Food Chemistry 136 (2013) 576–584
Contents lists available at SciVerse ScienceDirect
Food Chemistry
journal homepage: www.elsevier.com/locate/foodchem
Short-term comparative study of the influence of fried edible oils intake
on the metabolism of essential fatty acids in obese individuals
Carlos Ferreiro-Vera a,b, Feliciano Priego-Capote a,b,⇑, José M. Mata-Granados a,b,c,
María D. Luque de Castro a,b
a
b
c
Department of Analytical Chemistry, Annex C-3, Campus of Rabanales, University of Córdoba, E-14071 Córdoba, Spain
Institute of Biomedical Research Maimónides (IMIBIC), Reina Sofía Hospital, University of Córdoba, E-14071 Córdoba, Spain
Sanyres I+D+i Department, Sanyres Group, E-14012 Córdoba, Spain
a r t i c l e
i n f o
Article history:
Received 16 August 2011
Received in revised form 1 March 2012
Accepted 31 August 2012
Available online 8 September 2012
Keywords:
Deep frying
Dietary intervention
Eicosanoids
Essential fatty acids
Inflammation biomarkers
a b s t r a c t
The effect of breakfast intake of fried oils containing natural antioxidants or a synthetic autooxidation
inhibitor on the metabolism of essential fatty acids focused on obese individuals. Serum levels of eicosanoids were compared in individuals before and after intake of different breakfasts. Univariate descriptive
analysis was used to characterise the cohort selected for this study and multivariate analysis to reveal
statistical differences of normalised eicosanoids concentrations (determined by solid-phase extraction
coupled to LC–MS/MS) depending on the edible oil used for breakfast preparation. The results showed
that the intake of breakfast prepared with pure sunflower oil subjected to deep frying causes an effect
over the eicosanoids profile that enables discrimination versus the rest of individuals. The effect was a
significant increase in the concentration of hydroxyoctadecadienoic acid (HODE) metabolites, indicative
markers of the intake of fried oils. The concentration of HODE metabolites was lower when the oil contained either natural antioxidants from olive-oil pomace or a synthetic autooxidation inhibitor as
dimethylsiloxane. The comparison of the effect of fried sunflower oils with fried extra virgin olive oil
shows the benefits associated to the consumption of the latter.
Ó 2012 Elsevier Ltd. All rights reserved.
1. Introduction
Dietary fat plays a major nutritional role as a key source of energy despite the concerns about fat intake in developed countries
(fat provides 9 kcal/g of metabolizable energy versus an average
energy of 4 kcal/g for carbohydrates and proteins). Dietary fat is
also the source of essential fatty acids (EFAs) and thus, it must be
present in the diet. EFAs derive from two families of fatty acids,
namely, n-6 and n-3 unsaturated fatty acids.
Linoleic acid (LA; 18:2 n-6), which accounts for more than 50%
of the fatty acids in many vegetable oils (e.g. sunflower oil), is the
main n-6 fatty acid in the diet. LA accounts for between 85% and
Abbreviations: AA, arachidonic acid; ASO, refined high-oleic sunflower oil
enriched with an olive-pomace extract of phenols; BI, before intake; COX,
cyclooxygenase; CYP450, cytochrome P450 monooxygenase; DSO, refined higholeic sunflower oil enriched with 400 lg/mL of dimethylsiloxane; EFA, essential
fatty acid; HETE, hydroxyeicosatetraenoic acid; HODE, hydroxyoctadecadienoic
acid; LA, linoleic acid; LNA, a-linolenic acid; LOX, lipoxygenase; LT, leukotriene;
MANOVA, multivariate analysis of variance; PC, principal component; PCA, principal component analysis; PG, prostaglandin; SFA, saturated fatty acid; SO, refined
sunflower oil; SRM, selected reaction monitoring; TX, thromboxane; VOO, extravirgin olive-oil.
⇑ Corresponding author at: Department of Analytical Chemistry, Annex C-3,
Campus of Rabanales, University of Córdoba, E-14071 Córdoba, Spain. Tel./fax: +34
957218615.
E-mail address: [email protected] (F. Priego-Capote).
0308-8146/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.foodchem.2012.08.081
90% of the n-6 fatty acids in the diet with the balance coming from
arachidonic acid (AA; 20:4 n-6) and c-linolenic acid (18:3 n-6). By
contrast, a-linolenic acid (LNA; 18:3 n-3) is a member of the n-3
family of fatty acids. Like LA, a-linolenic acid is the main n-3 fatty
acid in the diet, although common vegetable oils contain LNA concentrations below 1%.
Essential fatty acids are important constituents of biological
membranes, which surround cells and subcellular particles
(McDonald & Eskin, 2006). EFAs also serve as precursors of a variety of biologically active compounds referred to collectively as
eicosanoids (e.g. prostaglandins, thromboxanes, and leukotrienes),
which are key regulators of a host of physiological reactions,
including constriction and dilation of blood vessels, contraction
of smooth muscle, platelet aggregation and regulation of immune
and inflammatory functions (Wang & DuBois, 2007). Eicosanoids
are formed through the action of a set of oxygenase-type enzymes
such as cyclooxygenases (COXs), lipoxygenases (LOXs), and cytochrome P450 monooxygenases (CYP450s). Thus, metabolism of
C-20 fatty acids by COX enzymes leads to the formation of prostanoids, including prostaglandins (PGs) and thromboxanes (TXs),
generating three series of compounds depending on the original
fatty acid (Fitzpatrick & Soberman, 2001; Peters-Golden & Brock,
2003; Yang et al., 2006) (Supplementary Fig. 1). Eicosanoids of
the 2- and 3-series are of clinical interest because they derive from
C. Ferreiro-Vera et al. / Food Chemistry 136 (2013) 576–584
two competitive pathways (n-3 and n-6), which could justify their
opposite functions, as suggested by Schmitz and Ecker (2008).
Thus, 2-series eicosanoids produced from arachidonic acid possess
pro-inflammatory, pro-aggregating, vasoconstriction action and
immunosuppressive properties, as recently reported by Wang
and DuBois (2010). On the other hand, eicosanoids of the 3-series
produced from eicosapentaenoic acid have anti-inflammatory,
anti-aggregating, vasodilatory and anti-arythmic actions and
immunomodulating properties (González-Périz & Clària, 2010;
Groeger et al., 2010).
Lipoxygenases convert AA, LA and other PUFAs into bioactive
metabolites such as leukotrienes (LTs), hydroxyeicosatetraenoic
acids (HETEs) and hydroxyoctadecadienoic acids (HODEs). LOXcatalysed products, LTs, HETEs, and HODEs also exert profound
biological effects on inflammation processes being involved in
the development and progression of specific human cancers such
as colorectal or pancreatic cancer (Xian-Zhong, Wei-Gang, & Thomas, 2001).
Deep fat frying is one of the most common processes used
worldwide for preparation of cooked food. Complex patterns of
oxidative and thermolytic reactions take place in fats and oils during heating and deep-fat frying, including polymerisation, hydrolysis, isomerization, and cyclisation (Dobson, Christie, Brechany,
Sebedio, & Le Quere, 1995; White, 1991; William & Dobson,
2000). Secondary oxidation products are mainly oxidised triglyceride monomers, dimers, and polymers that define the thermal oxidised compounds of the polar material fraction (Dobarganes,
Márquez-Ruiz, & Velasco, 2000; Romero, Bastida, & Sánchez-Muniz, 2006). Oxidation modifies the organoleptic properties of oils
and affects their shelf life, mainly owing to formation of oxidation
products of cholesterol and phytosterols. Degradation results in
loss of nutritional value of food as well as changes in its physiological properties (Saguy & Danaa, 2003; Tyagi & Vasishtha, 1996),
which cause rejection from the consumers and losses to the target
industries, as a result.
The presence of antioxidants, naturally existing in (or added to)
oils, exerts beneficial effects by avoiding or delaying oxidation during frying of compounds such as sterols, fatty alcohols, triterpenic
dialcohols and unsaturated fatty acids. As happens with other food
additives, natural antioxidants such as phenol compounds have
demonstrated an antioxidant activity superior to that of synthetic
oxidation inhibitors; therefore, there is an increased trend to replace the latter with natural antioxidants. The enrichment of edible
oils with phenols protects them, for example, against oxidation
that means better oil quality and prevention from the formation
of toxic products such as cholesterol oxides (Dobarganes & Márquez-Ruiz, 2006).
The aim of the present research was to evaluate the nutritional
impact of the intake of four breakfasts prepared with oils subjected
to deep frying on the profile of eicosanoids in human serum. The
target metabolites were selected taking into account the direct
implication of their metabolism in the inflammatory cascade. Vegetable edible oils with natural or added content of antioxidants
were selected for this purpose. Breakfast muffins made with these
oils were ingested by 26 obese volunteers who consumed the different muffins throughout eight weeks. Liquid-chromatography
coupled to mass detection was used to monitor eicosanoids profile
in human serum extracted from all individuals at three different
sampling times.
2. Materials and methods
2.1. Oils and heating procedure
The edible oils were: (1) extra-virgin olive oil (VOO) as such
with a total natural phenols concentration of 400 mg/L, expressed
577
as caffeic acid, 70.5% monounsaturated fatty acids (MUFAs), 11.1%
PUFAs, 18.4% saturated fatty acids (SFAs); (2) refined high-oleic
sunflower oil enriched with an olive–pomace extract of phenols
(ASO) up to 400 mg/L, also expressed as caffeic acid, 76.7% MUFAs,
17.6% PUFAs and 5.8% SFAs; (3) refined high-oleic sunflower oil enriched with 400 mg/L of dimethylsiloxane as a synthetic oxidation
inhibitor (DSO), 71.8% MUFAs, 18.0% PUFAs and 10.2% SFAs; and (4)
pure refined sunflower oil (SO), 34.3% MUFAs, 58.3% PUFAs and
7.3% SFAs. Koipesol (SOS Cuétara S.A., Madrid) provided the oils
for subsequent enrichment in the laboratory. The composition of
phenolic compounds in VOO and in the extract from olive pomace
is listed in Supplementary Table 1 (Girón, Ruiz-Jimenéz, & Luque
de Castro, 2009).
Two litres of the target oil was placed in a stainless-steel deep
fryer without cover (Fagor F-206, Barcelona, Spain). The oil was
heated at 180 ± 5 °C for 5 min 10 times every day for two days (total heating cycles: 20) with 30 min cooling intervals between heating cycles. This protocol simulates a conventional use of oils for
frying.
2.2. Subjects and samples
Twenty six obese individuals with a body mass index between
30 and 40 kg/m2 formed the cohort in this study. All of them gave
their informed consent and underwent a comprehensive medical
history, physical examination and clinical chemistry analysis before enrolment. Participants with evidence of kidney, pancreas,
lung, liver or thyroid disease were excluded. All subjects were
non-diabetics, non-smokers, without clinical manifestations of cardiovascular disease and off treatment. The target cohort was composed by 17 post-menopausal women, age 48–70 years, and 9
men, age 39–70 years. None of the subjects was taking medication
or supplementary vitamins with influential effect on serum
lipidome.
All volunteers received four breakfasts in muffin format prepared with the four different oils (0.45 mL of oil per kilogram of
body weight), previously subjected to the simulated frying process.
The administration of each breakfast was held at randomization
and cross following a Latin square design, which increased the
power of the study. The volunteers ate each breakfast every two
weeks (4 oils, 8 weeks). All steps from blood extraction to analysis
were performed in compliance with the guidelines dictated by the
World Medical Association Declaration of Helsinki (2004), which
were supervised by the ethical review board of Reina Sofia Hospital
(Córdoba, Spain) that approved the experiments.
2.3. Serum extraction
Blood extraction was planned just before breakfast intake and 2
and 4 h after it. Venous blood was collected into evacuated sterile
tubes for whole blood haematology determination VacutainerÒ
(Becton Dickinson, Franklin Lakes, NJ, USA) and centrifuged at
1500g 4 °C for 10 min to isolate the serum fraction (processing
within 2 h after collection), which was placed in a plastic ware tube
and stored at 80 °C until analysis.
2.4. Analytical method
A fast and automatic method for quantitative analysis of PGE1,
PGE2, PGE3, PGD2, PGF2a, 15keto-PGF2a, TXB2, 9-HODE, 13-HODE,
5-HETE, 8-HETE, 11-HETE, 12-HETE and 15-HETE in serum samples
was developed. The approach is based on a hyphenated system
composed by an SPE workstation (Prospekt-2 unit) on-line coupled
to a liquid chromatograph–triple quadrupole-tandem mass-spectrometer with an ESI source (6410 QqQ Agilent, Palo Alto, CA,
USA). Briefly, 100 lL of human serum is injected in the analytical
578
C. Ferreiro-Vera et al. / Food Chemistry 136 (2013) 576–584
system. Then, the SPE–LC–MS/MS analysis is initiated with the following sequence of automatic operations: (1) Solvation of the stationary phase of C18 cartridge with methanol (2 mL). (2)
Conditioning and equilibration with water (2 mL). (3) Sample loading into the cartridge with water (2 mL) as carrier. Under these
conditions, the target compounds are retained in the sorbent and
20% methanol aqueous solution (2 mL) is used as washing solution
to remove interferences. The chromatographic step starts by
switching the left clamp valve and eluting for 30 s the content of
the cartridge into the chromatographic column with the isocratic
mobile phase – 76:22:2:0.02 (v/v) methanol–water–acetonitrile–
acetic acid – pumped by the chromatographic pump at 0.8 mL/
min. The temperature of the column compartment is set at 25 °C.
Mass-spectrometry detection is performed in negative ESI mode
with 4 kV capillary voltage, 315 °C source temperature and 60 psi
pressure nebulizer. Nitrogen as dessolvation gas is flowed at
10 mL/min. The optimum values for parameters involved in the
determination of target metabolites by selected reaction monitoring (SRM) in tandem mass spectrometry are exposed in Supplementary Table 2. The entire analytical process is completed
within 30 min, which enables to synchronise the SPE protocol with
the chromatographic step. MassHunter Quantitative Analysis
Workstation software version B04.00 from Agilent Technologies
was used to integrate peak areas of the chromatographic signals
corresponding to the target compounds.
Due to the relevance of eicosanoids stability on the quality of
the obtained results, a small experimental planning was done to
assess the stability of analytes under experimental conditions.
For this propose, a serum sample pool stored at 80 °C was thawed
at room temperature and spiked with eicosanoids at concentration
(20 ng/mL). Aliquots were analysed every hour up to 8 h. The results obtained for 12-HETE, 9-HODE, and PGE3 are exposed in Supplementary Fig. 2. No statistical variation was observed (95%
confidence limit) for the target eicosanoids in the studied period,
which was set as the maximum time for sequences of analyses.
Therefore, antioxidants spiking was not required.
2.5. Statistical analysis
A total of 312 samples were obtained from the 26 obese volunteers who received the four breakfasts prepared with the four different oils subjected to the simulated frying process. Blood was
extracted before breakfast intake and 2 and 4 h after it.
The data for each target analyte were studied by one-variable
analysis, where standardised skewness and kurtosis were used to
determine whether the samples belong or not to a normal distribution. Data distributions were log-transformed for normalisation to
meet the assumption of the statistical tests. Antropometric data
not were statistically transformed. Descriptive statistics are presented as means ± SD or as percentages. Univariate and multivariate analyses were carried out using the Statgraphics Plus software
version 5.10 from Manugistics (Rockville, MD, USA) and The
Unscrambler v7.6 (Oslo, Norway).
3. Results and discussion
3.1. Characteristics of the cohort selected for the study
There are several events that can turn on inflammatory responses. The most obvious are microbial invasion, injuries and
burns. However, it is important to understand how diet can also
activate the same inflammatory responses. Obesity is a multifactorial condition resulting from improper balances of hormones and
gene expression induced by the diet. It is becoming more evident
that inflammation plays an important role in the metabolic consequences of obesity as well as other chronic degenerative conditions, as recently reported by Sears and Ricordi (2011).
Inflammation is primarily driven by the production of pro-inflammatory eicosanoids derived from AA, the levels of which are entirely controlled by the diet. Therefore, it is expected that
monitoring eicosanoid metabolites, with special emphasis on AA
metabolites, can reveal some evidences about the influence of
the diet. In fact, the intake of fried foods has been linked to obesity
in a selected cohort of adult Spaniards (Guallar-Castillón et al.,
2007).
After preparation of the data set, the first purpose was the statistical characterisation of the cohort selected for this study prior
to the experiment. The anthropometric characteristics of the participants, such as age, weight, height, waist perimeter and body
mass index, depending on the individuals’ gender, are exposed in
Table 1. Mean, standard deviation and concentration range for each
metabolite before the intake of the target breakfast, depending on
the individuals’ gender, are also included in Table 1, which enables
to set concentration ranges for obese subjects. It is worth emphasising that PGE1, 15keto-PGF2a, 5-HETE, 8-HETE, 11-HETE and
15-HETE were below their detection limits in serum samples
Table 1
Anthropometric characteristics and average concentrations of the target analytes before the intervention study in the selected cohort of 17 women and 9 men (expressed as
mean ± standard deviation). Serum concentrations of target metabolites are expressed as ng/mL.
Parameter
Age (years)
Weight (kg)
Height (cm)
Waist perimeter (cm)
Body mass index (kg/m2)
PGE1
PGE2
PGE3
PGD2
PGF2a
15keto-PGF2a
TXB2
9-HODE
13-HODE
5-HETE
8-HETE
11-HETE
12-HETE
15-HETE
Mean ± SD
Range
N.D.
0.13 ± 0.10
12.86 ± 9.95
0.33 ± 0.27
12.50 ± 8.25
N.D.
0.43 ± 0.48
7.05 ± 4.47
5.99 ± 4.57
N.D.
N.D.
N.D.
12.54 ± 6.96
N.D.
39–63
79–125
164–176
104–145
29.4–41.3
–
0.01–0.54
2.53–55.27
0.07–0.87
2.32–55.27
–
2.32–50.78
1.99–23.36
1.94–27.23
–
–
–
0.98–23.44
–
Concentrations below LLOQ were considered non-detected (N.D.).
Mean ± SD
Range
N.D.
0.31 ± 0.43
11.12 ± 9.23
0.61 ± 0.45
12.37 ± 8.29
N.D.
0.75 ± 0.90
7.81 ± 4.23
6.96 ± 4.16
N.D.
N.D.
N.D.
12.35 ± 9.76
N.D.
48–70
69–120
146–164
88–140
30.7–46.9
–
0.01–2.06
0.14–47.83
0.17–1.33
2.19–44.3
–
0.01–5.87
0.69–26.35
1.5–24.73
–
–
–
1.78–33.95
–
C. Ferreiro-Vera et al. / Food Chemistry 136 (2013) 576–584
579
Fig. 1. Scores graphs of principal component analyses of individuals after intake of each prepared breakfast versus control individuals before intake (BI) attending to serum
levels of monitored eicosanoids. Symbol code is as follows: control individuals before breakfast intake (star) and intake of breakfasts prepared with sunflower oil (circle),
high-oleic sunflower oil enriched with antioxidants from olive pomace (square), high-oleic sunflower oil enriched with dimethylsiloxane (triangle), extra-virgin olive-oil
(hexagon).
and, for this reason, were not taken into account for statistical
analysis. The cohort was initially characterised as a normal distribution with standardised skewness and kurtosis coefficients below
±2.0. The t-Student test revealed no differences in serum levels of
the eicosanoids panel for male and female individuals. This result
enabled to continue the multivariate analysis without gender discrimination with a unique data set for development of the statistical models.
The first statistical test was focused on finding univariate correlations between pairs of quantitative variables, anthropometric
variables versus normalised concentration of metabolites prior to
breakfast intake. In particular, the statistical analysis selected in
this research was the univariate Pearson test by estimation of a
99% confidence level taking into account the biological variability.
As a result, PGE2 serum levels showed strong correlations with
weight (R = 0.8386, p-value = 0.0047), as also does TXB2
(R = 0.8286, p-value = 0.0065) with height. These correlations
are in agreement with other studies reported in the literature such
as that carried out by Hetu and Riendeau (2007), who found PGE2
synthase down-regulation in adipose tissue of obese mices. No
similar correlation for TXB2 has previously been described in the
literature.
Despite these correlations, the eicosanoids profile in the selected cohort was examined according to the anthropometric
parameters of the individuals prior to consumption of breakfasts.
Supplementary Fig. 3 illustrates the results of PCA for the different
anthropometric parameters, revealing no differences between
individuals in terms of eicosanoids concentration with these
anthropometric characteristics. Therefore, the studied cohort was
homogeneous in serum levels of the target eicosanoids with regards to anthropometric variables before the study, which is an
essential fact to proceed with statistical analysis. As a result, the
eicosanoids levels in serum after intake of breakfasts could be correlated to the oil used to prepare them.
3.2. Influence of the breakfast intake on serum levels of the eicosanoids
panel
Principal components analysis was carried out by setting the
eicosanoids profile as quantitative parameter and the intervention
breakfast as categorical variable. Samples were labelled for each
breakfast administered and the control group (red circles in Supplementary Fig. 4A), that corresponded to the analysis of serum
sampled prior to breakfasts intake. For this analysis, no distinction
was made between sampling times after breakfast consumption.
Supplementary Fig. 4A illustrates the scores graph by representing
the three principal components with higher contribution to explain
the observed variability (91%). The colour of symbols indicates the
different breakfasts prepared with the four heated vegetable oils:
VOO (orange circles), ASO (dark green), DSO (light green) and SO
(blue). As can be seen in the three-dimensional plot, no clear discrimination can be observed between each individual before and
after breakfast intake. However, control samples (red) can be perfectly differentiated from individuals after breakfast intake based
580
C. Ferreiro-Vera et al. / Food Chemistry 136 (2013) 576–584
Fig. 2. Loadings graphs of principal component analyses of individuals after intake of each prepared breakfast versus control individuals attending to serum levels of
monitored eicosanoids.
on heated SO (blue) along the first principal component. However,
the high variability of the model does not allow visual discrimination among the other individuals, which were dispersed between
both groups. Despite this variability, Supplementary Fig. 4B shows
the loading graph associated to the previous PCA (for PC 1 and 2)
with the influence of monitored eicosanoids on the observed variability (83% total explained variability: PC1 61%, PC2 22%). The
main contribution to the variability explained by PC1, which enables to differentiate control individuals from those after intake
of SO-breakfast, can be associated to both HODE metabolites. Nevertheless, there is also contribution with the opposite effect for PGs
F2a and E3, and 12-HETE, the last two eicosanoids with major influence on PC2.
After this preliminary analysis, independent PCAs were carried
out for each intervention breakfast versus control samples taken
before breakfast intake. In this sense, the purpose was to evaluate
serum levels of monitored eicosanoids in the initial situation and
after each intake, which enables comparison between both physiological states. Fig. 1 shows the scores graphs corresponding to
each pair control/intervention breakfast. Attending to these graphs,
two behaviours can be differentiated between individuals after intake of VOO- and ASO-based breakfasts (Fig. 1A and B) and individuals after intake of DSO- and SO-based breakfasts taking as
reference the control individuals (prior to intake of any breakfast)
(Fig. 1C and D). It can be clearly visualised the high overlapping in
the serum levels of eicosanoids after consumption of VOO- and
ASO-prepared breakfasts. Although it was similar to VOO-based
intervention diet, the overlapping observed in the case of ASO
was slightly lower than for VOO. This result can be interpreted
pointing out that the metabolism of EFAs is not affected by the intake of these two breakfasts. On the contrary, the control individuals can be perfectly discriminated from individuals after intake
of SO- and DSO-prepared breakfasts attending to their PCAs. Additionally, it can be checked that discrimination is higher for individuals after intake of SO-based breakfast. The lower discrimination
observed in individuals after DSO-based breakfast as compared
to pure sunflower oil could be ascribed to the activity of the synthetic auto-oxidation inhibitor, dimethylsiloxane. The non-polar
character of this additive contributes notably to the inhibition of
EFAs oxidation.
In the case of individuals after VOO- and ASO-prepared breakfasts, there is no statistical incidence on the EFAs metabolism. Taking into account that the refined high-oleic sunflower oil employed
for preparation of ASO and DSO was the same, the behaviour of
individuals after ASO- and VOO-based breakfasts has to be attributed to their composition in hydrophilic antioxidants by enrichment with an extract from olive pomace. To explain the
differences observed between individuals after consumption of
breakfasts, it is important to review the hypothesis in the literature
that compares the efficiency of hydrophilic and lipophilic antioxidants against thermal degradation of oils and subsequent formation of oxidising artifacts such as free radicals. The presence of
hydrophilic compounds in edible oils and their high antioxidant
activity can be explained by the so-called ‘‘polar paradox’’ (Porter,
Black, & Gutfinger, 1981), which establishes that ‘‘polar antioxidants are more effective in non-polar lipids whereas non-polar
antioxidants are more active in polar-lipid emulsions’’. This means
that hydrophilic antioxidants protect more effectively against oxidation than lipophilic antioxidants, because the latter remain dissolved in the oil while phenolic compounds are located at the
air–oil interface (Frankel, 1996). With these premises, the formation of oxidation artifacts may be superior in DSO and SO after frying and, therefore, these artifacts directly affect the metabolism of
EFAs. The assertion of this hypothesis would explain the differ-
C. Ferreiro-Vera et al. / Food Chemistry 136 (2013) 576–584
581
Fig. 3. Scores graphs of principal component analyses of individuals after intake of each prepared breakfast versus control individuals attending to serum levels of monitored
eicosanoids. Distinction was made between the two sampling times after breakfast intake. Symbol code is as follows: control individuals before breakfast intake (circle) and
sampling time 2 h after breakfast intake (square), sampling time 4 h after breakfast intake (triangle).
ences in serum levels of individuals after consumption of the four
breakfasts prepared with fried edible oils. Similar results were
found by Quiles et al. (2002) by analysis of liver microsomes from
rats exposed to the intake of fried sunflower and olive oils. The levels of lipid peroxidation in rats after intake of fried sunflower oil
were higher than after intake of olive oil.
Scores graphs plotted in Fig. 1 were complemented with loading
graphs in these PCA tests for individual breakfasts. Fig. 2 presents
the loadings graphs obtained by qualitative data analysis of control
individuals together with individuals after intake of SO- and DSOprepared breakfasts. Similarly to Supplementary Fig. 4B, the influence of eicosanoids can be evaluated from these graphs. Thus, in
the case of intervention breakfasts based on SO y DSO with a clear
discrimination from control individuals, this variability, expressed
mainly along PC1, can be equally ascribed to HODE metabolites
with an opposite trend to other less significant metabolites such
as 12-HETE, PGs F2a and E3. On the contrary, no contribution to
the statistical model can be attributed to TBX2 and PGD2.
Eicosanoids are synthesised by the action of two different enzymes, COX and LOX, and are derived from two different biosynthetic pathways: n-6 and n-3. The two compounds with higher
contribution to explain the variability of the model, HODE metabolites, are synthesised from linoleic acid (C18:2n-6) by activity of
an LOX enzyme. Linoleic acid is the main fatty acid present in sunflower oil as triglyceride, with concentrations normally above 50%
(w/w). However, this concentration falls down in high-oleic sunflower oil to values of 20% (w/w) or below. Therefore, linoleic acid
passes to a second place in high-oleic sunflower oil being its position occupied by oleic acid. With these premises, a second hypoth-
esis to complete the discussion dealing with the role of hydrophilic
antioxidants is the source of EFAs in the diet. Thus, the precursor
for biosynthesis of HODE metabolites, linoleic acid, is more concentrated in the pure sunflower oil used to prepare the SO-based
breakfast than in the high-oleic sunflower oil used to prepare the
DSO-based breakfast. In summary, the levels of linoleic acid in
the edible oil used for preparation of breakfast could support the
discrimination between individuals after intake of SO- and DSOprepared breakfasts. The up-regulation in the synthesis of HODE
metabolites seems to be accompanied by a down-regulation of
the activity of LOX and COX enzymes using arachidonic acid as
substrate. This could explain the apparent contrary activity of
12-HETE and PGF2a. The deregulation of the biosynthetic pathways
not only affects to the n-6 route, but also to the n-3 route. In fact,
an opposite effect of PGE3 metabolites versus HODE metabolites
was also found as a consequence of the intake of SO- and DSO-prepared breakfasts. This result could be explained by the inflammatory properties attributed to eicosanoids synthesised in n-6 and
n-3 pathways, pro-inflammatory and anti-inflammatory properties, respectively.
3.3. Influence of sampling time after intake of breakfasts prepared with
fried edible oils
After comparison of individuals subjected to intervention
breakfasts versus control physiological state, the next study was
aimed at evaluating the sampling time, defined as the interval
within breakfast intake and blood extraction. Two sampling times
were set, 2 and 4 h, to assess the initial response as a consequence
582
C. Ferreiro-Vera et al. / Food Chemistry 136 (2013) 576–584
Fig. 4. MANOVA results of the study breakfast type versus serum levels of monitored eicosanoids for significant variables at a 99% confidence level. X-axis: (1) SO-based
breakfast; (2) DSO-based breakfast; (3) ASO-based breakfast; (4) VOO-based breakfast; and (5) before breakfast intake.
of intake. Fig. 3 shows the scores graphs obtained after PCAs of
eicosanoids levels in serum. In this case, serum analysis of individuals prior to breakfast intake is plotted by circles. Squares and triangles are used for serum analysis after 2 and 4 h, respectively, of
breakfast intake. Similarly to the results in Fig. 1, a high overlapping is observed with VOO- and ASO-breakfasts. Attending to the
results obtained by PCAs in the evaluation of the sampling time,
there are not significant differences in the serum levels of eicosanoids 2 and 4 h after the intake of breakfasts. Therefore, the time
scale selected for this experimental planning is not enough to have
an impact on the serum levels of eicosanoids for the two sampling
times selected.
3.4. Multivariate analysis of variance (MANOVA)
General linear multivariate models were used to evaluate the
influence of breakfast intake and sampling time together with
anthropometric variables on the serum concentration of target
metabolites as response variables. Breakfasts were categorised
according to type of oils used to prepare the muffins as VOO-,
ASO-, DSO-, and SO-breakfast. Time for serum extraction was categorised as previous to the intake, 2 or 4 h after it. Analysis of variance with a lack-of-fit test was used to determine whether the
model was adequate to describe the observed data. Statistical significance was defined as a p value less than 0.05. A multivariate
model variance analysis (MANOVA) was used to evaluate the
parameters that may have an effect on the eicosanoids concentration. Fisher’s least significant difference procedure was used to discriminate among the means of eicosanoids concentration after the
different breakfasts in MANOVA studies for 99% of confidence level.
Fig. 4 summarises the MANOVA results obtained for those significant variables correlated with breakfast intake that coincide with
those deregulated eicosanoids (9-HODE, 13-HODE, PGs D2 and E3)
according to PCA loadings graphs excepting for PGF2a. As can be
seen, serum concentrations of HODEs were clearly increased in
individuals after SO-breakfast intake versus individuals after intake
of breakfasts with the other oils. Means of logarithm values (log) of
13-HODE serum levels in obese individuals after SO-based breakfast were statistically higher (1.84 ± 0.04) than in the other breakfasts with a 99% confidence level. Additionally, logarithmic values
of 13-HODE serum levels were not significantly different in individuals treated with the DSO diet (0.98 ± 0.04), ASO diet (1.19 ± 0.04)
and VOO diet (1.09 ± 0.04) with a confidence level of 99%. The same
trend was observed for 9-HODE, the logarithmic levels of which
were statistically higher in obese individuals after intake of SO prepared breakfast (1.89 ± 0.04) as compared to the other individuals
with a 99% confidence level. Logarithmic values of 9-HODE serum
levels were significant higher in individuals after intake of ASObased breakfast (1.32 ± 0.04) versus individuals after intake of
DSO-based breakfast (1.17 ± 0.04) and VOO-based breakfast
(1.13 ± 0.04), with a confidence level of 99%. Thus, serum levels of
HODE metabolites in SO-based breakfast reveal that LOX enzyme
using linoleic acid as substrate is up-regulated after intake of this
prepared breakfast. By contrast, no discrimination was observed
after intake of DSO-, ASO- and VOO-based breakfasts, although
the eicosanoid profile versus control individuals was affected.
For the rest of metabolites, differences in serum levels were detected for PGD2 with a confidence level of 99% in individuals after
C. Ferreiro-Vera et al. / Food Chemistry 136 (2013) 576–584
583
confidence level in the MANOVA test. As can be seen, HODEs serum
levels were considerably increased after intake of SO-based breakfast with a 99% confidence level. Additionally, differences in HODEs
levels were detected with 99% confidence level between 2 and 4 h
after breakfast consumption. The opposite situation was found for
PGE3, the serum levels of which were statistically decreased after
intake of the SO-based breakfast with a 99% confidence level. In
this case, no distinction was made between both sampling times
after breakfast intake. This behaviour could be justified by the
pro-inflammatory activity associated to HODE metabolites and
the anti-inflammatory capability frequently linked to series-3
PGs. These results are in agreement with those provided by other
statistical tests in this research and are an evidence of the activation of inflammatory cascade caused by diet.
4. Concluding remarks
The effect of the intake of breakfasts prepared with four edible
oils subjected to a simulated deep frying protocol has been assessed in terms of serum levels of well-known biomarkers of the
inflammatory cascade. Multivariate analysis has led to discrimination between different breakfasts depending on the content of
hydrophilic antioxidants after oils heating. Additionally, the fatty
acids profile has been hypothesised as a critical aspect to detect
deregulation of the metabolism of EFAs. This study supports both
the accepted high-value of extra virgin olive oil as base fat source
from a clinical and nutritional point of view, and also the proposal
of addition of natural antioxidants as a promising alternative to
improve the nutritional properties of other edible oils.
Acknowledgments
The Spanish Ministerio de Ciencia e Innovación (MICINN) and
FEDER Program are thanked for financial support through projects
CTQ2009-07430 and SAF2007-62005. F.P.-C. is also grateful to the
MICINN for a Ramón y Cajal contract (RYC-2009-03921).
Appendix A. Supplementary data
Fig. 5. MANOVA results of the study of sampling time after breakfast intake versus
serum levels of monitored eicosanoids for significant variables at a 99% confidence
level. X-axis: (0) before breakfast intake; (2) 2 h after breakfast intake (4) 4 h after
breakfast intake.
intake of SO-based breakfasts ( 0.02 ± 0.05) than in individuals before intake of any breakfasts as reference group ( 0.65 ± 0.11).
Similarly, PGE3 serum levels also reported differences between
individuals after breakfasts intake versus those prior intake. In this
context, multivariate analysis of the influence of the diet on the
serum concentrations of eicosanoids reported a low incidence in
obese individuals after intake of VOO-based breakfast that, of
course, minimally activated the inflammatory cascade. The inflammation process was activated at the highest extent in individuals
who consumed the breakfast prepared with SO. According to the
MANOVA results, ASO and DSO consumption through breakfast
caused a lower inflammatory response than after consumption of
the cake prepared with pure sunflower oil.
MANOVA tests also enabled the assessment of the influence of
sampling time to check if the biological effect on the eicosanoids
profile was temporally modified. Fig. 5 summarises the results obtained for individuals before and after intake of SO-based breakfast, which statistically affects to the eicosanoids profile. Only
three metabolites were found statistically significant with a 99%
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.foodchem.2012.
08.081.
References
Dobarganes, M. C., & Márquez-Ruiz, M. (2006). Formation and analysis of oxidized
monomeric, dimeric and higher oligomeric triglycerides. In M. D. Erickson (Ed.),
Deep frying-Chemistry, nutrition, and practical applications (pp. 87–110). Illinois:
AOCS-PRESS.
Dobarganes, M. C., Márquez-Ruiz, G., & Velasco, J. (2000). Interactions between fat
and food during deep-frying. European Journal of Lipid Science and Technology,
102, 521–528.
Dobson, G., Christie, W. W., Brechany, E. Y., Sebedio, J. L., & Le Quere, J. L. (1995).
Silver ion chromatography and gas chromatography–mass spectrometry in the
structural analysis of cyclic dienoic acids formed in frying oils. Chemistry and
Physics of Lipids, 75, 171–182.
Fitzpatrick, F. A., & Soberman, R. (2001). Regulated formation of eicosanoids. Journal
of Clinical Investigation, 107, 1347–1351.
Frankel, E. N. (1996). Antioxidants in lipid foods and their impact on food quality.
Food Chemistry, 57, 51–55.
Girón, M. V., Ruiz-Jimenéz, J., & Luque de Castro, M. D. (2009). Dependence of fattyacid composition of edible oils on their enrichment in olive phenols. Journal of
Agricultural and Food Chemistry, 57, 2797–2802.
González-Périz, A., & Clària, J. (2010). Resolution of adipose tissue inflammation.
Scientific Word Journal, 10, 832–856.
Groeger, A. L., Cipollina, C., Cole, M. P., Woodcock, S. R., Bonacci, G., Rudolph, T. K.,
et al. (2010). Cyclooxygenase-2 generates anti-inflammatory mediators from
omega-3 fatty acids. Nature Chemical Biology, 6, 433–441.
Guallar-Castillón, P., Rodríguez-Artalejo, F., Schmid-Fornés, N., Banegas, J. R.,
Amiano Etxezarreta, P., Ardanaz, E., et al. (2007). Intake of fried foods is
associated with obesity in the cohort of Spanish adults from the European
584
C. Ferreiro-Vera et al. / Food Chemistry 136 (2013) 576–584
prospective investigation into cancer and nutrition. American Journal of Clinical
Nutrition, 86, 198–205.
Hetu, P. O., & Riendeau, D. (2007). Down-regulation of microsomal prostaglandin E2
synthase-1 in adipose tissue by high-fat feeding. Obesity, 15, 60–68.
McDonald, B. E., & Eskin, M. N. A. (2006). Role of fat in the diet. In M. D. Erickson
(Ed.), Deep frying-Chemistry, nutrition, and practical applications (pp. 167–171).
Illinois: AOCS-PRESS.
Peters-Golden, M., & Brock, T. G. (2003). 5-Lipoxygenase and FLAP. Prostaglandins
Leukotrienes and Essential Fatty Acids, 69, 99–109.
Porter, W. L., Black, E. D., & Gutfinger, T. (1981). Polyphenols in olive oils. Journal of
the American Oil Chemists’ Society, 58, 966–968.
Quiles, J. R., Huertas, J. R., Battino, M., Ramírez-Tortosa, M. C., Cassinello, M., Mataix,
J., et al. (2002). The intake of fried virgin olive or sunflower oils differentially
induces oxidative stress in rat liver microsomes. British Journal of Nutrition, 88,
57–65.
Romero, A., Bastida, S., & Sánchez-Muniz, F. J. (2006). Cyclic fatty acid monomer
formation in domestic frying of frozen foods in sunflower oil and high oleic acid
sunflower oil without oil replenishment. Food and Chemical Toxicology, 44,
1674–1681.
Saguy, I. S., & Danaa, D. (2003). Integrated approach to deep fat frying: Engineering,
nutrition, health and consumer aspects. Journal of Food Engineering, 56,
143–152.
Schmitz, G., & Ecker, J. (2008). The opposing effects of n-3 and n-6 fatty acids.
Progress in Lipid Research, 47, 147–155.
Sears, B. & Ricordi, C. (2011). Anti-inflammatory nutrition as a pharmacological
approach to treat obesity. Journal of Obesity, article ID 431985, 14 pages.
Tyagi, V. K., & Vasishtha, A. K. (1996). Changes in the characteristics and
composition of oils during deep-fat frying. Journal of the American Oil
Chemists’ Society, 73, 499–506.
Wang, D., & DuBois, R. N. (2007). Measurement of eicosanoids in cancer tissues.
Methods in Enzymology, 433, 27–50.
Wang, D., & DuBois, R. N. (2010). Eicosanoids and cancer. Nature Reviews Cancer, 10,
181–193.
White, P. J. (1991). Methods for measuring changes in deep-fat frying oils. Food
Technology, 45, 75–80.
William, W. C., & Dobson, G. (2000). Formation of cyclic fatty acids during the frying
process. European Journal of Lipid Science and Technology, 102, 515–520.
Xian-Zhong, D., Wei-Gang, T., & Thomas, E. A. (2001). Cyclooxygenases and
lipoxygenases as potential targets for treatment of pancreatic cancer.
Pancreatology, 1, 291–299.
Yang, P., Chan, D., Felix, E., Madden, T., Klein, R. D., Shureiqi, I., Chen, X., Dannenberg,
A. J., & Newman, R. A. (2006). Determination of endogenous tissue inflammation
profiles by LC/MS/MS: COX- and LOX-derived bioactive lipids. Prostaglandins
Leukotrienes and Essential Fatty Acids, 75, 385–395.