foods

Article
Occurrence of Furosine and Hydroxymethylfurfural
in Breakfast Cereals. Evolution of the Spanish Market
from 2006 to 2018
Marta Mesías, Laura Sáez-Escudero, Francisco J. Morales and Cristina Delgado-Andrade *
Institute of Food Science, Technology and Nutrition, ICTAN-CSIC, José Antonio Novais 10, 28040 Madrid, Spain;
[email protected] (M.M.); [email protected] (L.S.-E.); [email protected] (F.J.M.)
* Correspondence: [email protected]; Tel.: +34-91-549-2300




Received: 22 April 2019; Accepted: 8 May 2019; Published: 10 May 2019


Abstract: The demand for healthier products has led the breakfast cereal sector to develop new
formulations to improve the nutritional profile of breakfast cereals; however, the increase in chemical
risks should also be evaluated. Amadori compounds and 5-hydroxymethylfurfural (HMF) are
Maillard reaction products applied as heat damage indices in breakfast cereals. Furosine (a synthetic
amino acid formed by acid hydrolysis of Amadori compounds) is linked to the loss of protein quality,
while HMF has exhibited toxicological effects in cells and animals. Furosine and HMF content was
evaluated in Spanish breakfast cereals whereas the effect of protein, fibre, and sugar content, the type
of grain, the presence of honey, and the manufacturing process were discussed, as well as compared
with a previous prospective study. The average furosine and HMF contents were 182 mg/kg and
21.7 mg/kg, respectively. Protein and fibre content were directly related to the furosine content,
whereas sugar level, honey addition, and the manufacturing process affected the content of HMF.
Occurrence of furosine and HMF decreased nearly 40% in a decade (2006–2018). These findings are
relevant in terms of nutritional score, since lysine availability is preserved, but also from a toxicological
point of view, due to the decreased daily exposure to both compounds, which dropped 30%.

Keywords: breakfast cereals; Maillard reaction; amadori compound; furosine; hidroxymethylfurfural;

consumption; exposure

1. Introduction
In recent decades, the consumption of breakfast cereals has increased in both infant and adult
populations. In 2017, the intake was estimated at 1.57 kg/person/year for the Spanish population [1].
Breakfast cereals are generally rich in carbohydrates and poor in fat. Some of them have high bran
content, with high levels of proteins, and appreciable amounts of certain vitamins and minerals [2].
Looking for healthier products and to meet the well-being criteria of the current consumer, innovative
formulations of breakfast cereals are today drowning the market. Thus, traditional cereals are commonly
replaced by new ones, pseudo-cereals, or mixed with different ingredients to offer new products,
in some cases gluten-free alternatives intended for people with celiac disease. New formulations
including kamut, quinoa, spelt, or teff allow consumers to choose those that best suit their lifestyle [3].
Whatever the composition, the main manufacturing process undergone by these products is
extrusion, which adequately modifies the texture of the final products [4]. In addition, drying and
toasting are also conducted during the elaboration of breakfast cereals. During these steps, starch and
non-reducing sugars, such as sucrose, may be hydrolysed to form reducing sugars, which promotes
non-enzymatic browning, including both Maillard reaction and caramelisation reactions. Polymerisation
reactions can also occur, forming new polymers through non-enzymatic transglycosylation reactions
known as melanoidinas, which are associated with several health benefits [5,6].

Foods 2019, 8, 158; doi:10.3390/foods8050158 www.mdpi.com/journal/foods

Foods 2019, 8, 158 2 of 13

The development of Maillard and caramelisation reactions is related to the aroma, flavour,
and colour of processed foods and depends on the type of substrate, temperature, water activity
(Aw ), and pH, among other factors [7]. The Maillard reaction includes a set of chain reactions that
involves reducing sugar and protein or amino acids, and is usually developed at low Aw , pH 4–7,
and temperatures over 50 ◦ C or even at room temperature, although with a low reaction rate [8].
On the other hand, caramelisation is the direct degradation of the sugars at temperatures higher than
120 ◦ C, lower Aw, and pH 3–9 [7], where the reducing sugars are submitted directly to 1–2 enolisation,
dehydration, and cycling reactions. The extrusion (with intermediate Aw and temperatures over
80–95 ◦ C) and the drying-toasting steps (with low Aw and temperatures higher than 150 ◦ C) encourage
both the Maillard reaction and caramelisation.
Furosine is an amino acid generated during acid hydrolysis of the Amadori compound, which is
formed in the early stage of the Maillard reaction once the free amino group of amino acids, peptides,
or proteins reacts with a reducing sugar to form the Schiff base [9]. Furosine has been considered as a
useful indicator of the degree of thermal damage during the initial steps of the Maillard reaction in
cereal products and is an indirect measure of the available lysine in the food [10]. Since lysine is usually
the limiting amino acid in cereal products, furosine levels can help to set the adequate manufacturing
conditions to ensure the highest nutritional value for products [11].
HMF (5-hydroxymethylfurfural) is formed as an intermediate product of the Maillard reaction
and, moreover, it is generated by the caramelisation of sugars at high temperature [12]. HMF has
been analysed in breakfast cereals and used as an index to control browning reactions and thermal
damage [13,14]. Breakfast cereals commonly include in their formulation ingredients such as caramel
and/or honey, with variable amounts of HMF, contributing to those in situ formed during cereal
manufacturing. Based on animal studies, this compound is suspected to have potential genotoxic and
mutagenic effects through its metabolism product sulphoxymethylfurfural [15]. Therefore, HMF is
considered a chemical process contaminant.
Taking into account the improvements introduced in the industrial processing of breakfast cereals,
there is a growing interest in the evaluation of those changes in terms of toxicological and nutritional
effects. The purpose of this study was to assess the current furosine and HMF levels in breakfast
cereals marketed in Spain. The effects of the type of cereal, protein, dietary fibre, and sugar levels as
well as the presence of honey and the manufacturing process were evaluated. Results were compared
with data previously reported by our research group in 2006, in order to discuss the evolution of the
Spanish breakfast cereals over the last decade.

2. Material and Methods

2.1. Reagents and Chemicals

The HMF standard was purchased from Sigma (St. Louis, MO, USA), and furosine was obtained
from Neosystem Laboratories (Strasbourg, France). Acetonitrile and formic acid were purchased from
Merck (Darmstadt, Germany). The Milli-Q water used was produced using an Elix3 Millipore water
purification system coupled to a Milli-Q module (model Advantage10) (Millipore, Molsheim, France).
All other chemicals, solvents, and reagents were of analytical grade.

2.2. Samples
Sixty commercial packaged breakfast cereal products from more than 20 producers were purchased
from different Spanish supermarkets. Most commercial samples and manufacturers included in the
sampling were brands widely distributed throughout Europe and America. Breakfast cereals containing
dried fruits, nuts, or chocolate were excluded from this study to avoid bias during data interpretation.
Average nutritional composition was calculated based on the information declared on the package
(Table 1). Samples were classified according to several variables: Cereal (the predominant cereal or a
mixture of them), protein content (below or above 7.5% w/w), type of grain (refined or wholegrain),
Foods 2019, 8, 158 3 of 13

fibre content (below or above 5% w/w), sugar content (below or above 20% w/w), the presence of
honey in the recipe (yes or no), and the type of processing (flaked or puffed). The target consumer
(children or general population) was also considered. Samples were mixed and thinly grinded, placed
in polyethylene containers, sealed under vacuum, and stored at 4 ◦ C until analyses.

Table 1. Average nutritional composition of the whole dataset of breakfast cereals as provided by
the manufacturer and grouped according to the predominant cereal. Data are expressed per 100 g
of sample.

Energy Fat Protein Carbohydrates Sugars Fibre Honey

Sample
(Kcal) (g) (g) (g) (g) (g) (g)
Wheat-based 364 ± 34 3.1 ± 1.5 11.0 ± 4.1 66.0 ± 17.4 24.6 ± 15.1 15.0 ± 11.6 0.76 ± 1.34
Corn-based 379 ± 6 1.0 ± 0.4 7.0 ± 1.4 83.7 ± 3.0 16.2 ± 13.4 3.6 ± 1.1 0.69 ± 1.31
Oat-based 383 ± 38 8.2 ± 4.5 12.0 ± 2.1 62.2 ± 7.7 7.2 ± 8.9 11.1 ± 3.9 0.03 ± 0.07
Rice-based 375 ± 5 1.4 ± 0.8 8.3 ± 2.4 81.0 ± 3.6 14.2 ± 8.8 3.2 ± 1.7 -
Spelt-based 391 ± 43 6.9 ± 7.8 11.8 ± 2.1 67.4 ± 5.0 10.1 ± 8.4 6.6 ± 1.2 -
Barley-based 334 2.1 10.6 63.3 2.0 9.8 -
Rye-based 323 1.7 9.5 60.7 0.9 13.2 -
Kamut-based 365 2.0 16.0 70.0 0.9 n.a. -
Teff-based 297 1.8 11.7 55.3 n.a. 26.2 -
Quinoa-based 357 0.9 5.0 80.3 2.8 3.7 -
Cereal-mixture 368 ± 18 2.7 ± 1.0 8.1 ± 1.7 74.6 ± 7.8 16.4 ± 12.1 6.5 ± 3.3 1.24 ± 1.90
Mean value 370 ± 28 3.2 ± 3.3 9.5 ± 3.3 72.3 ± 13.0 15.8 ± 13.5 9.0 ± 8.2 0.70 ± 1.40
Values are mean ± standard deviation (SD). n.a. indicates not available data. Hyphens indicate absence of honey in
the formulation.

2.3. Ion-Pairing HPLC Determination of Furosine

The determination was performed following the traditional method described by Delgado-
Andrade et al. [16] with some modifications. Briefly, 30 mg of the sample were hydrolysed with
4 mL of 7.95 M HCl at 110 ◦ C for 23 h in a Pyrex screwcap vial with polytetrafluoroethylene -faced
septa. Hydrolysis tubes were sealed under nitrogen. The hydrolysates were aerated and cooled at
room temperature and subsequently centrifuged at 14,000× g for 10 min. A 0.5 mL portion of the
supernatant was applied to a Sep-pak Plus C18 cartridge (Waters Millipore, WAT020515) prewetted
with 5 mL of methanol and 10 mL of deionized water and was then eluted with 3 mL of 3 M HCl.
The sample was dried in a speed-vac for 2 h 30 min at 65 ◦ C and dissolved in 1 mL of solution 0.2% of
formic acid. Mobile phase was prepared with 5 mM sodium heptane sulphonate including 20% of
acetonitrile and 0.2% of formic acid. An Extrasyl-ODS2 analytical column (25 × 0.40 cm, 5-µm particle
size, Tecknokroma, Barcelona, Spain) was used at 35 ◦ C. The elution was isocratic, and flow rate was
1.0 mL/min. The injection volume was 20 µL and detection at 280 nm. Furosine was quantified by the
external standard method. Calibration curves were built from a stock solution (1.2 mg/mL of furosine)
in the ranges 0.1–5.0 mg/L and in 0.2% of formic acid. Analysis was conducted with a Shimadzu HPLC
system (Kyoto, Japan) equipped with an LC-20AD pump, an SIL-10ADvp autosampler, a CTO-10ASVP
oven, and an SPD-M20A diode array detector. The limit of quantification (LOQ) was set at 5 mg/kg.
Analyses were done in duplicate, and results were expressed as mg/kg sample.

2.4. Determination of HMF

Aqueous and clarified extracts of samples were used for HMF determination following the HPLC
method described by Mesías et al. [17]. Analysis was conducted with a Shimadzu HPLC system,
as previously described. The chromatographic separation was carried out on a Mediterranean Sea
ODS-2 (250 × 4.0 mm, 5 µm, Tecknokroma, Barcelona, Spain). The mobile phase was a mixture of
acetonitrile in water (5 mL/100 mL) at a flow rate of 1 mL/min under isocratic conditions. The total
running time was 20 min, the ultraviolet detector was set at 280 nm, and 20 µL of the extract was
Foods 2019, 8, 158 4 of 13

injected. HMF was quantified using the external standard. The LOQ was set at 0.3 mg/kg. Analyses
were done in duplicate, and results were expressed as mg/kg sample.

2.5. Food Consumption Data and Exposure

Dietary exposure to Amadori products and HMF from the whole breakfast cereal category as
well as grouped according to different types of cereals was estimated combining the data of total per
capita consumption of breakfast cereals (1.57 kg/person/year) established by the Spanish Ministry of
Agriculture, Food and Environment [1] and the Amadori product and HMF content in the samples.
In the case of Amadori products, their presence in the breakfast cereals was previously calculated taking
into account that the rate of transformation into furosine under acid hydrolysis is nearly 36% [18].

2.6. Statistical Analysis

Statistical analyses were performed using SPSS version 23 (SPSS, Chicago, IL, USA). Data were
expressed as mean ± standard deviation (SD). One-way ANOVA followed by a least significant
differences (LSD) test or Student’s t-test, according to needs, were used to identify the overall
significance of differences. All statistical parameters were evaluated at a p < 0.05 significance level.

3. Results and Discussion

3.1. Nutritional Composition of Breakfast Cereals

The overall nutritional composition of breakfast cereals grouped according to the main cereal in
the formulation and based on the information provided by the manufacturer is depicted in Table 1.
This is shown only for descriptive purposes and for a better understanding of the raw matter and its
behaviour during processing. Samples represented the real situation of the breakfast cereal market not
only in Spain but also in other European countries. According to this, barley, rye, kamut, teff, and
quinoa-based cereal groups were only represented by a unique sample. In contrast, as a reflection of
the commercial options, most of the samples were wheat-based products (23%).
Breakfast cereals exhibited close values for energy, with the teff-based sample being the lowest
in caloric intake (297 Kcal/100 g). Within different groups, protein ranged from 5.0 to 16.0 g/100 g,
carbohydrates from 55.3 to 83.7 g/100 g, sugars from 0.9 to 24.6 g/100 g, and fibre from 3.6 to 26.2 g/100
g. The overall fat content was below 8.5%, with the quinoa-based sample being that with a minor
presence. Thirteen of the sixty samples contained honey in their formulation. Based on the differences
observed in their composition and in line with our previous sampling [14,16], products were grouped
according to the predominant cereal, type of grain (refined or wholegrain), protein content (above
or below 7.5%), fibre content (above or below 5%), sugar content (above or below 20%), and the
presence/absence of honey in their formulation. Additionally, the target consumers (children or general
population) and the manufacturing process (flaked or puffed) were considered as interesting factors to
establish comparisons in the dataset. This allows providing recommendations for consumption in a
certain range of ages as well as information about the technology applied.

3.2. Furosine Levels in Breakfast Cereals. Effect of Type of Cereal in the Formulation and Other Factors
Affecting Its Occurrence
The mean furosine content in the studied samples was 182 mg/kg with a median value of 92 mg/kg,
ranging from lower than 5 mg/kg (LOQ) to 1247 mg/kg; only one sample was below the LOQ. Figure 1
shows furosine detected in breakfast cereals grouped according to the predominant cereal. Wheat-based
products depicted the highest levels, followed by those mainly composed of rice, oat, a mixture of
cereals, and corn (300, 268, 211, 142, and 91 mg/kg, respectively). Teff, quinoa, kamut, rye, barley, and
spelt formulations exhibited a low furosine content, although great variability was detected in this last
group (n = 6). The only significant differences were observed between wheat and corn-based products
(p < 0.05). The rising presence of furosine detected in corn, rice, and wheat groups could be at least
Foods 2019, 8, x FOR PEER REVIEW 5 of 13
Foods 2019, 8, 158 5 of 13

this last group (n = 6). The only significant differences were observed between wheat and corn-based
products (p < 0.05).
partially related The rising
to their proteinpresence
content of furosine
(7.0, 8.3, anddetected in corn, rice, (Table
11.0%, respectively) and wheat
1) andgroups could be
the presence of
at least partially related to their protein content (7.0, 8.3,
lysine in that protein (2.6, 4.0, and 3.7%, respectively) [19]. and 11.0%, respectively) (Table 1) and the
presence of lysine in that protein (2.6, 4.0, and 3.7%, respectively) [19].

(A)

(B)
Furosine (A)
Figure 1. Furosine (A) and
and HMF
HMF (5-hydroxymethylfurfural)
(5-hydroxymethylfurfural) (B)(B) content
content in
in breakfast
breakfast cereals
cereals grouped
according to the predominant cereal. Symbols: ∗∗ strong outlier; ◦◦ mild
mild outlier.
outlier.

Current studies
Current studies describing
describing furosine
furosine in in breakfast
breakfast cereals
cereals are
are scarce. Guerra-Hernández et
scarce. Guerra-Hernández et al.
al. [10]
[10]
established furosine values ranging from 143 to 1026 mg/100 g protein for
established furosine values ranging from 143 to 1026 mg/100 g protein for infant cereal foods after infant cereal foods after
different types
different typesof processing,
of processing,with soy-containing
with soy-containing productsproducts
being those with those
being the highest
withconcentrations
the highest
concentrations due to the greater lysine content in this legume. The sampling carried out byetRada-
due to the greater lysine content in this legume. The sampling carried out by Rada-Mendoza al. [4]
in a heterogeneous
Mendoza et al. [4] ingroup of 10 ready-to-eat
a heterogeneous group ofcereals describedcereals
10 ready-to-eat levelsdescribed
between 87 andbetween
levels 1203 mg/100
87 andg
protein. If samples containing dried milk were excluded, the range was 87–172
1203 mg/100 g protein. If samples containing dried milk were excluded, the range was 87–172 mg/100 g mg/100 g protein,
values much
protein, values closer
much tocloser
our mean
to ourdata
meanexpressed by the protein
data expressed content content
by the protein (176 mg/100 g protein),
(176 mg/100 a usual
g protein),
away to present
usual the furosine
way to present level inlevel
the furosine foods. In the In
in foods. present study,study,
the present formulations with dried
formulations milk were
with dried milk
avoided since considerable amounts of furosine have been detected in milk
were avoided since considerable amounts of furosine have been detected in milk and milk products and milk products [20].
Bastos et al. [21] established a lower range for furosine content in breakfast
[20]. Bastos et al. [21] established a lower range for furosine content in breakfast cereals cereals commercialised in
Portugal (50–119inmg/100
commercialised Portugalg protein).
(50–119 mg/100 g protein).
Together with the processing
Together with the processing conditions, conditions, the the formulation
formulation of of the
the breakfast
breakfast cereal
cereal is
is aa key
key factor
factor
influencing furosine
influencing furosine formation
formation since
since it
it definitively
definitively affects
affects the
the concentration
concentration of of precursors.
precursors. As As already
already
mentioned, based on the previous study by our research group, besides
mentioned, based on the previous study by our research group, besides the predominant cereal, the predominant cereal,
samples were grouped according to seven additional factors: Protein content
samples were grouped according to seven additional factors: Protein content (below or above 7.5% (below or above 7.5%
w/w), type
w/w), type of of grain
grain (refined
(refined oror wholegrain),
wholegrain), fibre fibre content
content (below
(below or or above
above 5%5% w/w),
w/w), sugar
sugar content
content
(below or above 20% w/w), honey presence in the recipe (yes or no), target
(below or above 20% w/w), honey presence in the recipe (yes or no), target consumer (children or consumer (children or
general population), and type of processing (flaked or
general population), and type of processing (flaked or puffed) (Table 2).puffed) (Table 2).
Foods 2019, 8, 158 6 of 13

Table 2. Average furosine and HMF content in breakfast cereals grouped according to different factors.

Factor n Furosine HMF

Protein content
<7.5% 20 114 ± 85 a 31.0 ± 37.1 a
>7.5% 40 216 ± 292 b 17.0 ± 20.8 a
Type of grain
Refined 29 105 ± 116 a 19.8 ± 32.1 a
Wholegrain 32 249 ± 307 b 23.3 ± 23.9 a
Fibre content
<5% 26 121 ± 165 a 27.8 ± 34.4 a
>5% 34 229 ± 288 b 17.0 ± 20.9 a
Sugar content
<20% 39 200 ± 267 a 14.2 ± 17.1 a
>20% 21 165 ± 220 a 34.4 ± 37.3 b
Presence of honey
No 47 197 ± 261 a 16.0 ± 18.8 a
Yes 13 128 ± 183 a 42.0 ± 43.3 b
Target consumer
Children 16 95 ± 78 a 34.2 ± 38.3 a
General population 44 214 ± 279 b 17.1 ± 21.7 a
Manufacturing process
Flaked 34 181 ± 222 a 14.6 ± 16.7 a
Puffed 26 184 ± 281 a 30.9 ± 36.1 b
Analyses were performed in duplicate. Data are mean ± standard deviation (SD) (mg/kg). Different letters within
the same factor indicate statistical differences (p < 0.05). HMF: 5-hydroxymethylfurfural.

A protein content above 7.5% doubled the furosine formation in the product (p < 0.05). Thus,
the higher the protein content in the formulation, regardless of the protein source, the higher was the
furosine formed, with a subsequent reduction of protein quality in the product. In fact, furosine and
protein content were significantly correlated (r = 0.5493; p = 0.0192). The use of wholegrain cereals in
the formulation increased the furosine content 2.5-fold, and fibre levels above 5% raised the formation
of the compound in the product as well (p < 0.05). A significantly higher protein content was detected
among samples with fibre content above 5% (11.3 vs. 7.3% of protein, respectively), a fact that could
explain the favoured formation of furosine. In addition, there was a significant difference (p < 0.05)
in furosine content between breakfast cereals intended for the general population (214 mg/kg) and
those for children (95 mg/kg). Again, the protein level in both groups helped to understand the data,
since samples intended for the general population showed a significantly greater content than those
for children (10.6 vs. 6.6%, respectively). It is worth mentioning that our previous prospective study in
2006 also established increased furosine occurrence in relation with the protein level, the amount of
fibre, and the target population [16].
A high sugar content in the marketed breakfast cereals is the consequence of adding this ingredient
to the recipe or to the final product. In general, in the modern breakfast cereal industry, after mixing
the cereal flours, they are submitted to a cooking-extrusion and shaping step, usually within the same
instrument. Then, the puffed or flaked product is transported to a coating device where sugar syrup is
sprayed. At this moment, other ingredients such as cocoa or honey can be applied, or even vitamins
and minerals if a fortification strategy is pursued to optimize the nutritional value. Afterwards, coated
products are discharged onto a belt dryer to reach targeted levels of moisture for an adequate texture.
Hot coated cereals are cooled to prevent water condensation and are led to the packaging step [22].
The addition of honey, glucose, fructose, and other reducing sugars during the coating step and not
before prevents excessive darkening and the formation of undesirable compounds. This fact helps to
explain why the presence of honey or a sugar content higher than 20% did not influence the levels of
furosine in breakfast cereals, as it was previously described [16].
Foods 2019, 8, 158 7 of 13

Based on our former study, an effect of the manufacturing process (puffed or flaked products) was
expected. The puffing process involves more extreme conditions than the flaking one since puffed
cereal are always obtained by extrusion. In the classic method of producing flakes, grinded grains
are boiled (80–95 ◦ C), dried (220 ◦ C) to decrease moisture content, laminated, toasted (160–200 ◦ C),
and coated, if appropriate [23]. In the modern extrusion method, cereal flours are directly cooked in
the extruder at 140–180 ◦ C, dried, and shaped thanks to the sudden decompression at the exit of the
extruder. Then, they are toasted at temperatures up to 330 ◦ C. Finally, they can be fortified and coated
as in the traditional process [22]. Differences in moisture, temperatures, and residence times could
account for the highest furosine content in puffed breakfast cereals. However, in the present study,
flaked ready-to-eat cereals exhibited the same furosine content as puffed ones. The average protein
content in those samples was less than in the flaked group (8.4 vs. 10.0%). Thus, although the puffing
process involves more extreme conditions and would lead to higher heat damage, the lower presence
of protein in the puffed group was the limiting factor for the furosine formation.

3.3. HMF Levels in Breakfast Cereals. Effect of Type of Cereal in the Formulation and Other Factors Affecting
Its Occurrence
HMF content ranged from lower than 0.3 mg/kg (LOQ) to 159.6 mg/kg, with a mean value of
21.3 mg/kg and a median value of 12.0 mg/kg. Eight of the sixty samples exhibited levels lower
than the LOQ. This variability probably depends on differences in the composition of the cereal and
the type of processing. García-Villanova et al. [13] found higher levels of HMF in breakfast cereals
(3.67–193.34 mg/kg) whereas similar values have been reported by Teixidó et al. [24] (12.6–46.2 mg/kg)
and Mankowska et al. [25] (below 0.4–85.10 mg/kg). In these studies, rice-based cereals had the lowest
HMF levels, and the highest concentrations corresponded to corn and wheat-based cereals, which is in
agreement with the results of the present study if these predominant cereals are compared (Figure 1).
When all the groups of cereals were considered, maximum HMF levels were detected in breakfast
cereals made up of wheat (44.9 mg/kg), followed by those mainly composed by spelt (26.0 mg/kg) and
corn (22.8 mg/kg). The rest of the samples showed levels lower than 12 mg/kg, with concentrations
below the LOQ for oat- and rye-based products. Significant differences were only found between the
HMF content in breakfast cereals made from wheat and those made from rice, oat, corn, and a mixture
of cereals, while the other combinations did not show statistical differences in the mean values.
HMF is formed as an intermediate product of the Maillard reaction between reducing sugars and
amino acids, but it is also generated by the caramelisation of sugars at a high temperature and in slightly
acidic media, for which the presence of amino groups is not needed [12,26]. Specifically, in breakfast
cereals, composition, pH, and processing conditions promote nonenzymatic browning [4]. Regarding
composition and similarly to furosine, the formulation of breakfast cereals and, in consequence,
the concentration of precursors influences HMF formation. The samples with a protein content above
7.5% presented lower levels of HMF. However, variability according to this classification was very
wide and these differences were not significant (Table 2). A similar trend was observed in samples
with a fibre content above 5%, which also exhibited lower levels of HMF but again without significant
differences, in agreement with previously reported data [14]. These results may be expected since,
as mentioned before, samples with high levels of protein also showed high levels of fibre. In accordance
with these results, the use of wholegrain cereals in the formulation of breakfast cereals did not affect
the formation of HMF, showing similar results to refined samples (23.3 and 19.8 mg/kg, respectively).
These findings are contrary to those observed by Teixidó et al. [24], who reported that breakfast cereals
showing the highest values of HMF were bran flakes (46.2 mg/kg).
Differences were significant when samples were grouped according to sugar content. In this
case, levels of HMF were almost 2.5 times higher in breakfast cereals with sugar levels above 20%,
which demonstrates that sugars are a key factor in the formation of this compound compared with
protein levels [27]. As mentioned before, it is common that honey, glucose, fructose, and other reducing
sugars are added during the coating step and not before and, therefore, these ingredients cannot be
Foods 2019, 8, 158 8 of 13

considered as precursors of HMF [14]. However, it has to be taken into account that some of these
ingredients may contribute to the HMF concentration in breakfast cereals not as precursors but due
to their own content in HMF, as in the case of honey or caramel. The dehydration process that these
products undergo in their processing leads to non-enzymatic browning, mainly by carbohydrate
degradation, and then to HMF formation [28]. In this study, honey presence in analysed breakfast
cereals significantly increased the concentration of HMF (42.0 mg/kg) in comparison to samples without
honey (16.0 mg/kg). These results confirm the information reported in other studies, such as the
one by Rufián-Henares et al. [14] who described average levels for HMF of 43.44 and 34.24 mg/kg,
respectively for breakfast cereals with and without honey. A similar level was found by Teixidó et
al. [24] in honey rings (41.0 mg/kg), whereas concentrations up to 85.10 mg/kg have been displayed
in honey wheat loops [25]. Consequently, although HMF is a traditional indicator of the browning
reaction, it would not be useful in monitoring changes in colour, flavour, or nutritional composition
during the manufacture and storage of breakfast cereals when ingredients such as honey or caramel
are included in the formulation [14]. These findings could also explain the higher levels of HMF
exhibited in breakfast cereals intended for children (34.2 mg/kg) compared with those intended for
the general population (17.1 mg/kg), although without significant differences. Products for children
presented an average content of sugars of 30.8 g/100 g and, among them, eleven samples contained
honey. In contrast, the mean sugar content in cereals for the general population was 10.08/100 g, with
only three of the forty-three samples containing honey. Therefore, the sugar level accounted for the
HMF concentrations observed in products intended for these consumers.
Regarding the manufacturing process, puffed samples doubled the levels of HMF (30.9 mg/kg)
compared with flaked cereals (14.6 mg/kg). These results verify that the extreme conditions applied
during the puffing treatment, in order to obtain the extruded cereal, lead to higher heat damages [22].
This promotes the advance of both Maillard reaction and caramelisation and, consequently, the formation
of HMF [11]. Similar findings have been reported by Teixidó et al. [24], describing mean values of
36.6 mg/kg for HMF in puffed corn and 24.0 mg/kg in corn flakes. This fact confirms the use of HMF
as an indicator to monitor processing conditions in the industry [14]. However, if the predominant
cereal was considered, inside the puffed samples, mean HMF results were < LOQ in rye and barley,
2.3 mg/kg in multi-cereal group, 4.1 mg/kg in teff, 4.7 mg/kg in oat, 8.2 mg/kg in rice, 22.4 mg/kg in
corn, 24.9 mg/kg in wheat, and 28.3 mg/kg in spelt-based cereals. In a similar way, Teixidó et al. [24]
reported values close to 12 mg/kg in puffed rice breakfast cereals, lower than the content found in corn
flakes and puffed corn previously mentioned. In conclusion, the effects of processing must be analysed
together with the formulation composition, since high temperatures may promote HMF development
as long as there are precursors that allow it.

3.4. Evaluation of Amadori Compounds and HMF Exposure from Spanish Commercialised Breakfast Cereals
As described previously, furosine is a non-physiological amino acid coming from the acid
hydrolysis of the Amadori compound. Only 36% of this product is efficiently transformed into furosine
in those conditions [18]; therefore, this transformation rate must be taken into account to assess the
presence of Amadori compounds in foods. Using that information and the data provided by the Spanish
Ministry of Agriculture, Food and Environment (MAPAMA) [1] on breakfast cereal consumption in
Spain (1.57 kg/person/year), the average intake of Amadori compounds through breakfast cereals was
estimated. When different predominant grains were considered, exposure was calculated assuming
that all the breakfast cereals consumed in a year came from the same group (Table 3).
Daily exposure to Amadori compounds reflected the furosine levels detected in different groups
of cereals. Wheat, rice, and oat-based groups represented the highest exposure level (3.59, 3.24, and
2.52 mg/day, respectively), followed by a mixture of cereals and corn breakfast cereals with medium
levels (1.69 and 1.08 mg/day). Consumption of spelt, barley, rye, kamut, teff, and quinoa led to a low
supply of Amadori compounds to the diet, ranging from 0.52 to 0.68 mg/day. Taking into account
the mean furosine content in the global sampling, the average exposure to Amadori compounds was
Foods 2019, 8, 158 9 of 13

2.17 mg/day. This figure represents a fall of 30% compared with the estimation that can be drawn from
the study performed in 2006 (3.0 mg/day) [16], using the Spanish breakfast cereal consumption at that
moment (1.28 kg/person/year) [29].

Table 3. Daily exposure to Amadori compounds and HMF considering the whole dataset for breakfast
cereals and grouped for the predominant cereal.

Predominant Cereal Amadori Compound (mg/day) HMF (mg/day)

Wheat 3.59 ± 4.31 0.19 ± 0.17
Corn 1.08 ± 0.50 0.10 ± 0.08
Oat 2.52 ± 3.26 0.03 ± 0.04
Rice 3.21 ± 3.69 0.03 ± 0.02
Spelt 0.68 ± 0.43 0.11 ± 0.08
Barley 0.60 0.001
Rye 0.52 0.001
Kamut 0.57 0.05
Teff 0.61 0.02
Quinoa 0.60 0.001
Mixture 1.69 ± 1.24 0.04 ± 0.04
Mean value 2.17 ± 2.95 0.09 ± 0.12
HMF: 5-hydroxymethylfurfural.

In the case of HMF, daily exposure also followed the same behaviour described for the HMF
content in breakfast cereals (Table 3). The highest exposure scenario takes place when wheat-based
products are consumed (0.19 mg/day), whereas the medium exposure scenario occurs in the case of spelt
and corn groups (0.11 and 0.10 mg/day, respectively), and very low exposure occurs for the remaining
groups (0.05–0.001 mg/day). From the mean HMF content in the whole study, the average daily
exposure for HMF was calculated as 0.09 mg/day. This again indicates a drop of around 30% from the
estimation deducted in our previous sampling in Spanish marketed breakfast cereals (0.13 mg/day) [14].
Rufián-Henares and de la Cueva [30] established the standard HMF intake in the Spanish diet at 10
mg/day, so that the current contribution of breakfast cereals to the daily HMF exposition would be
very low (0.9%).

3.5. Evolution of the Spanish Breakfast Cereal Market in Terms of Furosine and HMF Occurrence from 2006
to 2018
Technological processes in the food industry have progressed during the last decade.
Manufacturers are adopting newer technologies and equipment to enhance different aspects.
For example, innovative enzyme technologies and bioprocessing coupled with high-pressure processing
technology are being used to improve the overall safety, quality, and nutritional traits of oat-based
foods [31,32]. The reassessment of furosine and HMF occurrence in breakfast cereals is a proper tool to
evaluate the impact of the improvements, since they are good indices to estimate the degree of thermal
damage [10,13]. Figure 2A compares the mean furosine level determined in the sampling carried out
by our research group in 2006 [16] with the reassessed value in 2018. A significant decrease of above
40% in furosine occurrence was detected. Figure 2B depicts data according to the cereal type. To be
consistent with the categories studied in 2006, only wheat, corn, rice, and multi-cereal-based breakfast
cereal have been included in the comparison shown in this graphic. In our previous prospective
study, the amount of furosine measured in corn, rice, and wheat-based products were in the same
order of magnitude as those measured in the present sampling (94 vs. 91, 226 vs. 268, and 339 vs.
300 mg/kg, respectively for 2006 and 2018). However, the current levels of the compound in the cereal
mixture group are noticeably lower than those reported in 2006 (142 vs. 473 mg/kg for 2018 and 2006,
respectively) (p < 0.05). Therefore, the one responsible for the decline in the mean furosine level was
the multi-cereal group. Aside from the improvements implemented by the breakfast cereal industry
Foods 2019, 8, x FOR PEER REVIEW 10 of 13

339 vs. 300 mg/kg, respectively for 2006 and 2018). However, the current levels of the compound in
the cereal mixture group are noticeably lower than those reported in 2006 (142 vs. 473 mg/kg for 2018
Foods2006,
and 2019, 8,respectively)
158 10 of 13
(p < 0.05). Therefore, the one responsible for the decline in the mean furosine
level was the multi-cereal group. Aside from the improvements implemented by the breakfast cereal
industry during processing,
during processing, the presence theofpresence ofvarieties
new grain new grainand varieties and pseudo-cereals
pseudo-cereals (spelt,
(spelt, teff, quinoa, teff,
kamut,
quinoa, kamut, etc.) with little furosine formation potential could contribute to that
etc.) with little furosine formation potential could contribute to that decline (Figure 1). decline (Figure
1).

(A)
700 a

600 a b
2006 2018
500
Furosine (mg/kg)

400

300
a a
200
a a
100

0
Wheat Corn Rice Mixture

(B)
Figure
Figure 2.
2. (A)
(A)Box-and-whisker
Box-and-whiskerplotplotof
ofglobal
globalfurosine
furosinecontent
contentin inbreakfast
breakfastcereals
cerealssampled
sampledin in2006
2006vs.
vs.
2018. Symbols:∗∗ strong
2018. Symbols: outlier; ◦◦mild
strong outlier; mildoutlier
outlier(B)
(B) Comparison
Comparison of of the
the furosine
furosine content
content in
in breakfast
breakfast
cereals grouped
cereals grouped according to the predominant
predominant cereal
cereal in
in 2006
2006 vs.
vs. 2018. Different
Different letters
letters indicate
indicate
significant
significant differences between years (p < 0.05).
< 0.05).

In the
In thesame
sameline
lineto to
that described
that for furosine,
described the average
for furosine, HMF occurrence
the average in the current
HMF occurrence in thesampling
current
decreased decreased
sampling 40% compared 40%with the value
compared established
with the value in established
our previousinprospective study
our previous [14] (Figure
prospective 3A).
study
This time, although the mixture group showed a decrease in its mean HMF content compared
[14] (Figure 3A). This time, although the mixture group showed a decrease in its mean HMF content to 2006
(8 vs. 26 mg/kg;
compared p >(8
to 2006 0.05),
vs. the
26 rice and pcorn-based
mg/kg; > 0.05), thegroups
rice were the major contributors
and corn-based groups were to the
thedecline
major
of the globalto
contributors HMF occurrence
the decline in global
of the breakfast
HMF cereals currently
occurrence marketedcereals
in breakfast in Spain (7 vs. 32
currently mg/kg for
marketed in
rice-based and 23 vs. 43 for corn-based, p < 0.05; Figure 3B).
Spain (7 vs. 32 mg/kg for rice-based and 23 vs. 43 for corn-based, p < 0.05; Figure 3B).
Foods 2019, 8, 158 11 of 13
Foods 2019, 8, x FOR PEER REVIEW 11 of 13

(A)
180
a
160
2006 2018
140 a
HMF (mg/kg)

120

100
a b
80
b
60
a
40
a
20 a
0
Wheat Corn Rice Mixture

(B)
Figure 3. (A)
(A) Box-and-whisker
Box-and-whisker plotplot of
of global
global HMFHMF content
contentinin breakfast
breakfast cereals
cereals sampled
sampledin in 2006
2006 vs.
vs.
2018. Symbols: ∗∗ strong
2018. Symbols: strong outlier;
outlier; ◦◦mild
mildoutlier
outlier (B)
(B) Comparison
Comparison of of the
the furosine
furosine and
and HMF
HMF content
content in
breakfast cereals
cerealsgrouped
groupedaccording
according to the predominant
to the predominant cerealcereal
in 2006
invs. 2018.
2006 vs.Different letters indicate
2018. Different letters
significant
indicate differences
significant differences years (p <
between between 0.05).(p < 0.05).
years

4. Conclusions
4. Conclusions
This work
This work compiled
compiled thethe occurrence
occurrence ofof furosine
furosine and
and HMF
HMF inin commercial
commercial breakfast
breakfast cereals
cereals of
of the
the
current Spanish
current Spanishmarket
markettotomonitor
monitortheir
their evolution
evolution from
from 2006
2006 to 2018
to 2018 in terms
in terms of thermal
of thermal damage.
damage. The
The protein and fibre contents were directly related to the formation of furosine, whereas
protein and fibre contents were directly related to the formation of furosine, whereas sugar level, sugar level,
honey addition,
honey addition, and
and the
the manufacturing
manufacturing process
process were
were key
key factors
factors that
that increased
increased thethe presence
presence ofof HMF
HMF
in this
in this food
food matrix.
matrix. Occurrence
Occurrence of of both
both compounds
compounds significantly
significantly decreased
decreased around
around 40%40% inin the
the new
new
prospective study compared with the one performed in 2006. These are interesting findings
prospective study compared with the one performed in 2006. These are interesting findings in terms in terms of
nutritional
of value,
nutritional since
value, lysine
since lysineavailability
availabilityhas
hasbeen
beenimproved
improvedthrough
throughlower
lower heat
heat damage,
damage, but also
but also
from a toxicological point of view, due to the decreased daily exposure to both
from a toxicological point of view, due to the decreased daily exposure to both compounds which compounds which
dropped 30%
dropped 30% in
in this
this food
food category.
category.

Author Contributions: Conceptualization, M.M. and C.D.-A.; Methodology, L.S.-E.; Software, M.M.; Validation,
Author Contributions:
M.M. and F.J.M.; FormalConceptualization,
Analysis, L.S.-E.;M.M. and C.D.-A.;
Investigation, Methodology,
M.M. and C.D.-A.;L.S.-E.; Software,
Resources, M.M.;
F.J.M.; Validation,
Data Curation,
M.M.
M.M. and C.D.-A.;
F.J.M.; Formal Analysis, L.S.-E.;
Writing-Original Investigation,
Draft Preparation, M.M.Writing-Review
L.S.-E.; and C.D.-A.; Resources, F.J.M.;
and Editing, M.M.Data
andCuration,
C.D.-A.;
Supervision,
M.M. F.J.M.;Writing-Original
and C.D.-A.; Funding Acquisition,
DraftF.J.M.
Preparation, L.S.-E.; Writing-Review and Editing, M.M. and C.D.-A.;
Supervision, F.J.M.; Funding
Funding: Comunidad Acquisition,
de Madrid: Project F.J.M.
S2018/BAA-4393, AVANSECAL-II.
Funding: Comunidad de Madrid: Project S2018/BAA-4393, AVANSECAL-II.
Foods 2019, 8, 158 12 of 13

Acknowledgments: This work was partially supported by the Comunidad of Madrid and European funding
from FSE and FEDER programs (project S2018/BAA-4393, AVANSECAL-II-CM).
Conflicts of Interest: The authors declare that they have no conflict of interest.

References
1. MAPAMA (Ministerio de Agricultura, Alimentación y Medio Ambiente). Informe Del Consumo De
Alimentos En España 2017. Available online: https://www.mapa.gob.es/es/alimentacion/temas/consumo-y-
comercializacion-y-distribucion-alimentaria/informeconsumoalimentacionenespana2017_prefinal_tcm-
30-456186.pdf (accessed on 21 April 2019).
2. Bertrais, S.; Luque, M.L.P.; Preziosi, P.; Fieux, B.; De Flot, M.T.; Galan, P.; Hercberg, S. Contribution of
Ready-to-Eat Cereals to Nutrition Intakes in French Adults and Relations with Corpulence. Ann. Nutr.
Metab. 2000, 44, 249–255. [CrossRef]
3. Boukid, F.; Folloni, S.; Sforza, S.; Vittadini, E.; Prandi, B. Current Trends in Ancient Grains-Based Foodstuffs:
Insights into Nutritional Aspects and Technological Applications. Compr. Rev. Food Sci. Food Saf. 2018, 17,
123–136. [CrossRef]
4. Rada-Mendoza, M.; García-Baños, J.L.; Villamiel, M.; Olano, A. Study on nonenzymatic browning in cookies,
crackers and breakfastcereals by maltulose and furosine determination. J. Cereal Sci. 2004, 39, 167–173.
5. Moreira, A.S.; Nunes, F.M.; Simões, C.; Maciel, E.; Domingues, P.; Domingues, M.R.M.; Coimbra, M.A.
Transglycosylation reactions, a main mechanism of phenolics incorporation in coffee melanoidins: Inhibition
by Maillard reaction. Food Chem. 2017, 227, 422–431. [CrossRef]
6. Moreira, A.S.P.; Nunes, F.M.; Domingues, M.R.; Coimbra, M.A. Coffee melanoidins: structures, mechanisms
of formation and potential health impacts. Food Funct. 2012, 3, 903. [CrossRef]
7. Ramírez-Jiménez, A.; García-Villanova, B.; Guerra-Hernández, E.; Ramírez-Jiménez, A.; García-Villanova, B.;
Guerra-Hernández, E. Effect of toasting time on the browning of sliced bread. J. Sci. Food Agric. 2001, 81,
513–518. [CrossRef]
8. Valdés-Martínez, S.E. Hidratos de carbono. In Química De Los Alimentos; Badui, S., Ed.; Pearson Educación:
Naucalpan de Juárez, México, 2006; pp. 29–118. ISBN 970-26-0670-5.
9. Nursten, H. Recent developments in studies of the maillard reaction. Food Chem. 1981, 6, 263–277. [CrossRef]
10. Guerra-Hernández, E.; Corzo, N.; García-Villanova, B. Maillard Reaction Evaluation by Furosine
Determination during Infant Cereal Processing. J. Cereal Sci. 1999, 29, 171–176. [CrossRef]
11. Friedman, M. Food Browning and Its Prevention: An Overview. J. Agric. Food Chem. 1996, 44, 631–653.
[CrossRef]
12. Kroh, L.W. Caramelisation of food and beverages. Food Chem. 1994, 45, 1570–1573. [CrossRef]
13. Garcia-Villanova, B.; Guerra-Hernandez, E.; Martinez-Gomez, E.; Montilla, J. Liquid chromatography for
the determination of 5-(hydroxymethyl)-2-furaldehyde in breakfast cereals. J. Agric. Food Chem. 1993, 41,
1254–1255. [CrossRef]
14. Rufián-Henares, J.Á.; Delgado-Andrade, C.; Morales, F.J. Analysis of heat-damage indices in breakfast
cereals: Influence of composition. J. Cereal Sci. 2006, 43, 63–69.
15. Høie, A.H.; Svendsen, C.; Brunborg, G.; Glatt, H.; Alexander, J.; Meinl, W.; Husøy, T. Genotoxicity of three
food processing contaminants in transgenic mice expressing human sulfotransferases 1A1 and 1A2 as
assessed by the in vivo alkaline single cell gel electrophoresis assay. Environ. Mol. Mutagen. 2015, 56, 709–714.
[CrossRef]
16. Delgado-Andrade, C.; Rufián-Henares, J.Á.; Morales, F.J. Lysine availability is diminished in commercial
fibre-enriched breakfast cereals. Food Chem. 2007, 100, 725–731. [CrossRef]
17. Mesias, M.; Holgado, F.; Marquez-Ruiz, G.; Morales, F. Effect of sodium replacement in cookies on the
formation of process contaminants and lipid oxidation. LWT Food Sci. Technol. 2015, 62, 633–639. [CrossRef]
18. Erbersdobler, H.F. Twenty years of furosine-better knowledge about the biological significance of Maillard
reaction in food and nutrition. In Amino-Carbonyl Reactions in Food and Biological Systems; Fujimaki, M.,
Namiki, M., Kato, M.H., Eds.; Elsevier: Amsterdam, The Netherlands, 1986; pp. 481–491. ISBN
978-4062023672.
19. Kent, N.L. Cereales. In Tecnología De Los Cereales; Acribia Editorial: Zaragoza, España, 1987; pp. 79–83. ISBN
978-8420006086.
Foods 2019, 8, 158 13 of 13

20. Olano, A.; Corzo, N.; Delgado, T.; Troyano, E. Ratio of Lactulose to Furosine as Indicator of Quality of
Commercial Milks. J. Food. Prot. 1994, 57, 737–739.
21. Bastos, D.H.M.; Shibao, J.; Ferreira, E.L.; Bombo, A.J. Produtos da reação de Maillard em alimentos. J. Brazilian
Soc. Food Nutr. 2011, 36, 63–78.
22. Fast, R. Breakfast cereals. In Cereals Processing Technology; Owens, G., Ed.; CRC Woodhead Publishing
Limited: Cambridge, UK, 2001; pp. 158–172. ISBN 1-85573-561-X.
23. Guy, R. Extrusión De Los Alimentos: Tecnología y Aplicaciones; Acribia Editorial: Zaragoza, España, 2002;
pp. 131–160. ISBN 978-84-200-0981-0.
24. Teixidó, E.; Núñez, O.; Santos, F.J.; Galceran, M.T. 5-Hydroxymethylfurfural content in foodstuffs determined
by micellar. Food Chem. 2011, 126, 1902–1908. [CrossRef]
25. Mańkowska, D.; Majak, I.; Bartos, A.; Słowianek, M.; Łacka,˛ A.; Leszczyńska, J. 5-hydroxymethylfurfural
content in selected gluten- and gluten-free cereal food products. Biotechnol. Food Sci. 2017, 81, 11–21.
26. Berg, H.E.; Van Boekel, M.A.J.S. Degradation of lactose during heating of milk. Neth. Milk Dairy J. 1994, 48,
157–175.
27. Morales, F.J. Hydroxymethylfurfural (HMF) and related compounds. In Process-Induced Food Toxicants.
Occurrence, Formation, Mitigation, and Health Risks; Stadler, R., Lineback, D.R., Eds.; John Wiley & Sons, Inc.:
Hoboken, NJ, USA, 2008; pp. 135–174. ISBN 978-0-470-07475-6.
28. Fernández-Artigas, P.; Guerra-Hernandez, E.; Garcia-Villanova, B. Browning Indicators in Model Systems
and Baby Cereals. J. Agric. Food Chem. 1999, 47, 2872–2878. [CrossRef]
29. MAPAMA (Ministerio de Agricultura, Alimentación y Medio Ambiente). Informe del consumo de alimentos
en España 2006. Available online: https://www.mapa.gob.es/app/consumo-en-hogares/resultado1.asp
(accessed on 21 April 2019).
30. Rufián-Henares, J.Á.; De La Cueva, S. Assessment of hydroxymethylfurfural intake in the Spanish diet.
Food Addit. Contam. Part A 2008, 25, 1306–1312. [CrossRef]
31. Hüttner, E.K.; Bello, F.D.; Poutanen, K.; Arendt, E.K. Fundamental evaluation of the impact of high Hydrostatic
Pressure on oat batters. J. Cereal Sci. 2009, 49, 363–370. [CrossRef]
32. Sibakov, J.K.; Kirjoranta, S.J.; Alam, S.A.; Kokkonen, H.; Jurvelin, J.S.; Jouppila, K.; Poutanen, K.S.; Sozer, N.
Effect of oat bran fractions on extrudates made of defatted oats. Food Bioprocess Tech. 2015, 8, 445–458.
[CrossRef]

© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).