Journal of Chromatography A, 1214 (2008) 90–99

Contents lists available at ScienceDirect

Journal of Chromatography A

journa l homepage: www.e lsev ier .com/ locate /chroma

Improved liquid chromatography tandem mass spectrometry method for thedetermination of phenolic compounds in virgin olive oil

Manuel Suárez, Alba Macià, Maria-Paz Romero, Maria-José Motilva ∗

Food Technology Department, CeRTA-TPV, Escuela Técnica Superior de Ingeniería Agraria, Universidad de Lleida, Av/Alcalde Rovira Roure 191, 25198 Lleida, Spain

a r t i c l e i n f o

Article history:Received 5 November 2007Received in revised form 13 October 2008Accepted 15 October 2008Available online 6 November 2008

Keywords:Phenolic compoundsHPLCUPLCMS/MSOlive oil

a b s t r a c t

An improved liquid chromatography (LC) tandem mass spectrometry (MS/MS) method has been devel-oped for the determination of phenolic compounds (phenyl alcohols, phenyl acids, secoiridoid derivatives,lignans and flavonoids) in virgin olive oil. The used LC technique was ultra-performance LC with columnspacked with 1.7 �m particles. The obtained results (retention times, linearity, reproducibility, detectionlimits (LODs) and quantification limits (LOQs) for the analysis of 14 phenolic compounds in standard solu-tions were compared with those obtained by high-performance LC (HPLC)–fluorescence and UPLC–diodearray detection (DAD). When the 1.7 �m column was used, the retention times were decreased threetimes with respect to conventional HPLC (5 �m). The reproducibility of these methods, expressed as rel-ative standard deviation (RSD) in terms of concentration ranged from 0.4–5.0%. In general, the LODs andLOQs were lower in UPLC–MS/MS than the other two methodologies for all the analytes, with the excep-tion of vanillic acid and pinoresinol which values of LODs and LOQs by HPLC–fluorescence were similar tothe values obtained by UPLC–MS/MS. Afterwards, the improved UPLC–MS/MS methodology was used todetermine the studied compounds in spiked refined olive oil (ROO) by combining a liquid–liquid extrac-tion (LLE) as a sample pre-treatment technique. The recoveries of the analytes were higher than 70%, withthe exception of pinoresinol and 3,4-DHPEA-EDA, which were 61% and 67%, respectively. The LODs and

the LOQs ranged from 0.44–127.78 �g/kg, and from 1.11–427.78 �g/kg, respectively for all the analytes.The reproducibility of the method was lower than 3.2%. The LLE–UPLC–MS/MS was successfully applied

ound

1

wptenaiosleo(a

Dva

nas(adac(

0d

to analyze phenolic comp

. Introduction

Polyphenols are an important group of natural compounds,hich are produced in the secondary metabolism of manylants in nature. The importance of these compounds resides inheir antioxidant activity (demonstrated in in vivo and in vitroxperiments) [1], their possible effects against degeneration ill-ess, and some pharmaceutical effects, such as anti-carcinogenic,nti-atherogenic and anti-microbial properties [2–4]. The mostmportant phenolic compounds that have been identified in oliveil are phenolic alcohols (such as hydroxytyrosol and tyrosol)ecoiridoid derivates (such as the dialdehydic form of eleno-ic acid linked to tyrosol (p-HPEA-EDA), the aldehydic form of

lenolic acid linked to tyrosol (p-HPEA-EA), the dialdehydic formf elenolic acid linked to hydroxytyrosol (3,4-DHPEA-EDA), 4-acetoxyethyl)-1,2-dihydroxybenzene (3,4-DHPEA-AC), oleuropeinglycone (3,4-DHPEA-EA) and its methylated form (methyl 3,4-

∗ Corresponding author. Tel.: +34 973 702817; fax: +34 973 702596.E-mail address: motilva@tecal.udl.es (M.-J. Motilva).

ubnCguTe

021-9673/$ – see front matter © 2008 Elsevier B.V. All rights reserved.oi:10.1016/j.chroma.2008.10.098

s in a virgin olive oil sample within 18 min.© 2008 Elsevier B.V. All rights reserved.

HPEA-EA)), phenolic acids and derivates (such as vanillic acid andanillin respectively), lignans (pinoresinol and acetoxypinoresinol)nd flavonoids (including luteolin and apigenin) [5].

The qualitative and quantitative determination of these phe-olic compounds in oil samples is very important and severalnalytical methodologies have been reported. In early years, non-pecific analytical methods, such as thin layer chromatographyTLC) [6] and UV spectroscopy (Folin) [7,8], were applied for thenalysis of polyphenols with limited success. Afterwards, these tra-itional methods were replaced by other more specific ones, suchs high performance liquid chromatography (HPLC) [9–17] and gashromatography (GC) [18,19], and later, capillary electrophoresisCE) [20–23] given the need to profile and identify the individ-al phenolic compounds in olive oil samples. The results obtainedy GC are very reliable and interesting, but the use of this tech-ique is less common because a derivatization step is necessary.

arrasco-Pancorbo et al. [21–23] developed different methodolo-ies to determine phenolic compounds in olive oil samples by CEsing both UV and mass spectrometry (MS) as the detector system.he results were very attractive, with short analysis time and highfficiency peak separation, but the downside of this technique, as

matog

iHtbptmtp

brbtinr(m

patpeAltio

mn(t1Atitaetc(utdpp

2

2

gpl(

3Ect

msws

aW

2

2

UalT((u((sotBtflsaat

2

wAe(cma5t1ltg

awwta(34Mps

M. Suárez et al. / J. Chro

s known, is the low concentration sensitivity. On the other hand,PLC methods with UV [12–14], fluorescence [15], MS [16,17] and

andem MS (MS/MS) [9,10] detectors have also been reported toe highly efficient. All of them, MS and MS/MS have the greatestotential for the analysis of phenolic compounds in olive oil, sincehese compounds are found at trace levels in this complex sample

atrix. Additionally, apart of its high sensitivity, this detection sys-em has the capability to both determine the molecular weight androvide structural information.

Recently, ultra-performance liquid chromatography (UPLC) haseen developed as a result of the improvement in the packing mate-ials used for the chromatographic separation. This technique isased in the van Deemter equation which shows that, as the par-icle sizes decreases to less than 2.5 �m, there is a significant gainn efficiency (in this case the particle size is 1.7 �m) and it doesot diminish at increased flow rates [24]. The principle advantageseported for UPLC are the increase in the signal-to-noise ratio (S/N)narrower peaks), a reduction in the analysis time and an enhance-

ent in peak resolution [25,26].As well as the analytical separation technique, olive oil sam-

les have to be pre-treated in order to clean up the sample matrixnd preconcentrate the polyphenols, since these are found atrace levels in the oil. The most common technique to accom-lish this objective has been to perform either a liquid–liquidxtraction (LLE) [7,27,28] or a solid-phase extraction (SPE) [12,29].lthough SPE is an attractive isolation technique in terms of speed,

ow cost and solvent consumption, Hrncirik et al. [30] concludedhat it was problematic because of the selectivity towards thendividual phenolic compounds, particularly the aglycone typenes.

The aim of this study was to develop a rapid and sensitiveethod based on UPLC–MS/MS for the determination of phe-

olic compounds in olive oil. Previously, the obtained resultsretention time and the quality parameters) in standard solu-ions were compared with those obtained for the analysis of4 phenolic compounds by HPLC–fluorescence and UPLC–DAD.fterwards, UPLC–MS/MS with an LLE as a sample pretreatment

echnique was used to determine the studied compounds spikedn refined olive oil (ROO). To our knowledge, for the first timehe phenolic compounds present in the olive oil were analysednd quantified by UPLC–MS/MS. In addition, the quality param-ters (linearity, reproducibility, LOD, LOQ, and matrix effect) ofhe developed method were studied for the analysis of theseompounds in spiked olive oil. Finally, the developed methodLLE–UPLC–MS/MS) was applied to determine the compoundsnder study in a virgin olive oil sample. A tentative quantifica-ion was also performed for the determination of some secoiridoiderivates (3,4-DHPEA-AC, 3,4-DHPEA-EA, methyl 3,4-DHPEA-EA,-HPEA-EA and ligstroside and oleuropein derivates) in this sam-le.

. Experimentation

.1. Chemicals and reagents

Apigenin, apigenin 7-O-glucoside, luteolin, luteolin 7-O-lucoside, oleuropein, hydroxytyrosol, tyrosol, and vanillin wereurchased from Extrasynthese (Genay, France). Caffeic and vanil-

ic acids were purchased from Fluka Co. (Buchs, Switzerland) and+)-pinoresinol was acquired from Arbo Nova (Turku, Finland).

The secoiridoid derivatives (3,4-DHPEA-AC, 3,4-DHPEA-EDA,,4-DHPEA-EA, methyl 3,4-DHPEA-EA, p-HPEA-EDA and p-HPEA-A) and the lignan acetoxypinoresinol were not availableommercially and were isolated from virgin olive by semiprepara-ive HPLC [31].

cwtws

r. A 1214 (2008) 90–99 91

Standard stock solutions of each compound were prepared inethanol. All the solutions were stored in a dark flask at 4 ◦C. A

tandard stock mixture of the phenolic compounds was preparedeekly at a concentration of 50 mg/l. Fig. 1 shows the chemical

tructure of the studied phenolic compounds.Methanol (HPLC grade), acetonitrile (HPLC grade), n-hexane and

cetic acid were all provided by Scharlau Chemie (Barcelona, Spain).ater was of Milli-Q quality (Millipore Corp, Bedford, MA, USA).

.2. Instrumentation

.2.1. HPLCHPLC analyses were performed in a Waters system (Milford,

SA) consisted of an in-line degasser (with high-purity helium),pump (Waters 600 E), a Waters 717 Plus autosampler with a 20 �l

oop injector, a multi lambda fluorescence detector Waters 2475.he column was an Inertsil ODS-3 column (5 �m, 15 cm × 4.6 mm)GL Sciences Inc.) equipped with a Spherisorb S5 ODS-2 precolumn5 �m, 1 cm × 4.6 mm) (Technokroma, Barcelona). The softwaresed to manage the chromatographic separation was EmpowerMilford, USA). The mobile phase was MilliQ water/acetic acid100/0.2, v/v) as eluent A and acetonitrile as eluent B. The elutiontarted at 5% of eluent B for 2 min, then was linearly increased 25%f eluent B in 8 min, further increased to 40% B in 10 min, an addi-ional increased to 50% B in 10 min, and more increased to 100%

in 10 min. Then, it was kept isocratic for 5 min and back to ini-ial conditions for 5 min. The reequilibration time was 5 min. Theow-rate was 1.5 ml/min, the injected volume was 20 �l and all theamples were filtered through 0.45 �m before the chromatographicnalyses. The fluorescence characteristics were 339 nm emissionnd 278 nm excitation. All the analyses were carried out at roomemperature.

.2.2. UPLCThe UPLC system consisted of an AcQuityTM UPLC equipped

ith a binary pump system Waters (Milford, MA, USA) using ancQuity UPLCTM BEH C18 column (1.7 �m, 100 mm × 2.1 mm i.d.)quipped with a VanGuardTM Pre-Column AcQuity UPLCTM BEH C182.1 mm × 5 mm, 1.7 �m) also from Waters. During the analysis, theolumn was kept at 30 ◦C and the flow-rate was 0.4 ml/min. Theobile phase was eluent A, MilliQ water/acetic acid (100/0.2, v/v)

nd eluent B, acetonitrile. The elution started at 5% of eluent B formin, then was linearly increased 40% of eluent B in 20 min, fur-

her increased to 100% of eluent B in 0.1 min and kept isocratic for.9 min. Then, back to initial conditions in 0.1 min, and the reequi-ibration time was 1.9 min. The injection volume was 2.5 �l, and allhe samples were filtered through 0.22 �m before the chromato-raphic analyses.

The UPLC was coupled to a PDA detector AcQuity UPLCTM andTQDTM mass spectrometer (Waters, Milford, MA, USA). The soft-are used was MassLynx 4.1. The wavelengths in the PDA detectorere set at 278 and 339 nm. Ionization was achieved using elec-

rospray (ESI) interface operating in the negative mode [M–H]−

nd the data were collected in the selected reaction monitoringSRM). The ionization source parameters were capillary voltagekV, source temperature 150 ◦C and desolvation gas temperature00 ◦C, with a flow-rate of 800 l/h. Nitrogen (99.99% purity, N2 LC-S nitrogen generator, Claind, Como, Italy) and argon (≥99.99%

urity, Aphagaz, Madrid, Spain) were used as cone and colli-ion gases respectively. The SRM transitions and the individual

one voltage and collision energy for each phenolic compoundere evaluated by infusing 10 mg/l of each compound in order

o obtain the best instrumental conditions. Two SRM transitionsere studied in order to find the most abundant product ions,

electing the most sensitive transition for quantification and a

92 M. Suárez et al. / J. Chromatogr. A 1214 (2008) 90–99

Fig. 1. Chemical structures of the studied phenolic compounds.

M.Suárez

etal./J.Chrom

atogr.A1214

(2008)90–99

93

Table 1Optimized SRM conditions for the analyses of the studied phenolic compounds by UPLC–MS/MS.

Phenolic group Compound Precursor ion (m/z) Quantitation (SRM1) Confirmation (SRM2)

Product ion Cone voltage (V) Collision energy (eV) Product ion Cone voltage (V) Collision energy (eV) Ion ratio

Phenyl alcohols Hydroxytyrosol 153 123 35 10 95 35 25 35Tyrosol 137 106 40 15 119 40 15 1.9

Phenyl acids and derivates Vanillic acid 167 123 30 10 152 30 15 1.0Caffeic acid 179 135 35 15 117 35 20 25Vanillin 151 136 20 10 92 20 15 6.5

Flavonoids Luteolin-7-O-Glucoside 447 285 50 25 256 50 40 23Apigenin-7-O-Glucoside 431 269 85 20 311 85 25 2.5Luteolin 285 133 55 25 151 55 25 1.9Apigenin 269 123 30 10 152 30 15 1.6

Secoiridoid Oleuropein 539 377 35 15 275 35 20 1.0

Secoiridoid derivates 3,4-DHPEA-AC 195 135 30 10 107 30 20 3.03,4-DHPEA-EDA 319 195 40 5 183 40 10 1.7Ligstroside derivate 335 199 40 10 155 40 15 5.0p-HPEA-EA 361 291 30 10 259 30 10 2.1p-HPEA-EDA 303 285 30 5 179 30 5 1.1Ligstroside derivate 393 317 40 15 257 40 15 4.1Methyl 3,4-DHPEA-EA 409 377 30 5 275 30 10 3.5Oleuropein derivate 365 229 35 10 185 35 15 2.83,4-DHPEA-EA 377 275 35 10 307 35 10 1.1

Lignans Pinoresinol 357 151 40 30 136 40 30 1.0Acetoxypinoresinol 415 235 45 15 151 45 15 3.0

Ion ratio: abundance SRM1/abundance SRM2.

9 matog

stt2otT

2

psoot

pswwscthcwaTn−

2

tfali

3

3

p[cfiTwvuo1

etoTita

wso[t1

btimoaievta

ttGalc7w

wosdrjmtpotptmdcztaifitoc

Ufmayps

4 M. Suárez et al. / J. Chro

econd one for confirmation purposes. For the positive identifica-ion of the analytes in the sample, the chromatographic retentionime of the analyte in the sample would not vary more than% compared to that of a standard, and the relative abundancef the two SRM transitions monitored had to be within 15% ofhe ratios obtained for the standards. The results are shown inable 1.

.3. Samples

Refined olive oil (ROO), which was obtained from an industrialrocess, was used as a blank for recovery and quality parameterstudies. To test the applicability of the developed method, virginlive oil samples were analyzed. This oil was obtained from an oliveil mill in Catalonia (Spain) during the harvest season and was fromhe Arbequina variety of olive fruit.

An LLE was used to isolate the phenolic fraction of the oil. The LLErotocol was similar to described by Morelló et al. [31–33] but withome modifications. Briefly, 20 ml of methanol:water (80:20, v/v)ere added to 45 g of virgin olive oil and homogenized for 2 minith a Polytron (Lutau, Switzerland). After that, two phases were

eparated by centrifugation at 637 × g for 10 min and the hydroal-oholic phase was transferred to a balloon. This step was repeatedwice and the extracts were combined in the balloon. Then, theydroalcoholic extracts were rotatory evaporated up to a syrupyonsistency at 31 ◦C and were dissolved in 5 ml of acetonitrile. After-ards, the extract was washed three times with 10 ml of n-hexane

nd the rejected n-hexane was treated with 5 ml of acetonitrile.he acetonitrile solution was finally rotatory evaporated to dry-ess and then re-dissolved in 5 ml of methanol and maintained at40 ◦C before the chromatographic analysis.

.4. Quality parameters

The instrumental quality parameters, such as linearity, calibra-ion curve, reproducibility, LODs and LOQs were determined firstlyor methanol spiked with known concentrations of polyphenols,nd secondly for ROO spiked with known concentrations of pheno-ic compounds and extracted according to the procedure describedn Section 2.3.

. Results and discussion

.1. Analysis of phenolic compounds in standard solutions

The initial conditions for the analysis of the studied com-ounds were those reported in our previous studies by HPLC31–33]. Then, in order to convert the HPLC gradient to UPLConditions, the AcQuity UPLCTM Columns calculator (Waters, Mil-ord, USA) was used. Then, the gradient was slightly modifiedn order to improve the separation of the studied compounds.he flow-rate was 0.4 ml/min. In our previous reports, methanolas used as eluent B, but in the present study this organic sol-

ent was modified by acetonitrile because when methanol wassed at a flow-rate of 0.4 ml/min, in some percentages of thisrganic solvent in the mobile phase, the pressure was higher than5000 psi.

The SRM transitions and the corresponding acquisition param-ters (cone voltage and collision energy) were selected accordingo the results obtained from infusing 10 mg/l of a standard solution

f each compound into the mobile phase in the negative ESI mode.he selected parameters were those which had a better responsen each case and the results are shown in Table 1. As can be seen inhis table, two product ions were studied in SRM: one was used tollow the quantification of the phenolic compound and the other

pttdu

r. A 1214 (2008) 90–99

as used as a confirmation ion. This is necessary when complexamples, such as virgin olive oil, are analyzed. The product ionsbtained were in agreement with those reported in the literature17]. The SRM ratio of each compound was used to confirm the iden-ification of the analyte. All the SRM ratio calculated were within5% of the ratio calculated upon standards.

Related with the secoiridoid derivatives it was possible to obtainy semipreparative HPLC enough quantity for all the compoundso optimize SRM conditions (by the infusion of a solution of eachndividual compound into the mobile phase in the negative ESI

ode). However the quantity of 3,4-DHPEA-EDA and p-HPEA-EDAbtained was only enough to study the parameters of the validationssay. The lower concentration of the other secoiridoid derivativesn the phenolic extract of virgin olive oil did not permit to obtainnough quantity of these compounds to carry out the completealidation procedure. In consequence these compounds were ten-atively quantified in reference of 3,4-DHPEA-EDA and p-HPEA-EDAs standards.

Fig. 2 shows the total ion chromatogram (TIC) obtained forhe analysis of 14 selected phenolic compounds (hydroxytyrosol,yrosol, vanillic acid, caffeic acid, vanillin, luteolin-7-G, apigenin-7-, 3,4-DHPEA-EDA, oleuropein, luteolin, pinoresinol, p-HPEA-EDA,cetoxypinoresinol and apigenin) in the negative mode. The 14 ana-ytes were resolved in within 18 min. The concentration of theseompounds was 5 mg/l, except to luteolin 7-O-glucoside, apigenin-O-glucoside, luteolin and apigenin which was 1 mg/l, tyrosolhich was 2 mg/l and p-HPEA-EDA which was 15 mg/l.

As it can been seen in Fig. 2 3,4-DHPE-EDA and p-HPEA-EDAere broad peaks. This fact could be attributed to the presence

f isomeric forms of these compounds resulting of the hydroly-is of oleuropein and ligstroside, the main phenols of olive fruit,uring the olive oil extraction process. Phenolic molecules areeactive chemical species, vulnerable to hydrolysis, oxidation, con-ugation, polymerization and complexation. During crushing and

alaxation, steps that break cell walls and expose the olive pasteo enzymes, oxygen and mild heat, many chemical and enzymaticrocesses take place that could explain the presence in the virginlive oil of the different secoiridoids and their isomers resulting ofhe hydrolysis of oleuropein and ligstroside. In order to show theurity of 3,4-DHPEA-EDA and p-HPEA-EDA and the other phenolshat were isolated by semipreparative HPLC, an analysis in full-scan

ode was performed and the MS spectrum at the front, at the mid-le and at the tail were compared. As example, Fig. 3 shows the MShromatogram in full-scan mode of 3,4-DHPEA-EDA at the threeones. The figure shows that the MS spectrum were similar at thehree zones, obtaining in all the cases the precursor ion m/z 319nd the products ions 195 and 183 which were the quantitationon and the confirmation ion respectively. Therefore, this fact con-rmed the purity of the peak and demonstrated that the width ofhis peak could be due to the presence of the isomeric forms. Anal-gous behavior was obtained in the analysis of the other phenolicompounds.

The chromatograms obtained with this methodology,PLC–DAD and UPLC–MS/MS, were compared to that obtained

or the analysis of these compounds by HPLC–fluorescence (chro-atograms not shown). When the studied compounds were

nalysed by HPLC–fluorescence, only eight of them (hydrox-tyrosol, tyrosol, vanillic acid, oleuropein, 3,4-DHPEA-EDA,inoresinol, acetoxypinoresinol, and p-HPEA-EDA) were detected,ince the use of this detector is limited to a reduced group of

henolic compounds. As it can be seen in Table 2, the retentionimes in UPLC methodologies were shorter than the retentionimes in HPLC method, as it was expected. In our study, an averageecrease in retention time about three times was observed bysing the 1.7 �m column.

M. Suárez et al. / J. Chromatogr. A 1214 (2008) 90–99 95

F f the 1w ic acid( nol, 51

3

tLoHfU

btwsc(

ilsw05

chbHp

ig. 2. Total ion chromatograms (TIC) in SRM acquisition obtained from the analysis oas: (1) hidroxytyrosol, 5 mg/l; (2) tyrosol, 2 mg/l; (3) vanillic acid, 5 mg/l; (4) caffe

8) 3,4-DHPEA-EDA, 5 mg/l; (9) oleuropein, 5 mg/l; (10) luteolin, 1 mg/l; (11) pinoresimg/l.

.2. Quality parameters

Standards solutions were analyzed by UPLC–MS/MS in ordero determine the linearity range, reproducibility, LODs andOQs. Then, the obtained results were compared with to thosebtained for the analysis of these compounds by UPLC–DAD andPLC–fluorescence. Table 2 show the quality parameters obtained

or the analysis of these 14 compounds by HPLC–fluorescence,PLC–DAD and UPLC–MS/MS.

The linearity range of the analytical procedure was performedy serial dilution of a stock solution of the studied compounds. Allhe calibrations curves (obtained based on the integrated peak area)ere linear over the range of study and were calculated by using

even points at different concentrations. Furthermore, each con-entration was injected three times. The determination coefficientsR2) were higher than 0.996 for all analytes both for HPLC and UPLC.

wis

Fig. 3. MS chromatogram and MS spectrum obtained at the front (A), a

4 phenolic compounds in standard solution. Peak designation and its concentration, 5 mg/l; (5) vanillin, 5 mg/l; (6) luteolin 7-O-G, 1 mg/l; (7) apigenin 7-O-G, 1 mg/l;mg/l; (12) p-HPEA-EDA, 15 mg/l; (13) acetoxypinoresinol, 5 mg/l; and (14) apigenin,

The reproducibility, as the relative standard deviation (RSD%),n terms of concentration, was determined at two concentrationevels, 50 mg/l and the lowest value of the range of study (data nothown) on three different days. As can be seen in Table 2, good RSDsere achieved in the three methodologies, and these ranged from.4 to 4.0% for all the analytes, except to p-HPEA-EDA, which was% in HPLC–fluorescence.

The LODs and LOQs, calculated using the signal-to-noise ratioriterion of 3 and 10, respectively, were between low �g/l andundreds �g/l, respectively for the analysis of these compoundsy UPLC–MS/MS. These values were lower than the obtained byPLC–fluorescence and UPLC–DAD. In contrast, for some com-ounds such as tyrosol, vanillic acid, and pinoresinol, these values

ere similar than the obtained by HPLC–fluorescence. By compar-

ng these results with to those reported in the literature, these wereimilar or lower [15,29].

t the middle (B) and at the tail (C) of the peak 3,4-DHPEA-EDA.

96 M. Suárez et al. / J. Chromatogr. A 1214 (2008) 90–99

Fig. 4. Extracted ion chromatograms of the identified phenolic compounds in a virgin olive oil sample by LLE–UPLC–MS/MS.

M. Suárez et al. / J. Chromatogr. A 1214 (2008) 90–99 97

Table 2Retention time, linearity, reproducibility (RSD%), LODs and LOQs for the analysis of the 14 phenolic compounds by HPLC–fluorescence, UPLC–DAD, and UPLC–MS/MS instandard solutions.

Compound Retention time (min) Linearity (mg/l) RSD% (n = 3), 50 mg/l LOD (mg/l) LOQ (mg/l)

HPLC–fluorescenceHydroxytyrosol 8.2 0.45–50 1.6 0.21 0.69Tyrosol 11.1 0.014–50 2.1 0.009 0.031Vanillic acid 12.7 0.014–50 1.3 0.006 0.021Oleuropein 16.8 0.27–30 1.8a 0.14 0.473,4-DHPEA-EDA 18.9 2.5–50 0.7 1.5 5.0Pinoresinol 20.5 0.006–50 1.3 0.002 0.007Acetoxypinoresinol 20.8 0.2–18 2.2b 0.1 0.4p-HPEA-EDA 22.8 5–50 5.0 2.0 6.5

UPLC–DADHydroxytyrosol 2.5 0.3–50 0.4 0.17 0.60Tyrosol 4.05 0.4–50 0.9 0.22 0.70Vanillic acid 5.28 0.2–50 0.5 0.07 0.20Caffeic acid 5.66 0.2–50 1.4 0.12 0.40Vanillin 7.52 0.1–50 0.7 0.04 0.12Luteolin-7-glucoside 11.39 0.15–50 1.3 0.08 0.30Apigenin-7-glucoside 12.42 0.08–36 0.6c 0.04 0.103,4-DHPEA-EDA 13.25 6–50 0.5 3.00 15.00Oleuropein 13.69 0.5–50 1.5 0.20 0.60Luteolin 15.06 0.09–50 2.6 0.06 0.20Pinoresinol 15.65 0.2–50 0.4 0.1 0.50p-HPEA-EDA 16.02 7–50 0.7 0.7 9.00Acetoxypinoresinol 16.41 3–50 0.5 0.5 5.00Apigenin 17.21 0.08–50 1.4 0.05 0.10

UPLC–MS/MSHydroxytyrosol 2.5 0.06–42 1.8d 0.03 0.08Tyrosol 4.05 0.02–50 1.6 0.01 0.06Vanillic acid 5.28 0.03–50 0.9 0.02 0.06Caffeic acid 5.66 0.03–50 1.1 0.02 0.06Vanillin 7.52 0.03–50 0.4 0.008 0.03Luteolin-7-glucoside 11.39 0.01–50 0.9 0.004 0.01Apigenin-7-glucoside 12.42 0.025–36 0.9c 0.006 0.023,4-DHPEA-EDA 13.25 0.6–10 1.3e 0.4 1.4Oleuropein 13.69 0.04–30 1.2a 0.02 0.06Luteolin 15.06 0.004–50 1.4 0.002 0.007Pinoresinol 15.65 0.03–50 2.1 0.01 0.08p-HPEA-EDA 16.02 0.6–15 3.0f 0.3 1.0Acetoxypinoresinol 16.41 0.3–5 4.0g 0.1 0.8Apigenin 17.21 0.003–5 2.1g 0.001 0.005

a 30 mg/l.b 18 mg/l.c 36 mg/l.d 42 mg/l.

3

aembaetrwahrarf[

lrTdRw1To4

tr

e 10 mg/lf 15 mg/l.g 5 mg/l.

.3. Analysis of oil sample

The developed method, UPLC–MS/MS, was validated for thenalysis of olive oil. In these samples, a pre-treatment is nec-ssary before the analytical separation technique to get rid ofatrix components and to enrich the analytes, usually performed

y LLE. Firstly, the extraction process was investigated with 22nd 45 g of ROO in order to test differences in recoveries. Thesextraction recoveries were estimated using this oil spiked withhe analytes at a concentration of 5 mg/l. No difference betweenecoveries was observed when the studied phenolic compoundsere extracted in 22 and 45 g of ROO and the recoveries were

cceptable. Therefore, we chose 45 g as the optimum amount sinceigher amounts can give lower LODs. Table 3 shows the extraction

ecoveries (R%) and these were above 70%, except with pinoresinolnd 3,4-DHPEA-EDA, whose recoveries were about 61% and 67%,espectively. These values were in agreement with the literatureor the extraction of these compounds both for LLE [30,34] and SPE29,30].

taoar

Then, the method was used to analyze ROO spiked with the ana-ytes at different concentrations in order to determine the linearange, the calibration curves, the RSDs, the LODs and the LOQs.he calibrations curves were linear over the range of study withetermination coefficients higher than 0.990 for all the analytes.eproducibility, expressed as RSD% and in terms of concentration,as obtained by analyzing three replicates of 45 g spiked analytes at

111 �g/kg. The RSDs were lower than 3.2% for all the compounds.he LODs and the LOQs, calculated using a signal-to-noise ratiof 3 and 10 respectively, were from 0.44 to 127, and from 1.11 to28 �g/kg, respectively.

Despite the increasing success of LC coupled to an MS detec-or, matrix effects have limited the ESI applications. Matrix effectsesult from co-eluting matrix components that compete for ioniza-

ion capacity. This competition produces significant errors in theccuracy and precision of the analytical method. Matrix effect isbserved by a decrease or increase of the analyte signal present inmatrix extract with the same analyte present in organic solvent,

espectively [35].

98 M. Suárez et al. / J. Chromatogr. A 1214 (2008) 90–99

Table 3Recovery (R%), linearity, reproducibility (RSD), LOD and LOQ for the analysis of 14phenolic compounds in spiked ROO UPLC–MS/MS.

Compound R (%) Linearity (�g/kg) RSD% (n = 3) (1111 �g/kg) LOD (�g/kg) LOQ (�g/kg)

Hydroxytyrosol 76 67–5555 1.8 3.33 55.52Tyrosol 84 6–5555 1.9 7.78 72.22Vanillic acid 87 2.7–1111 0.3 3.33 35.56Caffeic acid 75 3.3–5555 1.8 1.11 11.11Vanillin 73 2.7–5555 0.8 2.22 16.67Luteolin-7-O-glucoside 84 3.3–2222 2.3 1.11 11.11Apigenin-7-O-glucoside 82 6–555 1.3a 2.22 16.673,4-DHPEA-EDA 67 167–1333 3.0 72.22 244.44Oleuropein 89 2.2–5555 2.7 5.56 55.56Luteolin 104 0.7–1111 2.2 0.44 1.11Pinoresinol 61 6–5555 1.8 1.11 11.11p-HPEA-EDA 71 333–3333 3.1 127.78 427.78A 3.2 16.67 611.11A 1.3 0.67 2.22

iTtton1tbidtoi

3

civ3aobi

ywsecc

(EqDsd3two

f

Table 4Identified phenolic compounds in virgin olive oil and their quantification.

Compound Concentration (mg/kg olive oil)

Hydroxytyrosol 2.5Tyrosol 3.0Vanillic acid 0.9Caffeic acid n.d.Vanillin 0.5Luteolin-7-O-G n.q.Apigenin-7-O-G n.d.3,4-DHPEA-EDA 152Oleuropein n.d.Luteolin 4.1Pinoresinol 2.3p-HPEA-EDA 20Acetoxypinoresinol 0.9Apigenin 1.53,4-DHPEA-ACa 1.63,4-DHPEA-EAa 68Methyl 3,4-DHPEA-EAa 5.0p-HPEA-EAb 42Ligstroside derivatec 25Oleuropein derivatea 1.4

n.q.: not quantified, n.d.: not detected.

fpIDwa

HptoFwbcio

cetoxypinoresinol 80 33–1222pigenin 89 1.1–222

a 555.56 �g/kg.

In our study an evaluation of signal suppression was conductedn order to assess its effect on phenolic compounds quantification.his evaluation was done by comparing the detector response ofhe phenolic compounds spiked in organic solvent (methanol) withhose from ROO extracts, at different concentration levels. It wasbserved either a positive or negative effect, lower than 15% (dataot shown). In order to minimize this effect, the extract was diluted:2 to prevent it but this dilution produced an increase of LODs,hen no dilution of the sample was done because this effect coulde considered small. These results are consistent with other stud-

es previously reported in the literature which concluded that MSetection proceeded by an efficient UPLC separation is less suscep-ible to matrix effects, when this is only produced for the coelutionf two different compounds, and may contribute to a reduction inon suppression [36,37].

.4. Application of the developed method to virgin olive oil

In order to show the applicability of the developed method, aommercial virgin olive oil was analyzed. Fig. 4 shows the extractedon chromatograms of the selected 14 compounds included in thealidation procedure, and the other secoiridoids (3,4-DHPEA-AC,,4-DHPEA-EA, p-HPEA-EA, methyl 3,4-DHPEA-EA), and oleuropeinnd ligstroside derivates present in the phenolic fraction of virginlive oil. As it has been previously explained, the broad of the peakselonging to 3,4-DHPEA-EDA and p-HPEA-EDA could be due to their

someric forms.Table 4 presents the quantitative results obtained for the anal-

sis of virgin olive oil by LLE-UPLC-MS/MS. Analyte concentrationsere quantified by calibration curves for the phenolic compounds

piked in ROO. As it has been commented before, the SRM ratio ofach compound was used to confirm the identity of the phenolicompound. All the SRM ratio calculated were within 15% of the ratioalculated upon standards.

Some of secoiridoid derivates identified in the virgin olive oil3,4-DHPEA-EA, methyl 3,4-DHPEA-EA, 3,4-DHPEA-AC, p-HPEA-A, and oleuropein and ligstroside derivates) were tentativelyuantified. We attempted to quantify these compounds using 3,4-HPEA-EDA and p-HPEA-EDA as standards in the basis of the

imilarity in chemical properties of each compound with the stan-ard chosen for its quantification. In this way 3,4-DHPEA-AC,,4-DHPEA-EA and methyl 3,4-DHPEA-EA were tentatively quan-

ified with the calibration curve of 3,4-DHPEA-EDA; p-HPEA-EAith the calibration curve of p-HPEA-EDA; and the ligstroside and

leuropein derivatives with the calibration curve of oleuropein.The amount of the phenolic compounds in the olive oil varies

rom traces to relatively high levels, as it is known, and there-

4

f

a Quantified with the calibration curve of 3,4-DHPEA-EDA.b Quantified with the calibration curve of p-DHPEA-EDA.c Quantified with the calibration curve of oleuropein.

ore virgin olive oil contained low amounts of phenyl acids andhenyl alcohols and high concentrations of secoiridoid derivates.

n our study, the most abundant phenolic compounds were 3,4-HPEA-EDA, 3,4-DHPEA-EA, p-HPEA-EA and their concentrationsere 152, 68 and 42 mg/kg, respectively, although 3,4-DHPEA-EA

nd p-HPEA-EA were tentatively quantified.In order to quantify 3,4-DHPEA-EDA, luteolin, pinoresinol, p-

PEA-EDA, 3,4-DHPEA-AC, 3,4-DHPEA-EA, methyl 3,4-DHPEA-EA,-HPEA-EA and ligstroside derivate it was necessary to dilutehe sample because the concentrations of these compounds wereutwith the linear range of the corresponding calibration curve.urthermore, caffeic acid, apigenin 7-O-glucoside and oleuropeinere not detected and luteolin 7-O-glucoside was not quantifiedecause its concentration was between its detection and quantifi-ation limits. These results were in agreement with those reportedn the literature for the analysis of these compounds in virgin oliveil from the Arbequina cultivar [29,38].

. Conclusions

This paper develops a rapid, efficient and sensitive methodor the determination of phenolic compounds in virgin olive oil

matog

btotarpbaU

twipdbmnflo

A

ar

R

[

[

[

[[[

[

[[[[

[

[

[

[

[

[[

[[

[[

[

[

[

M. Suárez et al. / J. Chro

y UPLC–MS/MS. The developed methodology was compared inerms of speed, sensitivity and reproducibility with the resultsbtained by HPLC–fluorescence and UPLC–DAD in standard solu-ions. The use of columns packed with 1.7 �m particles allowednalysing the 14 phenolic compounds in within 18 min, whichepresented three-fold reduction in the analysis time in com-arison to conventional HPLC. Generally, the LODs and LOQsy UPLC–MS/MS were lower than the obtained by UPLC–DADnd HPLC–fluorescence, but for some analytes these values inPLC–MS/MS and HPLC–fluorescence were similar.

UPLC–MS/MS in combination with a LLE, as a sample pre-reatment technique, was developed by spiking the ROO sampleith the studied phenolic compounds. The extraction recover-

es ranged between 73–104%, except for 3,4-DHPEA-EDA andinoresinol which were 67% and 61%, respectively. The repro-ucibility was lower than 3.2% and the LODs and LOQs wereetween low �g/kg and hundred �g/kg. Finally, the developedethodology was successfully applied to a broad range of phe-

olic compounds, such as phenyl acids, phenyl alcohols, lignans,avonoids and secoiridoid derivates in a commercial virgin oliveil from the Arbequina cultivar.

cknowledgements

This work was supported by the Spanish Ministry of Educationnd Science (project AGL2005-07881-C02-01/ALI) and the granteceived by Manuel Suarez (BES-2006-14136).

eferences

[1] F. Visioli, C. Galli, Crit. Rev. Food Sci. Nutr. 42 (2002) 209.[2] D. Ryan, M. Antolovich, P. Prenzler, K. Robards, S. Lavee, Sci. Hort. 92 (2002) 147.[3] M. Patumi, R. D’Andria, V. Marsilio, G. Fontanazza, G. Morelli, B. Lanza, Food

Chem. 77 (2002) 27.[4] L. Bravo, Nutr. Rev. 56 (1998) 317.

[5] L.S. Artajo, M.P. Romero, M. Suárez, M.J. Motilva, Eur. Food Res. Technol. 225

(2007) 617.[6] R. Capasso, A. Evidente, F. Scognamiglio, Phytochem. Anal. 3 (1992) 270.[7] G. Montedoro, M. Servili, M. Baldioli, E. Miniati, J. Agric. Food Chem. 40 (1992)

1571.[8] M.J. Tovar, M.P. Romero, J. Girona, M.J. Motilva, J. Sci. Food Agric. 82 (2002) 892.

[[

[[

r. A 1214 (2008) 90–99 99

[9] A. Bianco, F. Buiarelli, G.P. Cartoni, F. Coccioli, R. Jasionowska, P. Margherita, J.Sep. Sci. 26 (2003) 409.

10] A. Bianco, F. Buiarelli, G.P. Cartoni, F. Coccioli, R. Jasionowska, P. Margherita, J.Sep. Sci. 26 (2003) 417.

11] A. Zafra, M.J.B. Juárez, R. Blanc, A. Novalón, J. González, J.L. Vilchez, Talanta 70(2006) 213.

12] R. Mateos, J.L. Espartero, M. Trujillo, J.J. Rios, M. León-Camacho, F. Alcudia, A.Cert, J. Agric. Food Chem. 49 (2001) 2185.

13] M. Tasioula-Margari, O. Okogeri, Food Chem. 74 (2001) 377.14] J.R. Morelló, M.J. Motilva, T. Ramo, M.P. Romero, Food Chem. 81 (2003) 547.15] R. Selvaggini, M. Servili, S. Urbani, S. Esposto, A. Taticchi, G.F. Montedoro, J. Agric.

Food Chem. 54 (2006 2832).16] S.M. Cardoso, S. Guyot, N. Marnet, J.A. Lopes-da-Silva, C.M.G.C. Renard, M.A.

Coimbra, J. Sci. Food Agric. 85 (2005) 21.17] M. Savarese, E. De Marco, R. Sacchi, Food Chem. 105 (2007) 761.18] D. Ryan, K. Robards, Analyst 123 (1998) 31R.19] L. Liberatore, G. Procida, N. D’Alessandro, A. Cichelli, Food Chem. 73 (2001) 119.20] M. Bonoli, A. Bendini, L. Cerretani, G. Lercker, T.G. Toschi, J. Agric. Food Chem.

52 (2004) 7026.21] A. Carrasco-Pancorbo, L. Cerretani, A. Bendini, A. Segura-Carretero, T. Gallina-

Toschi, A. Fernández-Gutiérrez, J. Sep. Sci. 28 (2005) 837.22] A. Carrasco-Pancorbo, D. Arráez-Román, A. Segura-Carretero, A. Fernández-

Gutiérrez, Electrophoresis 27 (2006) 2182.23] A. Carrasco-Pancorbo, C. Neusü�, M. Pelzing, A. Segura-Carretero, A. Fernández-

Gutiérrez, Electrophoresis 28 (2007) 806.24] E. Barceló-Barrachina, E. Moyano, M.T. Galcerán, J.L. Lliberia, B. Bagó, M.A.

Cortes, J. Chromatogr. A 1125 (2006) 195.25] C.C. Leandro, P. Hancock, R.J. Fussell, B.J. Keely, J. Chromatrogr. A 1103 (2006)

94.26] X. Li, Z. Xiong, X. Ying, L. Cui, W. Zhu, F. Li, Anal. Chim. Acta 580 (2006) 170.27] E. Gimeno, A.I. Castellote, R.M. Lamuela-Raventós, M.C. De La Torre, M.C. Lopez-

Sabater, Food Chem. 78 (2002) 207.28] L.S. Artajo, M.P. Romero, M.J. Motilva, J. Sci. Food Agric. 86 (2006) 518.29] K. De La Torre-Carbot, O. Jáuregui, E. Gimeno, A.I. Castellote, R.M. Lamuela-

Raventós, M.C. López-Sabater, J. Agric. Food Chem. 53 (2005) 4331.30] K. Hrncirik, S. Fritsche, Eur. J. Lipid Sci. Technol. 106 (2004) 540.31] L.S. Artajo, M.P. Romero, J.R. Morelló, M.J. Motilva, J. Agric. Food Chem. 54 (2006)

6079.32] J.R. Morelló, M.J. Motilva, M.J. Tovar, M.P. Romero, Food Chem. 85 (2004)

357.33] J.R. Morelló, S. Vuorela, M.P. Romero, M.J. Motilva, M. Heinonen, J. Agric. Food

Chem. 53 (2005 2002).34] A. Bendini, M. Bonoli, L. Cerretani, B. Biguzzi, G. Lercker, T.G. Toschi, J. Chro-

matogr. A 985 (2003) 425.

35] B.K. Choi, D.M. Hercules, A.I. Gusev, J. Chromatogr. A 907 (2001) 337.36] J. Castro-Perez, R. Plumb, J.H. Granger, I. Beattie, K. Joncour, A. Wright, Rapid

Commun. Mass Spectrom. 19 (2005) 843.37] M. Petrovic, M. Gros, D. Barcelo, J. Chromatogr. A 1124 (2006) 68.38] M.J. Oliveras-López, M. Innocenti, C. Giaccherini, F. Ieri, A. Romani, N. Mulinacci,

Talanta 73 (2007) 726.