Steroids 70 (2005) 896–906

Identification and quantitative analysis of�-sitosterol oxides invegetable oils by capillary gas chromatography–mass spectrometry

Xin Zhanga, Diane Julien-Davida, Michel Mieschb, Philippe Geoffroyb, Francis Raulc,Stamatiki Roussic, Dalal Aoude-Wernerd, Eric Marchionia,∗

a Laboratoire de Chimie Analytique et Sciences de l’Aliment (UMR 7512), Faculte de Pharmacie,Universite Louis Pasteur, 74 route du Rhin, 67400 Illkirch, France

b Laboratoire de Chimie Organique Synthetique (UMR 7123), Universite Louis Pasteur, 1 rue Blaise Pascal, 67008 Strasbourg, Francec Laboratoire d’Oncologie Nutritionnelle, Inserm UMR S392/IRCAD, 1 place de l’Hopital, 67091 Strasbourg, France

d Aerial, rue Laurent Fries—Parc d’Innovation, 67412 Illkirch, France

Received 28 October 2004; received in revised form 9 June 2005; accepted 9 June 2005Available online 20 July 2005

Abstract

f the mostic solventryive responseee recovere standard

instrikingly

Thermal-

y,ihy-

meenic,l

andlastnse-

ofducea.d on

raphy

As vegetable oils and phytosterol-enriched spreads are marketed for frying food or cooking purposes, temperature is one oimportant factors leading to the formation of phytosterol oxides in food matrix. A methodology based on saponification, organextraction, solid-phase extraction (SPE), followed by mass spectrometric identification and quantitation of�-sitosterol oxides using capillagas chromatography–mass spectrometry (GC–MS) in selected ion monitoring (SIM) mode was developed and characterized. Relatfactors of six�-sitosterol oxides, including 7�-hydroxy, 7�-hydroxy, 5,6�-epoxy, 5,6�-epoxy, 7-keto, and 5�,6�-dihydroxysitosterol, wercalculated against authentic standards of 19-hydroxycholesterol or cholestanol. Linear calibration data, limit of detection, and sampliesduring analytical process. Recoveries of these oxidation compounds in spiked samples ranged from 88 to 115%, while relativderivation (R.S.D.) values were below 10% in most cases. The analytical method was applied to quantify�-sitosterol oxides formedthermal-oxidized vegetable oils which were heated at different temperatures and for varying time periods. Sitosterol oxidation ishigher in sunflower oil relative to olive oil under all conditions of temperature and heating time.© 2005 Elsevier Inc. All rights reserved.

Keywords: �-Sitosterol oxides; Gas chromatography–mass spectrometry (GC–MS); Selected ion monitoring (SIM) mode; Quantification; Oil;oxidation

1. Introduction

Phytosterols in edible vegetable oils or plant sterol-enriched foods are structurally similar to cholesterol, theonly differences being the position of double bond and thealkyl side chain. Both cholesterol and phytosterols contain anunsaturated ring-structure and are thus susceptible undergooxidation in the presence of oxygen. Oxidation is acceleratedby heating, exposure to ionizing radiation, light, chemicalcatalysts, or enzymatic processes[1–5]. For cholesterol,more than 80 oxides have been identified and extensively

∗ Corresponding author. Tel.: +33 3 90244326; fax: +33 3 90244325.E-mail address: eric.marchioni@pharma.u-strasbg.fr (E. Marchioni).

studied and reviewed[6–9]. The main oxides are hydroxketo, and epoxy compounds. The latter may hydrate to ddroxy sterols[6]. The in vitro studies have shown that socholesterol oxides have cytotoxic, mutagenic, atherogand carcinogenic activities[10,11]. Due to the structurasimilarity with cholesterol and its oxides, phytosterolstheir oxides have received much attention during thedecades with respect to their biological properties and coquently, their performance in food safety evaluation[12–16].Some studies[9,12] have shown that supplementationfoods with phytosterols and phytosterol esters may inthe presence of their oxides in foods and human plasm

The analytical methods for phytosterol oxides are basethose developed for cholesterol oxides. Gas chromatog

0039-128X/$ – see front matter © 2005 Elsevier Inc. All rights reserved.doi:10.1016/j.steroids.2005.06.004

X. Zhang et al. / Steroids 70 (2005) 896–906 897

(GC) or high performance liquid chromatography (HPLC)is most often employed for the separation and quantificationof cholesterol oxides. Although HPLC provides good res-olution for cholesterol oxides[1,17], this technique is notsuitable for phytosterol oxides analysis because of excessiveinterference, modest sensitivity, and limited chromatographicresolution. For this reason, final separation and identifica-tion of phytosterol oxides in food matrix are always carriedout by GC–FID or GC–MS[18–24]. Guardiola et al.[25]and Diczfalusy[26] reviewed the analysis of sterol oxidesin foods and biological samples. Studies dealing with steroloxide mixtures from complex food samples using GC–MSoperating in selected ion monitoring (SIM) mode as quan-titative technique may increase the detection selectivity andavoid the numerous interfering problems[27,28].

Since individual standards of phytosterol oxides arenot available commercially, most research works, includ-ing chemical quantitative analysis, biochemical, and medicalstudies, have used mixtures of phytosterol oxides[15,16,20].Six individual �-sitosterol oxides have been synthesizedrecently[29]. In the present study, these compounds wereused as standards, a precise quantitative analytical methodof these products formed in thermal-oxidized oils are per-formed. Much attention was paid to the formation of oxideswhen heating vegetable oils for short periods of time (≤1 h).

2

2

-5-e sedf nter-n R( rcialp lo ,s5s -e( he-sd ks rds,c sepa-r

el-g romF l-deH Pro-l otas-s er-m( were

of analytical grade. Ultra-pure water was obtained bymeans of a Millipore Q water purification system (Milford,MA, USA). N-Methyl-N-(trimethylsilyl)trifluoroacetamide(MSTFA), used as the silylation reagent, was providedby Sigma-Aldrich (Steinheim, Germany). The SiOH-SPEcartridges (3 mL/500 mg) were purchased from Macherey-Nagel (Chromabond®, Duren, Germany).

2.2. Food samples

Sunflower oil (1 L, Lesieur Tournesol, France), olive oil(0.5 L, Puget, France), and butter (250 g, President, France)were of edible quality and purchased in a local supermarket.

2.3. Samples pretreatment

Firstly, 1 mg of�-sitosterol was weighed in an open glass-ware (10 mm× 60 mm i.d.) and heated in an oven (Memmert,Schwabach, Germany) which had been set at 100, 150, or200◦C for 5, 15, and 30 min. The final products were con-verted to TMS ethers (as described below) and analyzeddirectly by GC–MS to estimate the main oxides formed.After the model assay, approximately 200 mg of oil sam-ples (sunflower oil or olive oil) was weighted in the sametype of glassware to ensure efficient surface contact with air,a t 150o heg ereda omt ethyla dry-nn

2f

e forp hodspc

•) as0 mgofanddedandkerer-

5 h).• ep-

ed.thylere

. Experimental

.1. Materials and reagents

5�-Cholestan-3�-ol (cholestanol, 95%) and cholestne-3�,19-diol (19-hydroxycholesterol, 95%), purcha

rom Sigma (Steinheim, Germany), were used as ial standards.�-Sitosterol was purified from Generol 95Cognis, Saint-Fargeau-Ponthierry, France), a commehytosterol mixture, in the laboratory[29]. �-Sitosteroxides, e.g., sitost-5-ene-3�,7�-diol (7�-hydroxysitosterol)itost-5-ene-3�,7�-diol (7�-hydroxysitosterol), 5,6�-epoxy-�-sitostan-3�-ol (5,6�-epoxysitosterol), 5,6�-epoxy-5�-itostan-3�-ol (5,6�-epoxysitosterol), 3�-hydroxysitost-5n-7-one (7-ketositosterol), and sitost-5-en-3�,5�,6�-triol5�,6�-dihydroxysitosterol), were then chemically syntized [29]. The purity of each compound was∼95% asetermined by1H NMR, 13C NMR, and GC–MS. Stocolution (1 mg/mL) of each analyte and internal standaholestanol and 19-hydroxycholesterol, were preparedately in ethyl acetate and stored at 4◦C in the dark.

Ethyl acetate was obtained from Acros (Geel, Bium), ethanol from Carlo Erba (France), cyclohexane fluka (Buchs, Switzerland), diethyl ether from Riedeaen (Seelze, Germany), acetone and pyridine from

abo (Fontenay sous Bois, France), isooctane and pium hydroxide (KOH) from Merck (Darmstadt, Gany), and anhydrous sodium sulfate (Na2SO4) from SDS

Peypin, France). All organic solvents and reagents

nd then, they were placed into the oven and heated ar 200◦C. After various durations (from 5 to 60 min), tlasswares were removed from the oven, quickly covnd sealed with aluminium foil, and allowed to cool to ro

emperature. The samples were then dissolved in excesscetate, collected into 30 mL brown vials, evaporated toess under a gentle nitrogen flow, and stored at 4◦C underitrogen. Each sample was assayed in triplicate.

.4. Isolation and purification of phytosterol oxides fromood samples

The procedure for sample preparation described herhytosterol oxides analysis was adapted from the metroposed by Dutta[18] and Lampi et al.[20] with minorhanges. The detailed processes were as follows:

Cold saponification. Prior to saponification, 20�L of 19-hydroxycholesterol solution (1 mg/mL in ethyl acetatean internal standard was spiked into the heated 20of oil samples in a 30 mL brown vial. After removalthe ethyl acetate under nitrogen, 9 mL of ethanol,0.5 mL of saturated KOH aqueous solution were adinto the vial. The sample was purged with nitrogen,the tightly closed brown vial was put into a rotary sha(Edmund Buhler, Johanna Otta GmbH, Hechingen, Gmany) at ambient temperature in the dark overnight (1Extraction. The solution was transferred in a 100 mL saration funnel and 10 mL of distilled water were addThe unsaponifiable fraction was extracted with dieether (4× 15 mL). The combined organic extracts w

898 X. Zhang et al. / Steroids 70 (2005) 896–906

washed with 20 mL of 0.5 M aqueous KOH solution and0.2 M aqueous Na2SO4 solution (3× 15 mL). The organicsolvents were removed with a rotary vacuum evapora-tor (25◦C, 200 mbar), and the residue was dried undernitrogen flow. After additional 2 mL of diethyl ether, theunsaponifiable residue was carefully transferred using apipette to a 2 mL glass test tube and then evaporated to drymatter under nitrogen flow.

• Solid-phase extraction (SPE) purification. The SiOH-SPEcartridges were first conditioned with 5 mL of cyclohex-ane. The unsaponifiable matter was dissolved in 1 mLof a solution of cyclohexane/diethyl ether (9:1, v/v) andwas loaded onto the cartridge. Low-polarity lipids andnon-oxidized sterols were eluted with 5 mL of cyclohex-ane/diethyl ether (9:1, v/v) followed by 4 mL of cyclo-hexane/diethyl ether (1:1, v/v). Finally, the sterol oxideswere eluted out of the cartridge with 5 mL of acetone andcollected into a test tube to which 20�L of cholestanol(1 mg/mL in ethyl acetate) were added. This residue wasdried under a gentle nitrogen flow.

• Derivatization. The purified sterol oxides residue was con-verted to trimethylsilyl (TMS) ethers with 50�L of pyri-dine and 40�L of MSTFA at room temperature in the darkovernight and then, diluted with 410�L of isooctane. Onemicroliter of each sample was injected into the GC–MSsystem.

2a

400G ctorc ctor( stru-m oft-w umn( ane,t ec 5(a di-tt( s ac

ereo (4 ctedi ifieda

n

w thes spect d in

the sample and of internal standard added to the sample,respectively;τc andτ is the method recoveries of componentanalyzed in the sample and of internal standard added to thesample, respectively; and RRFc was the relative response fac-tor of each pure�-sitosterol oxide compared to the internalstandard, which was calculated as follows:

RRFc = n′c

n′is

A′is

A′c

RRFis

whereA′c andA′

is were the peak areas of standard oxide and ofinternal standard, respectively;n′

c andn′is the moles of stan-

dard oxide and of internal standard, respectively; and RRFiswas the response factor of the internal standard, which wasset to 1.

2.6. Linearity, limit of detection and recovery

For each oxide, standard solutions were made at six differ-ent concentrations between 0.1 and 100�g/mL, derivatized,and assayed in duplicate. Linear plots of peak areas ver-sus concentration were calculated. Recoveries of SiOH-SPEpurification were tested with mixtures containing the six�-sitosterol oxides as well as 19-hydroxycholesterol (10, 20,and 40�g of each one) spiked with 20�g of cholestanol asan internal standard. Recoveries of the complete analyticalmethod were carried out in 200 mg of butter spiked with thesh

3

3o

i7 edf ofG s oft apil-l esw ntiont anti-t rizedi

r typ-i witht si onedt , giv-i ulari7 asep -t .

.5. Gas chromatography–mass spectrometry (GC–MS)nalysis

GC–MS analyses were performed on a Varian STAR 3C instrument equipped with an on-column SPI injeoupled to a Varian SATURN 2000 mass sensitive deteVarian, France). Data acquisition and processing, and inental control were performed by Varian Saturn WS sare. Analytes were separated in a VF-5ms capillary col

phase stationary: 5% phenyl–95% dimethylpolysiloxhickness of 0.1�m, 60 m× 0.25 mm, Varian, France). Tholumn temperature gradient was programmed from 10◦Chold for 2 min) to 170◦C at 20◦C/min and then, to 320◦Ct 7◦C/min (hold for 15 min). The injector operating con

ions were as follows: injection volume 1�L; initial injectoremperature of 105◦C was increased to 300◦C at 100◦C/minhold for 40 min). Helium (purity 99.9995%) was used aarrier gas with a flow rate of 1 mL/min.

Electronic impact (EI) ionization mode mass spectra wbtained at 70 eV and monitored on the full-scan rangem/z0–600). Quantitative data were carried out using sele

on monitoring (SIM) analysis. The analytes were quants follows:

c = Ac

Aisnis

[RRFc

τis

τc

]

herenc andnis were the moles of component analyzed inample and of internal standard added to the sample, reively; Ac andAis the peak areas of component analyze

-

ix �-sitosterol oxides (10, 20, and 40�g) and 20�g of 19-ydroxycholesterol as an internal standard.

. Results and discussion

.1. Identification and characterization of β-sitosterolxides by GC–MS

A series of individual�-sitosterol oxides (∼95% purity),ncluding 7�-hydroxy, 7�-hydroxy, 5,6�-epoxy, 5,6�-epoxy,-keto, and 5�,6�-dihydroxysitosterol, were synthesiz

rom purified �-sitosterol and characterized by meansC–MS as TMS ethers. Chromatographic separation

hese compounds were well achieved on a VF-5ms cary column and identified by their MS data. Retention timere consistent within 1.2 s across runs. The typical rete

imes and selected ions chosen for identification and quation along with their relative abundances are summan Table 1.

The mass spectra of these oxides showed the similacal fragmentation patterns, which were in agreementhe data reported previously[2,30,31]with minor differencen the relative abundances. One point should be mentihat some mass spectra rounded numbers to the unitng occurrence to impaired mass fragments of molecon or base peak. For the mass spectra of 7�-hydroxy and�-hydroxysitosterol as di-TMS ether derivatives, the beak was observed atm/z 485 (M+ − TMSOH). The rela

ive intensity of molecular ions (M+) at m/z 575 was 1%

X. Zhang et al. / Steroids 70 (2005) 896–906 899

Table 1Retention time (tR) and relative retention time (RRT) of�-sitosterol and its oxides as TMS ethers in relation to that of 19-hydroxycholesterol, as well as theirspecific ions

Analytes tR (min) RRTa Mwb SIM ions (relative abundance, %)

Identification Quantitation

�-Sitosterol 31.42 1.06 486.7 382 (52), 397 (100), 472 (16), 487 (32) 397Cholestanol 29.28 0.99 460.7 215 (100), 356 (65), 446 (29), 461 (12) 35619-Hydroxycholesterol 29.55 1.00 546.7 73 (100), 353 (75), 367 (59), 547 (3) 3537�-Hydroxysitosterol 30.48 1.03 574.7 470 (4), 485 (100), 575 (1) 4857�-Hydroxysitosterol 32.03 1.08 574.7 470 (3), 485 (100), 575 (1) 4855,6�-Epoxysitosterol 33.08 1.11 502.7 73 (100), 395 (38), 413 (52), 503 (24) 4135,6�-Epoxysitosterol 33.32 1.12 502.7 73 (100), 395 (60), 413 (46), 503 (15) 3955�,6�-Dihydroxysitosterol 34.07 1.15 664.7 432 (100), 485 (84), 560 (34), 575 (13) 4327-Ketositosterol 35.58 1.20 500.7 396 (100), 486 (41), 501 (80) 396

a Relative to the retention time of 19-hydroxycholesterol TMS ether.b Mw: molecular weight as TMS ether.

The other typical fragments 470 (M+ − TMSOH-CH3), 395(M+ − 2TMSOH), 253 (M+ − 2TMSOH-side chain), and 129(�5-sterol) were of low intensity (1–6%), except form/z73 (22%). For the TMS ether derivatives of 5,6�-epoxyand 5,6�-epoxysitosterol, similar ion fragmentation patternswere found, apart from some differences in their relativepeak intensities. The molecular ion atm/z 503 was presentfor both compounds (15 and 24%), and the base peak wasat m/z 73. The molecular ion of 7-ketositosterol TMS etherwas atm/z 501 (M+, 80%), and the base peak was atm/z396 (M+ − TMSOH-CH3, 100%). This result was differentfrom previously reported data[28,31] in which an unspec-ified base peak atm/z 174 was observed. The other typicalfragments were consistent with the data reported previously[31,32].

Special attention should be taken with 5�,6�-dihydroxysitosterol. Indeed, depending on the silylatingreagents and on the reaction time used, bis- andtris-TMS ether derivatives may be produced from 5�,6�-dihydroxycholesterol or 5�,6�-dihydroxysitosterol becausethe derivatization at the 5�-hydroxyl group of the sterolis sterically hindered[31,32]. In the present work, whenderivatization was performed at 50◦C for 60 min, three

hydroxyl groups of 5�,6�-dihydroxysitosterol could notcompletely be derivatized to the tris-TMS ether derivative. Agas chromatogram showed two peaks, one (tR = 33.98 min)corresponding to tris-TMS ether (with three OTMS groups)and the other (tR = 37.21 min) to the bis-TMS ether (withone hydroxy and two OTMS groups) (Fig. 1(a)). Toavoid this phenomenon, the derivation reaction betweenanalytes and MSTFA was carried out at ambient temper-ature overnight in the dark. The three hydroxyl groupsof 5�,6�-dihydroxysitosterol could then be converted totris-TMS ether (Fig. 1(b)). Its base peak was observedat m/z 432 (M+ − TMSOH-side chain). The other majorfragments as well as their relative intensities were atm/z 575(M+ − TMSOH, 13%), 560 (M+ − TMSOH-CH3, 34%), 485(M+ − 2TMSOH, 84%), 470 (M+ − 2TMSOH-CH3, 28%),395 (M+ − 3TMSOH, 11%), and 380 (M+ − 3TMSOH-CH3,4%). Moreover, this method was suitable for the silylationof all other studied oxides present in the samples prior toGC analysis.

Relative response factors (RRF) of six�-sitosterol oxideswere calculated with SIM mode in the present workin relation to the internal standard, cholestanol or 19-hydroxycholesterol, are listed inTable 2.

Table 2R hers in a

A

Relativdroxyc ol

�

C .45± 0.17 .05± 0.7 .18± 0.5 .73± 0.5 .66± 0.5 .86± 0.7 .75± 0.

elative response factors (RRF) of�-sitosterol and its oxides as TMS et

nalytes RRF (TIC mode)

Relative to cholestanolhy

-Sitosterol 1.06± 0.04b –holestanol – 19-Hydroxycholesterol 0.69± 0.05 –�-Hydroxysitosterol 0.79± 0.05 1�-Hydroxysitosterol 0.85± 0.07 1,6�-Epoxysitosterol 1.99± 0.31 2,6�-Epoxysitosterol 1.25± 0.09 1�,6�-Dihydroxysitosterol 0.62± 0.29 0-Ketositosterol 1.31± 0.09 1a Mean± standard derivation (n = 10).b Mean± standard derivation (n = 3).c The characteristic ions for quantitation in SIM mode, seeTable 1.

relation to those of cholestanol or 19-hydroxycholesterol

RRF (SIM mode)c

e to 19-holesterol

Relative to cholestanol Relative to 19-hydroxycholester

0.17± 0.01b –08 – 1.29± 0.05

0.77± 0.03 –09 0.12± 0.01 0.16± 0.0108 0.13± 0.01 0.17± 0.0117 6.44± 0.20 8.29± 0.2814 3.54± 0.08 4.63± 0.1607 0.35± 0.02 0.49± 0.0213 1.00± 0.08 1.31± 0.12

900 X. Zhang et al. / Steroids 70 (2005) 896–906

Fig. 1. Influence of silylation conditions on the derivatization of 5�,6�-dihydroxysitosterol: (a) 50◦C for 60 min, (b) room temperature for overnight. The firstsilylation condition lead to the formation of two peaks as shown in GC chromatogram (a). The upper left mass spectrum corresponds to tris-TMS ether (peak1, with three OTMS groups). The upper right mass spectrum corresponds to bis-TMS ether (peak 2, with one hydroxy and two OTMS groups).

In previous works, because the phytosterol oxides were notcommercially available, the RRF values of these compoundswere usually adopted from those of the corresponding choles-terol oxides[19,20]or assumed to be a value of 1[33,34] inGC–FID or GC–MS analysis with total ion chromatogram(TIC) mode. However, although good separation and analyt-ical methods had been reported for cholesterol oxides[6,7,9],the chromatographic resolution problems associated with theanalysis of phytosterol oxides in foods of plant origin ormixed origin (animal/plant-based foods) are quite complexbecause of the high probabilities of co-elution[9,22]. Suchas, in the present GC separation system, 5,6�-epoxysitosterolwas co-eluted with 7�-hydroxycampesterol. Therefore, dueto high relative intensity of some characteristic ions, SIMmode has advantages in sensitivity, identification and quan-titation of sterol oxides in food matrix. Guardiola et al.[25]

reviewed the characteristic ions (m/z) used to quantify choles-terol oxides as trimethylsilyl ether derivatives by GC–MS.However, in relation to phytosterol oxides, the literature isscarce. The only work that has quantified 7-ketositosterolin food samples by GC–MS operated in SIM mode wasperformed by Turchetto et al.[28], who selected the char-acteristic ions atm/z 500, 484, 410, and 395 to quantify theTMSE derivative of this compound.

Since six �-sitosterol oxides were synthesized andcharacterized recently, their exact RRF values wereavailable. The ions selected for quantification werem/z 353 for 19-hydroxycholesterol,m/z 485 for 7�-hydroxy and 7�-hydroxy, m/z 413 for 5,6�-epoxy, m/z395 for 5,6�-epoxy, m/z 432 for 5�,6�-dihydroxy, andm/z 396 for 7-ketositosterol. In some cases, their basepeaks were chosen (7�-hydroxy, 7�-hydroxy, 5�,6�-

X. Zhang et al. / Steroids 70 (2005) 896–906 901

Table 3Analytical parameters of gas chromatography for each�-sitosterol oxide as TMS ethers assayed by the present method

Analytes Linear analysisa Limit of detection (�g/mL) % Recovery of SPE purificationb % Method recoveryc

Calibration curve R2

19-Hydroxycholesterol – – – 90.1± 6.9 87.2± 3.07�-Hydroxysitosterol y = (2.9x − 0.4)× 105 0.9966 0.5 97.5± 3.3 92.5± 5.07�-Hydroxysitosterol y = (1.7x + 0.9)× 105 0.9984 0.5 99.2± 8.4 90.2± 8.35,6�-Epoxysitosterol y = (1.7x − 1.3)× 105 0.9927 3 107.1± 10.9 115.3± 14.85,6�-Epoxysitosterol y = (1.6x + 0.7)× 105 0.9986 1 90.6± 7.8 102.3± 6.75�,6�-Dihydroxysitosterol y = (3.1x + 0.6)× 105 0.9997 1 88.7± 7.0 85.5± 5.37-Ketositosterol y = (2.6x − 1.0)× 105 0.9954 0.5 90.5± 6.5 87.2± 4.0

a Calibration curves were obtained by peak areas (y) vs. plotting concentration (x) (n = 10).b Mean± standard derivation (n = 10).c Mean± standard derivation (n = 7).

dihydroxy, and 7-ketositosterol). However, for 5,6�-epoxy, 5,6�-epoxysitosterol, 19-hydroxycholesterol, andcholestanol, the selected quantification ions were not thebase peaks (m/z 73 or 215) because these two ions werenot characteristic. Their detection was then not selectiveenough. Other characteristic ions (m/z 413, 395, 353, and356) were chosen for their quantitative analysis.

The results showed that their RRF values were sig-nificantly different. They ranged from 0.12 to 6.44 withcholestanol or from 0.16 to 8.29 with 19-hydroxycholesterolin SIM mode. Even in TIC mode, their RRF values were alsodifferent from 0.62 to 1.99 with cholestanol and from 0.86to 2.73 with 19-hydroxycholesterol. The response factors of5,6�-epoxy and 5,6�-epoxysitosterol were astonishingly dif-ferent. Then, the use of an arbitrary RRF equal to 1 within MSdetection (SIM mode) will necessarily lead to important ana-lytical errors. The RRF values were of utmost importance andtaken into account for quantification of�-sitosterol oxides inthe following analysis.

3.2. Linearity, limit of detection, and recovery

19-Hydroxycholesterol is a good choice as an internalstandard because it has quite the same analytical behaviouras cholesterol oxides[20,27,35,36]. Thus, it could be addedd ilars eten-t outb

fore n ofp ctions( doT eirT7ah sed.

l-a age

method recoveries between 85 and 115% for each oxide whenspiked in butter and assayed in complete analytical methodwith R.S.D. values below 10% (n = 7) in most cases indicatedgood recovery and repeatability of the method. In the quan-titative calculation, the method recovery values were takeninto account.

3.3. Optimization of the analytical method

Generally, methods for determination of sterol oxides con-sist of saponification, lipid extraction, purification, identifi-cation by GC–MS, and quantification. In order to minimizethe formation of artifacts and avoid the decomposition ofthe keto and epoxy derivatives during the saponification step[34,37,38], the saponification of lipids from oils was per-formed at room temperature overnight.

The possibility of artifact formation was studied duringthis cold saponification and following the workup processes.Two hundred milligrams of butter were spiked with 1 mg pure�-sitosterol and the analytical method was applied. Butterwas selected as a research medium because it contains no phy-tosterols and consequently, no phytosterol oxides. GC resultsshowed that no detectable level of�-sitosterol oxides waspresent in the extracts. Moreover, the possibility of intercon-version of each oxide during saponification and SPE purifi-cation was also evaluated. While each individual 20�g of� but-t od,n y thea

an-u car-t (con-t eu-t enti al.[ her( fromt thep L ofc 5,6e fth

irectly to the food samples and then withstand simaponification and purification processes. Moreover, its rion time was close to those of phytosterol oxides but witheing co-eluted with them in the GC chromatogram.

Table 3summarizes the linearity and limit of detectionach�-sitosterol oxide in the present study. The calibratioeak areas versus concentrations generated linear funcoefficient of correlationr2 = 0.9927–0.9997) for all studiexides within a range from limit of detection to 80�g/mL.he limits of detection (LOD) of the six oxides as thMS ethers were estimated to be 0.5�g/mL for 7�-hydroxy,�-hydroxy, and 7-ketositosterol, 1�g/mL for 5,6�-epoxynd 5�,6�-dihydroxy sitosterol, and 3�g/mL for 5,6�-ydroxysitosterol. A signal-to-noise ratio equal to 3 was u

As also shown inTable 3, most of the oxides had a retive recovery >90% in the SPE purification step. Aver

-sitosterol oxide was separately spiked into 200 mg ofer and followed workup by the complete analytical metho transformation happened between the oxides. Onldded oxide could be recovered in the GC analysis.

Due to the analytes complexity in the food matrix, a clep step prior to GC analysis was mandatory. SiOH-SPE

ridges were used to isolate and concentrate polar lipidsaining phytosterol oxides) from low-polarity lipids and nral lipid fraction (containing non-oxidized sterols) presn the unsaponifiable matter. In the work of Lampi et20], an elution volume of 5 mL of hexane/diethyl et1:1, v/v) was used to remove the non-oxidized sterolshe cartridge prior to the oxide fraction. However, inresent study, this elution volume was reduced to 4 myclohexane/diethyl ether because trace amounts of�-poxy and 5,6�-epoxysitosterol were detected in the fi

902 X. Zhang et al. / Steroids 70 (2005) 896–906

milliliter of this elution. Ten replicate experiments and GCresults showed no sterol oxides in the 4 mL of the cyclohex-ane/diethyl ether fraction (1:1, v/v), while all oxides werequantitatively removed from the SPE cartridge by the fol-lowing 5 mL of acetone.

3.4. Formation of β-sitosterol oxides by heatingβ-sitosterol

Phytosterols are basically stable compounds. However,under specific conditions, such as high temperature (>100◦C)in the presence of air, oxidation process of phytosterols mayoccur in the same way as for cholesterol[2,24,38,39]. Someauthors also reported the effect of refining and storage on theppm level of the sterol oxides formed[31,33,40].

In order to study the effect of high temperatures on theformation of sterol oxides, 1 mg of pure�-sitosterol washeated to 100, 150, or 200◦C. No detectable level of oxideswas found at 100◦C in 30 min. At 150◦C and 200◦C, themain �-sitosterol oxides represented during heating were7�-hydroxy, 7�-hydroxy, 5,6�-epoxy, 5,6�-epoxy, and 7-ketositosterol. No 5�,6�-dihydroxysitosterol was formedcertainly due to the lack of water in the system. The majoroxide formed at the two temperatures was 7-ketositosterol.The formation rate was quite constant during the first 30 mino lw eryq fter2 rmo-dS mo-o

3

plexm rols.� rols,a typesoo d to2 withco ts ofp liveo d0

veg-e eac-t ndi-t thep idesw pec-t odea

Fig. 2. Major oxides formed by heating�-sitosterol at different temper-atures. Each�-sitosterol oxide was quantified by capillary GC–MS withSIM mode: (a) 150◦C, (b) 200◦C. (�) 7�-hydroxysitosterol; (�) 7�-hydroxysitosterol; (�) 5,6�-epoxysitosterol; (×) 5,6�-epoxysitosterol; (�)7-ketositosterol.

Tables 4 and 5summarize the concentration of six�-sitosterol oxides formed before and after heating of sun-flower and olive oils at 150 and 200◦C for 60 min. Ini-tially, 7�-hydroxy (6.5�g/g of oil), 7�-hydroxy (7.0�g/gof oil), 7-keto (35.2�g/g of oil), and traces of 5�,6�-dihydroxysitosterol (2.1�g/g of oil) were found in the nativesunflower oil analyzed before heating. These oxides may beformed during oil processing and refinement[33,40]. Nodetectable levels of sterol oxides were found in the orig-inal olive oil. With heating, the amount of oxides formedin sunflower oil increased slowly during the first 30 min at150◦C and the first 15 min at 200◦C. The total content wasabout 75�g/g of oil. After these durations, the content of totaland individual�-sitosterol oxides increased quickly at bothtemperatures with a total level about three times higher at200◦C (815.1�g/g of oil) than that at 150◦C (240.9�g/g ofoil) after 60 min. The content of 5�,6�-dihydroxysitosterol(2–10�g/g) did not change significantly at two temperatures.

In olive oil, minor amounts of the oxides were formedafter heating the olive oil at 150◦C for 30 min or at 200◦C for

f heating at 150◦C (Fig. 2(a)). In contrast, when�-sitosteroas heated at 200◦C, the oxide content increased vuickly during the first 15 min, then slightly decreased a0–25 min. This phenomenon was probably due to theegradation and evaporation of the oxides formed (Fig. 2(b)).imilar results were shown in previous reports for the therxidation of stigmasterol or cholesterol[20,39].

.5. Quantification of β-sitosterol oxides in oils

Vegetable oils or plant sterol-enriched foods are comedia because they contain several different phytoste-Sitosterol is among the most representative phytostelong with campesterol, stigmasterol, and other variousf minor phytosterols. The initial content of�-sitosterol in theils used was quantified by applying the analytical metho00 mg of oil samples, and analyzing the fraction elutedyclohexane/diethyl ether (1:1, v/v) after spiking with 20�gf cholestanol as an internal standard. The total amounhytosterols were 2.6 and 1.6 mg/g in sunflower and oils, and the concentrations of�-sitosterol were 1.5 an.9 mg/g in the oils, respectively.

As mentioned above, the phytosterols contained intable oils are labile through thermolytic and oxidation r

ions when subjected to high temperatures or frying coions.Fig. 3presents the typical SIM chromatograms ofolar fractions in thermo-oxidized oils. The peaks of oxere identified by their retention times as well as mass s

ra, and quantitated with their characteristic ions of SIM mlong with their RRF, respectively, as shown inTables 1 and 2.

X. Zhang et al. / Steroids 70 (2005) 896–906 903

Fig. 3. Typical selection ion monitoring (SIM) chromatograms ofm/z 356 + 353 + 485 + 413 + 395 + 432 + 396 obtained by GC–MS analysis of (a) six�-sitosterol oxides and internal standards (20�g of each compound), as well as in (b) the polar fractions of the heated sunflower oil, and in (c) the polar fractionof the heated olive oil. The upper chromatograms are corresponding total ion chromatogram (TIC) of oxides standards and polar fractions in heated oils.Chromatographic peaks: (1) cholestanol (internal standard); (2) 19-hydroxycholesterol (internal standard); (3) 7�-hydroxysitosterol; (4) 7�-hydroxysitosterol;(5) 5,6�-epoxysitosterol; (6) 5,6�-epoxysitosterol; (7) 5�,6�-dihydroxysitosterol; (8) 7-ketositosterol.

904X

.Zhang

etal./Steroids70

(2005)896–906

Table 4Concentrations of�-sitosterol oxides (�g/g of oil) after heating sunflower oil at 150 or 200◦C for different durationsa

Analytes 150◦C 200◦C

0 min 5 min 15 min 30 min 60 min 0 min 5 min 15 min 30 min 60 min

7�-Hydroxysitosterol 6.5± 0.3 9.7± 0.6 9.8± 0.2 10.3± 0.4 31.6± 2.3 6.5± 0.3 10.1± 0.4 10.6± 0.3 33.1± 3.7 81.3± 9.37�-Hydroxysitosterol 7.0± 0.3 9.6± 0.7 9.4± 0.3 10.2± 0.5 49.5± 5.0 7.0± 0.3 9.7± 0.5 11.5± 0.4 46.2± 3.8 98.9± 8.75,6�-Epoxysitosterol n.d. 2.1± 0.2 2.9± 0.2 3.1± 0.6 45.0± 4.3 n.d. 2.9± 0.2 2.2± 0.2 46.9± 2.3 220.3± 11.25,6�-Epoxysitosterol n.d. 1.2± 0.0 2.0± 0.2 1.1± 0.1 24.3± 4.8 n.d. 1.1± 0.0 1.2± 0.1 20.2± 2.2 147.3± 21.55�,6�-Dihydroxysitosterol 2.1± 0.4 1.8± 0.1 1.8± 0.0 2.0± 0.2 3.9± 0.6 2.1± 0.4 1.8± 0.0 2.1± 0.2 3.8± 0.4 9.1± 0.87-Ketositosterol 35.2± 2.4 41.7± 1.3 44.0± 1.1 45.9± 2.2 86.6± 8.3 35.2± 2.4 41.8± 1.0 49.4± 0.7 72.9± 1.4 258.2± 16.5

Total amount (�g/g oil) 50.8 66.1 69.9 72.6 240.9 50.8 67.4 77.0 223.1 815.1% �-sitosterol oxidized 2.0 2.3 2.4 2.6 9.5 2.0 2.4 2.7 7.6 23.6

n.d. = not detected.a Mean± standard derivation (n = 3).

Table 5Concentration of�-sitosterol oxides (�g/g oil) after heating olive oil at 150 or 200◦C for different durationsa

Analytes 150◦C 200◦C

0 min 5 min 15 min 30 min 60 min 0 min 5 min 15 min 30 min 60 min

7�-Hydroxysitosterol n.d. 0.5± 0.0 0.6± 0.1 0.9± 0.1 5.4± 0.2 n.d. 0.5± 0.1 0.8± 0.1 21.1± 2.4 50.3± 1.17�-Hydroxysitosterol n.d. 0.5± 0.0 0.6± 0.1 1.1± 0.2 9.5± 0.2 n.d. 0.6± 0.2 1.1± 0.2 32.6± 3.0 54.8± 1.25,6�-Epoxysitosterol n.d. n.d. n.d. 1.5± 0.3 8.0± 0.4 n.d. n.d. 0.6± 0.2 31.5± 1.8 94.5± 5.75,6�-Epoxysitosterol n.d. n.d. n.d. 1.0± 0.1 3.8± 0.3 n.d. n.d. 0.2± 0.3 17.7± 0.2 56.6± 5.65�,6�-Dihydroxysitosterol n.d. n.d. n.d. n.d. 0.4± 0.3 n.d. n.d. n.d. 2.3± 0.0 3.2± 0.57-Ketositosterol n.d. 0.7± 0.1 1.0± 0.0 1.5± 0.2 10.0± 0.3 n.d. 0.7± 0.2 1.4± 0.2 20.5± 1.0 105.9± 9.3

Total amount (�g/g oil) n.d. 1.7 2.2 6.0 37.1 n.d. 1.8 4.1 125.7 365.3% �-sitosterol oxidized 0 1.0 1.3 3.0 7.2 0 1.1 2.4 4.8 20.3

n.d. = not detected.a Mean± standard derivation (n = 3).

X. Zhang et al. / Steroids 70 (2005) 896–906 905

15 min, but they contributed to less than 3% of the�-sitosteroloxidized. Conversely, the sunflower oil already containedsimilar levels of these oxides in its initial state. Trace levels of5�,6�-dihydroxysitosterol were found in olive oil (<4�g/g)after heating. The result also showed that higher tempera-ture will lead to more oxides formed. The total amounts of�-sitosterol oxides in olive oil were 37.1 and 365.3�g/g ofoil for heating at 150 and 200◦C after 60 min, respectively.

In the present heating assay of�-sitosterol and oils, the� configuration of the oxides formed more favorably thanthat of the corresponding� oxides. This result was in agree-ment with the previous report that formation of the�-epimerof cholesterol epoxide was favored over the�-epimer[6].The main oxides formed were the same at 150 and 200◦Cin the two oils. At 150◦C, 7�-hydroxy and 7-ketositosterolwere present in the highest concentrations, while at 200◦C,7-ketositosterol quantitatively predominated, followed by5,6�-epoxysitosterol. As the effect of cooking time on theoxidation of cholesterol in meats[36], the formation rate ofphytosterol oxides in plant oils differed according to the heat-ing time and temperatures. Lower temperature and shorterperiods might avoid the rapid formation of phytosterol oxides.However, significant amounts of oxides were detected in theoils heated at 200◦C for 60 min (20–25% of�-sitosterol oxi-dized). The total content of�-sitosterol oxides formed at200◦C for 60 min with sunflower oil was about two timesh

A

thep d bya tionN 02 P0

R

foructs

ofm

norOil

tos-

stry

pids

ids

lism.

[9] Guardiola F, Dutta PC, Savage GP, Codony R, editors. Cholesteroland phytosterol oxidation products in foods and biological samples:analysis, occurrence and biological effects. Champaign, IL: AOCSPress; 2002.

[10] Guardiola F, Codony R, Addis PB, Rafecas M, Boatell J. Bio-logical effects of oxysterols: current status. Food Chem Toxicol1996;34:193–211.

[11] Brown AJ, Jessup W. Oxysterols and atherosclerosis. Atherosclerosis1999;142:1–28.

[12] Ling WH, Jones PJH. Dietary phytosterols: a review of metabolism,benefits and side effects. Life Sci 1995;57:195–206.

[13] Piironen V, Lindsay DG, Miettinen TA, Toivo J, Lampi AM. Plantsterols: biosynthesis, biological function and their importance tohuman nutrition. J Sci Food Agric 2000;80:939–66.

[14] Moreau RA, Whitaker BD, Hicks KB. Phytosterol, phytostanols, andtheir conjugates in food: structural diversity, quantitative analysis,and health-promoting uses. Prog Lipid Res 2002;41:457–500.

[15] Adcox C, Boyd L, Oehrl L, Allen J, Fenner G. Comparative effectsof phytosterol oxides and cholesterol oxides in cultured macrophage-derived cell lines. J Agric Food Chem 2001;49:2090–5.

[16] Grandgirard A, Martine L, Demaison L. Oxyphytosterols are presentin plasma of healthy human subjects. Br J Nutr 2004;91:101–6.

[17] Shan H, Pang J, Li S, Chiang TB, Wilson WK, Schroepfer Jr GJ.Chromatographic behaviour of oxygenated derivatives of cholesterol.Steroids 2003;68:221–33.

[18] Dutta PC. Studies on phytosterol oxides. II. Content in some veg-etable oils and in French fries prepared in these oils. J Am Oil ChemSoc 1997;74:659–66.

[19] Plat J, Brzezinka H, Lutjohann D, Mensink RP, Bergmann K. Oxi-dized plant sterols in human serum and lipid infusions as measuredby combined gas-liquid chromatography–mass spectrometry. J Lipid

[ ionr B

[ prod-ectro-

[ O.entifi-riva-r A

[ of 14

[ ductslary

[ oxi-

[ og-

[ ationandr A

[ tion

[ df.

[ etry

Food

igher than that of olive oil under the same conditions.

cknowledgements

The authors thank Dr. Jennifer Wytko for her help inreparation of the manuscript. This work was supportegrant from the Ministere de la Jeunesse, de l’Educaationale et de la Recherche, France (RARE 015 No.640).

eferences

[1] Csallany AS, Kindom SE, Addis PB, Lee LH. HPLC methodquantitation of cholesterol and four of its major oxidation prodin muscle and liver tissues. Lipids 1989;24:645–51.

[2] Daly GG, Finocchiaro ET, Richardson T. Characterizationsome oxidation products of�-sitosterol. J Agric Food Che1983;31:46–50.

[3] Ferrari RA, Schulte E, Esteves W, Bruhl L, Mukherjee KD. Miconstituents of vegetable oils during industrial processing. J AmChem Soc 1996;73:587–92.

[4] Lutjohann D. Sterol autoxidation: from phytosterols to oxyphyterols. Br J Nutr 2004;91:3–4.

[5] Russell DW, Setchell KD. Bile acid biosynthesis. Biochemi1992;31:4737–49.

[6] Smith LL. Cholesterol autoxidation: 1981–1986. Chem Phys Li1987;44:87–125.

[7] Smith LL. Review of progress in sterol oxidation: 1987–1995. Lip1996;31:453–87.

[8] Schroepfer Jr GJ. Oxysterols: modulators of cholesterol metaboPhysiol Rev 2000;80:361–554.

Res 2001;42:2030–8.20] Lampi AM, Juntunen L, Toivo J, Piironen V. Determinat

of thermo-oxidation products of plant sterols. J Chromatog2002;777:83–92.

21] Johnsson L, Dutta PC. Characterisation of side-chain oxidationucts of sitosterol and campesterol by chromatographic and spscopic methods. J Am Oil Chem Soc 2003;80:767–76.

22] Grandgirard A, Martine L, Joffre C, Juaneda P, BerdeauxGas chromatographic separation and mass spectrometric idcation of mixtures of oxyphytosterols and oxycholesterol detives. Application to a phytosterol-enriched food. J Chromatog2004;1040:239–50.

23] Johannes C, Lorenz RL. Preparation and mass spectrometrypure and18O2-labeled oxidation products from the phytosterols�-sitosterol and stigmasterol. Anal Biochem 2004;325:107–16.

24] Johnsson L, Dutta PC. Separation of phytosterol oxidation proby combination of different polarity gas chromatography capilcolumns. J Chromatogr A 2005;1064:213–7.

25] Guardiola F, Bou R, Boatella J, Codony R. Analysis of steroldation products in foods. J AOAC Int 2004;87:441–66.

26] Diczfalusy U. Analysis of cholesterol oxidation products in biolical samples. J AOAC Int 2004;87:467–73.

27] Regueiro JAG, Maraschiello C. Procedure for the determinof eight cholesterol oxides in poultry meat using on-columnsolvent venting capillary gas chromatography. J Chromatog1997;764:279–93.

28] Turchetto E, Lercker G, Bortolomeazzi R. Oxysterol determinain selected coffees. Toxicol Ind Health 1993;9:519–27.

29] Zhang X, Geoffroy P, Miesch M, Julien-David D, Raul F, Aoue-Werner D, et al. Gram-scale chromatographic purification o�-sitosterol. Synthesis and characterization of�-sitosterol oxidesSteroids 2005;70:886–95.

30] Bortolomeazzi R, Zan MD, Pizzale L, Conte LS. Mass spectromcharacterization of the 5�, 7�, and 7�-hydroxy derivatives of�-sitosterol, campesterol, stigmasterol and brassicasterol. J AgricChem 1999;47:3069–74.

906 X. Zhang et al. / Steroids 70 (2005) 896–906

[31] Dutta PC, Appelqvist LA. Studies on phytosterol oxides. I. Effect ofstorage on the content in potato chips prepared in different vegetableoils. J Am Oil Chem Soc 1997;74:647–57.

[32] Pizzoferrato L, Nicoli S, Lintas C. GC–MS characterization andquantification of sterols and cholesterol oxidation products. Chro-matographia 1993;35:269–74.

[33] Bortolomeazzi R, Cordaro F, Pizzale L, Conte LS. Pres-ence of phytosterol oxide in crude vegetable oils and theirfate during refining. J Agric Food Chem 2003;51:2394–401.

[34] Louter AJH. Determination of plant sterol oxidation products in plantsterol enriched spreads, fat blend, and plant sterol concentrates. JAOAC Int 2004;87:485–92.

[35] Pie JE, Spahis K, Seillan C. Evaluation of oxidative degrada-tion of cholesterol in food and food ingredients: identification and

quantification of cholesterol oxides. J Agric Food Chem 1990;38:973–9.

[36] Pie JE, Spahis K, Seillan C. Cholesterol oxidation in meat prod-ucts during cooking and frozen storage. J Agric Food Chem1991;39:250–4.

[37] Park SW, Guardiola F, Park SH, Addis PB. Kinetic evaluationof 3�-hydroxycholest-5-en-7-one (7-ketocholesterol) stability duringsaponification. J Am Oil Chem Soc 1996;73:623–9.

[38] Ubhayasekera SJKA, Verleyen T, Dutta PC. Evaluation of GC andGC–MS methods for the analysis of cholesterol oxidation products.Food Chem 2004;84:149–57.

[39] Osada K, Kodama T, Yamada K, Sugano M. Oxidation of cholesterolby heating. J Agric Food Chem 1993;41:1198–202.

[40] Kochhar SP. Influence of processing on sterols of edible vegetableoils. Prog Lipid Res 1983;22:161–88.