food
Chem. 2003, 51, 1753-1757.
(11) Zyzak, D. V.; Sanders, R. A.; Stojanovic, M.; Tallmadge, D.
H.; Loye Eberhart, B.; Ewald, D. K.; Gruber, D. C.; Morsch, T.
R.; Strothers, M. A.; Rizzi, G. P.; Villagran, M. D. Acrylamide
formation mechanism in heated foods. J. Agric. Food Chem.
2003, 51, 4782-4787.
(23) Wedzicha, B. L.; Mottram, D. S.; Elmore, J. S.; Koutsidis, G.;
Dodson, A. T. Kinetic models as a route to control acrylamide
formation in food. In Chemistry and Safety of Acrylamide in
Food; M. Friedman and D. S. Mottram, Eds.; Springer: New
York, 2005; pp 235-253.
(24) Barber, D. S.; Hunt, J.; LoPachin, R. M.; Ehrich, M. Determi-
nation of acrylamide and glycidamide in rat plasma by reversed-
phase high performance liquid chromatography. J. Chromatogr.
B 2001, 758, 289-293.
(25) Peng, L.; Farkas, T.; Loo, L.; Dixon, A.; Teuscher, J.; Kallury,
K. Rapid and reproducible extraction of acrylamide in French
fries using a single solid-phase sorbent. Am. Lab. News. 2003,
10-14.
(26) Martins, S. I. F. S.; Marcelis, A. T. M.; Van Boekel, M. A. J. S.
Kinetic modelling of Amadori N-(1-deoxy-D-fructos-1-yl)glycine
degradation pathways. Part IsReaction mechanism. Carbohydr.
Res. 2003, 338, 1651-1663.
(27) Davidek, T.; Clety, N.; Aubin, S.; Blank, I. Degradation of the
Amadori compound N-(1-deoxy-D-fructos-1-yl)glycine in aque-
ous model systems. J. Agric. Food Chem. 2002, 50, 5472-
5479.
(28) Davidek, T.; Clety, N.; Devaud, S.; Robert, F.; Blank, I.
Simultaneous quantitative analysis of Maillard reaction precursors
and products by High-Performance Anion Exchange Chroma-
tography. J. Agric. Food Chem. 2003, 51, 7259-7265.
(29) Husek, P. Method for preparing sample for amino acid analysis
and kit for analyzing the same. Phenomenex Inc., EP 1,033,-
576, March 1999.
(30) Leong, L. P. Modelling of the Maillard reaction involving more
than one amino acid. Ph.D. Dissertation, University of Leeds,
Leeds, U.K., 1999.
(31) Leong, L. P.; Wedzicha, B. L. A critical appraisal of the kinetic
model for the Maillard browning of glucose with glycine. Food
Chem. 2000, 68, 21-28.
(12) Wenzl, T.; Beatriz de la Calle, M.; Anklam, E. Analytical
methods for the determination of acrylamide in food products:
a review. Food Addit. Contam. 2003, 20, 885-902.
(13) Svensson, K.; Abramsson, L.; Becker, W.; Glynn, A.; Hellena¨s,
K.-E.; Lind, Y.; Rose´n, J. Dietary intake of acrylamide in
Sweden. Food Chem. Toxicol. 2003, 41, 1581-1586.
(14) Mucci, L. A.; Dickman, P. W.; Steineck, G.; Adami, H.-O.;
Augustsson, K. Dietary acrylamide and cancer of the large bowel,
kidney, and bladder: Absence of an association in a population-
based study in Sweden. Br. J. Cancer 2003, 88, 84-89.
(15) Jung, M. Y.; Choi, D. S.; Ju, J. W. A novel technique for
limitation of acrylamide formation in fried and baked corn chips
and in French fries. J. Food Sci. 2003, 68, 1287-1290.
(16) Grob, K.; Biedermann, M.; Biedermann-Brem, S.; Noti, A.;
Imhof, D.; Amrein, T.; Pfefferle, A.; Bazzocco, D. French fries
with less than 100 µg/kg acrylamide. A collaboration between
cooks and analysts. Eur. Food Res. Technol. 2003, 217, 185-
194.
(32) Brands, C. M. J.; Wedzicha, B. L.; Van Boekel, M. A. J. S.
Quantification of melanoidin concentration in sugar-casein
systems. J. Agric. Food Chem. 2002, 50, 1178-1183.
(33) Stadler, R. H.; Robert, F.; Riediker, S.; Varga, N.; Davidek, T.;
Devaud, S.; Goldmann, T.; Hau, J.; Blank, I. In-depth mecha-
nistic study on the formation of acrylamide and other vinylogous
compounds by the Maillard reaction. J. Agric. Food Chem. 2004,
52, 5550-5558.
(34) http://www.athenavisual.com
(35) Brands, C. M. J.; Van Boekel, M. A. J. S. Reactions of
monosaccharides during heating of sugar-casein systems: build-
ing of a reaction network model. J. Agric. Food Chem. 2001,
49, 4667-4675.

Kinetic Modeling of Acrylamide Formation
J. Agric. Food Chem., Vol. 53, No. 15, 2005 6139
(36) Robert, F.; Vuataz, G.; Pollien, P.; Saucy, F.; Alonso, M.-I.;
Bauwens, I.; Blank, I. Acrylamide formation from asparagine
under low-moisture Maillard reaction conditions. 1. Physical and
chemical aspects in crystalline model systems. J. Agric. Food
Chem. 2004, 52, 6837-6842.
(37) Surdyk, N.; Rose´n, J.; Andersson, R.; Åman, P. Effects of
asparagine, fructose, and baking conditions on acrylamide content
in yeast-leavened wheat bread. J. Agric. Food Chem. 2004, 52,
2047-2051.
(38) Stadler, R. H.; Verzegnassi, L.; Varga, N.; Grigorov, M.; Studer,
A.; Riediker, S.; Schilter, B. Formation of vinylogous compounds
in model Maillard reaction systems. Chem. Res. Toxicol. 2003,
16, 1242-1250.
(39) Van Boekel, M. A. J. S. Statistical aspects of kinetic modeling
for food science problems. J. Food Sci. 1996, 61, 477-485, 489.
(40) Burham, K. P.; Anderson, D. R. Model selection and inference.
A practical information and theoretic approach; Springer-
Verlag: New York, 1998.
(41) Stewart, W. E.; Shon, Y.; Box, G. E. P. Discrimination and
goodness of fit of multiresponse mechanistic models. AlChE J.
1998, 44, 1404-1412.
Received for review March 7, 2005. Revised manuscript received May
20, 2005. Accepted May 21, 2005. This work was carried out partly
with support from a joint program of the Ministry of Economic Affairs,
the Ministry of Education, Culture and Science, and the Ministry of
Housing, Spatial and Environment of The Netherlands (EETK20038)
and with support from the European Commission, Priority 5 on Food
Quality and Safety (Contract FOOD-CT-2003-506820 Specific Targeted
Project), Heat-generated food toxicantssidentification, characterization
and risk minimization. This publication reflects the author’s views and
not necessarily those of the Dutch government or the EC. The
information in this document 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.
JF050504M