A

qobectimpvaH©

K

1

uioaa[

0d

Journal of Chromatography A, 1145 (2007) 222–228

Comparison of available analytical methods to measuretrans-octadecenoic acid isomeric profile and content by

gas–liquid chromatography in milk fat

Frederic Destaillats a,∗, Pierre-Alain Golay a, Florent Joffre b,Maureen de Wispelaere a, Bernadette Hug a, Francesca Giuffrida a,

Laetitia Fauconnot a, Fabiola Dionisi a

a Nestle Research Centre, Vers-chez-les-Blanc, P.O.Box 44, CH-1000 Lausanne 26, Switzerlandb Omega 21, Marsannay le Bois, France

Received 15 November 2006; received in revised form 22 January 2007; accepted 22 January 2007Available online 26 January 2007

bstract

Accurate quantification of trans-fatty acids (TFAs) could be achieved by infrared spectroscopy or by gas–liquid chromatography (GLC). Accurateuantification by GLC should be achieved using specific highly polar capillary columns such as 100 m CP-Sil 88 or equivalent. A pre-fractionationf cis and trans-fatty acids could be performed by silver-ion thin-layer chromatography (Ag-TLC), silver-ion solid-phase extraction (Ag-SPE), ory high-performance liquid-chromatography (HPLC). A pre-fractionation step allows accurate determination of the isomeric profile but it is notssential to achieve quantification of total trans-18:1 isomers nor to determine the level of vaccenic (trans-11 18:1) acid in dairy fat. TFA contentould also be calculated in milk fat based on the TAG profile determined by GLC. In this paper, different GLC methods suitable to measure theotal of trans-18:1 isomers, vaccenic acid and trans-18:1 acid isomeric distribution in milk fat were compared. Pre-separation of cis- and trans-18:1somers by Ag-TLC followed by GLC analysis under optimal conditions was selected as the reference method. Results obtained using alternative

ethods including pre-separation by HPLC followed by GLC analysis, direct quantification by GLC or calculation from the triacylglycerol (TAG)rofile were compared to data acquired using the reference method. Results showed that accurate quantification of total trans-18:1 isomers and

accenic acid could be achieved by direct quantification by GLC under optimal chromatographic conditions. This method represents a very goodlternative to Ag-TLC followed by GLC analysis. On the other hand, we showed that pre-fractionation of fatty acid methyl esters (FAMEs) byPLC represents a good alternative to Ag-TLC, even if some minor isomers are not selectively purified using this procedure.2007 Elsevier B.V. All rights reserved.

id; tr

tftflg

eywords: Gas–liquid chromatography; Dairy product; Milk fat; trans-Fatty ac

. Introduction

Accurate quantification of trans-18:1 acids in food prod-cts is an important issue since policies have been recentlymplemented in different countries to limit their consumption orccurrence in food products [1,2]. trans-Isomers of octadecenoic

cid are the major contributor of trans-fatty acid (TFA) intakend these compounds are found in both vegetable and animal fats3]. Animal fats and in particular milk fat are important contribu-

∗ Corresponding author. Tel.: +41 21 785 8937; fax: +41 21 785 8553.E-mail address: frederic.destaillats@rdls.nestle.com (F. Destaillats).

mtt1otp

021-9673/$ – see front matter © 2007 Elsevier B.V. All rights reserved.oi:10.1016/j.chroma.2007.01.062

ans-Octadecenoic acid; Vaccenic acid

or to TFA intake. TFA is a generic term that defines unsaturatedatty acid containing at least one ethylenic double bond in therans configuration. In milk fat, TFAs are produced by anaerobicermentation of polyunsaturated fatty acids in the rumen of theactating cows [4]. This fermentation process is called biohydro-enation and resulting TFAs could be further metabolized in theammary gland. Indeed, vaccenic (trans-11 18:1) acid, the main

rans-18:1 acid isomer produced in the rumen is mobilized inhe mammary gland to be transformed in rumenic (cis-9, trans-

1 18:2) acid. The formation of rumenic acid, one of a numberf conjugated isomers of linoleic acid (CLAs), is catalyzed byhe stearoyl-CoA desaturase. Non-ruminant animals could alsoroduce rumenic acid from vaccenic acid [5–7]. Therefore, the

mato

oIoa

eom(tctcfaiobtofpabmutmpf

pauHfoit

2

2

lleR

2t

Tflm

(izamt(hAa

2t

i[i1aattBtshwappa

2h

itwaat

2e

dTElI

F. Destaillats et al. / J. Chro

ccurrence of vaccenic acid is a very unique feature of milk fat.n partially hydrogenated vegetable oil, another dietary sourcef TFA, vaccenic acid represent less than 10% of the trans-18:1cid isomeric profile.

Analysis of trans-18:1 isomers could be achieved by differ-nt methods [8–27]. One of the most accurate method consistsf pre-fractionation of cis- and trans-18:1 isomers as theirethyl ester derivatives by silver-ion thin-layer chromatography

Ag-TLC) followed by the analysis of the collected frac-ions by gas–liquid chromatography (GLC). This pre-separationould also been performed using silver-ion solid-phase extrac-ion (Ag-SPE) or by reversed-phase high-performance liquidhromatography (HPLC) [22,23]. The trans-18:1 acid isomerractions could be analyzed by GLC using highly polar station-ry phases to obtain accurate isomeric distribution of trans-18:1somers. These two different fractionation techniques are basedn very different chromatographic principles and have nevereen compared previously. Similar accurate quantification ofotal trans-18:1 isomers can be achieved without pre-separationf geometrical isomers [24]. Direct quantification can be per-ormed by the use of long (i.e. 100 m) and highly polar stationaryhase (e.g. CP-Sil 88 or equivalent) and under optimal temper-ture programming conditions [24]. A third type of method haseen developed by Precht and Molkentin 15 years ago [25]. Thisethod consists of calculation of the content of total or individ-

al trans-18:1 isomers by means of mathematical formulas usinghe relative content of some triacylglycerols (TAGs). These for-

ulas have been obtained through statistically correlating TAGrofile and level of trans-18:1 isomers measured by Ag-TLCollowed by high-resolution GLC analysis [25].

In the present paper, Ag-TLC followed by GLC analysiserformed under optimal conditions for trans-18:1 isomers sep-ration were selected as the reference method. Results obtainedsing alternative methodologies, including pre-separation byPLC, direct quantification by GLC or estimation by TAG

ormulas, have been compared to the reference method. Thebjective was to determine the more accurate method to obtainsomeric distribution of trans-18.1 isomers, to quantify totalrans-18:1 isomers and vaccenic acid in milk fat.

. Materials and methods

.1. Samples and reference material

Liquid milk samples (n = 6) were collected in The Nether-ands, France and Germany. Lipids were extracted according toiterature procedures [26]. Certified CRM-519 butterfat refer-nce material was obtained from the EU Community Bureau ofeference (BCR).

.2. Preparation of fatty acid methyl esters (FAMEs) fromotal lipid extract

FAMEs were prepared according to literature procedure [24].he mixture of internal standards was prepared in a volumetricask (100 mL) by dissolving equal amounts of tritridecanoin andethyl undecanoate (200 ± 0.1 mg) in methyl tert-butyl ether

cOit

gr. A 1145 (2007) 222–228 223

MTBE). A sample (50 mg) was weighed (to the nearest 0.1 mg)nto a centrifuge tube (25 mL) with a screw cap. The derivati-ation was conducted at room temperature for 3.5–4.0 min afterdding internal standard solution (5 mL), methanolic sodiumethoxide (5% solution, 5 mL) to the sample and shaking con-

inuously for 10 s. The reaction was stopped by adding n-hexane2 mL), and by adding an aqueous solution (10 mL) of disodiumydrogen citrate (0.1 g/mL) and sodium chloride (0.15 g/mL).fter phase separation, the supernatant was diluted before GLC

nalysis.

.3. Fractionation of fatty acid methyl esters by silver-ionhin-layer chromatography (Ag-TLC)

FAMEs were fractionated by TLC on silica gel platesmpregnated with silver nitrate according to literature procedure20,21]. The plates were immerged in a 5% silver nitrate solutionn acetonitrile for 15 min in the dark, and activated at 100 ◦C forh [20]. Fractionation was performed according to the numbernd configuration of double bonds, using a mixture of hexanend diethyl ether (90:10, v/v) as developing solvent. At the end ofhe chromatographic runs, the plates were sprayed with a solu-ion of 2′,7′-dichlorofluorescein, and viewed under UV light.ands corresponding to, respectively, saturated FAs (RF = 0.51),

rans (RF = 0.41) and cis (RF = 0.30) monounsaturated FAs, werecraped off and transferred into a test tube. Methanol (1.5 mL),exane (2 mL) and an aqueous solution of sodium chloride (5%/v, 1.5 mL) were successively added with thorough mixing

fter each addition. After standing for ca. 1 min, the organichase was withdrawn, and hexane was evaporated N2. The sam-les were stored at −18 ◦C under inert conditions prior to GLCnalysis.

.4. Fractionation of fatty acid methyl esters byigh-performance liquid-chromatography

FAMEs were fractionated by reversed-phase HPLC accord-ng to the methodology published by Juaneda [23]. The fractionhat contained trans-18:1 isomers plus palmitic acid methyl esteras collected. The mobile phase (acetonitrile) was evaporated

nd the FAME sample was dissolved in hexane prior to GLCnalysis. The extracts were stored at −18 ◦C under inert condi-ions prior to analysis.

.5. Gas–liquid chromatography analysis fatty acid methylsters

GLC analysis was performed according to previouslyescribed conditions [24]. Analysis of total FAMEs and Ag-LC fractions were performed on a gas chromatograph (Carlorba, Milan, Italy), equipped with a fused-silica CP-Sil 88 capil-

ary column (100% cyanopropylpolysiloxane, 100 m × 0.25 mm.D., 0.25 �m film thickness, Varian, Palo Alto, CA, USA)

old on-column injector and flame-ionization detector (300 ◦C).ven temperature programming was 60 ◦C isothermal for 5 min,

ncreased to 165 ◦C at 15 ◦C/min, isothermal for 1 min at thisemperature then increased to 225 ◦C at 2 ◦C/min and held

2 mato

ig

2c

lud6wtwSa2uamTi8C0t5C0

2e

(piltma(a(tala(eanlvomr

2

Tlntmw

3

bItntbMCces1ap9

ao(baetfHAus1stsohtbctls

24 F. Destaillats et al. / J. Chro

sothermal for 17 min at 225 ◦C. The inlet pressure of the carrieras (H2) was 200 kPa.

.6. Triacylglycerol (TAG) profile by gas–liquidhromatography

The separation of milk TAG was performed according toiterature procedure [26]. The extracts were stored at −18 ◦Cnder inert conditions prior to analysis. Fat samples wereiluted in n-hexane (1 mg/mL) and analyzed using an Agilent890N gas chromatograph (Palo Alto, CA, USA) equippedith a DB-5HT capillary column (4.0 m × 0.25 mm I.D., film

hickness 0.10 �m; J&W, Palo Alto, CA, USA). This columnas prepared from a 15 m commercially available DB-5HT.plit injection (50:1) and flame-ionization detection (FID) werechieved at 380 ◦C. Initial oven temperature programming was00 ◦C, increased to 350 ◦C at 50 ◦C/min. Carrier gas (H2) wassed in constant flow mode. The initial pressure was 50 kPat 200 ◦C. The calibration was performed using the certifiedaterial (CRM-519) as described in the original procedure [25].he formula used for the quantification of the level of trans-18:1

somers (in g/100 g of fatty acids) is: 3.7190C26 + 2.4439C28 −.1505C30 + 3.5187C32 − 0.4819C34 − 0.3737C36 + 0.3514-40 + 1.3285C42 − 1.3429C44 − 0.2689C46 + 0.9557C48 −.7356C52 + 1.0775C54. The formula used for the quantifica-ion of the level of vaccenic acid (in g/100 g of fatty acids) is:.0994C28 − 5.4430C30 + 1.2730C32 − 0.5137C38 + 0.6394-40 + 0.7179C42 − 1.4086C44 + 0.3522C46 + 0.5527C48 −.5443C52 + 0.7896C54.

.7. Identification and quantification of fatty acid methylsters (FAMEs)

For optimal GLC separation, we recommend the use of longat least 100 m), polar, open-tubular capillary columns, whichermit accurate separation of FAMEs including the cis and transsomers [24]. Response factors for each FAME could be calcu-ated using commercially available FAME standard mixtures. Inhe present study, a quantitative mixture containing the followingethyl esters of butyric acid (4:0), caproic acid (6:0), caprylic

cid (8:0), capric acid (10:0), undecanoic acid (11:0), lauric acid12:0), tridecanoic acid (13:0), myristic acid (14:0), myristoleiccid (14:1 n-5), pentadecanoic acid (15:0), pentadecenoic acid15:1 n-5), palmitic acid (16:0), palmitoleic acid (16:1 n-7), hep-adecanoic acid (17:0), heptadecenoic acid (17:1 n-7), steariccid (18:0), elaidic acid (trans-9 18:1), oleic acid (18:1 n-9),inolelaidic acid (all trans-18:2 n-6), linoleic acid (18:2 n-6),rachidic acid (20:0), �-linoleic acid (18:3 n-6), eicosenoic acid20:1 n-9), linolenic acid (18:3 n-3), heneicosanoic acid (21:0),icosadienoic acid (20:2 n-6), behenic acid (22:0), eicosatrienoiccid (20:3 n-6), erucic acid (22:1 n-9), eicosatrienoic acid (20:3-3), arachidonic acid (20:4 n-6), docosadienoic acid (22:2 n-6),ignoceric acid (24:0), eicosapentanoic acid (20:5 n-3), ner-

onic acid (24:1 n-9), docosahexaenoic acid (22:6 n-3) wasbtained from Nu-Check-Prep (Elysian, MN). trans-18:1 Iso-ers were identified according to previous work and literature

esults.

ocdi

gr. A 1145 (2007) 222–228

.8. Statistics

Analytical data are often subject to potential outlying values.o reduce their influence in the precision characteristics calcu-

ation, therefore robust statistics were used. Robust statistics areot unduly affected by outliers and therefore, their main advan-age is to be less sensitive than classical statistics, and provide

ore reliable results. Q-Stat 2.0.1 software (Nestle, Switzerland)as used for statistical analyses.

. Results and discussion

The direct measurement of trans-18:1 acid isomeric distri-ution by GLC is not possible without a pre-fractionation step.ndeed, even with high-resolutive GLC conditions, and separa-ion of cis/trans positional isomers of octadecenoic acid couldot be fully achieved routinely. Some chromatographic condi-ions have been proposed to avoid this step, that is considered toe time-consuming for routine analysis. In particular, Precht andolkentin published interesting chromatographic conditions onP-Sil 88 capilary column (100 m) that allow the separation ofis and trans positional isomers of octadecenoic acid methylsters in milk fat. Using these conditions, only minor overlaps,uch as cis-6/8–trans-13/14 18:1 isomers or cis-14–trans-168:1 isomers occur [25]. However, these separative conditionsre difficult to maintain with the same highly polar stationaryhases and in particular the separation between trans-13/14–cis-18:1 isomers and cis-9–trans-15 18:1 acid.Two different types of method have been developed to sep-

ratate cis- and trans-18:1 acid methyl esters isomers. The firstne was Ag-TLC, which principle is the formation of a complexcharge-transfer type) between silver ion and ethylenic doubleonds [28]. Stability of the complex formed between a silver ionnd a cis double bond is higher than for a silver ion and a trans-thylenic double bond. Therefore, cis- could be separated fromrans-olefins on TLC impregnated with silver ions (preparedrom nitrate salts). This separation could also be performed byPLC using silver-ion loaded column [11]. The advantage ofg-TLC is that it allows to scrap-off a fraction that contains sat-rated FAMEs in addition to trans-18:1 isomers. Among theseaturated FAMEs, stearic acid, that elutes just before the trans-8:1 acid methyl esters in GLC could be used as an internaltandard to measure absolute content of trans-18:1 isomers inhe sample. As an alternative to TLC, silver-ion impregnatedolid phase extraction (Ag-SPE) cartridges have been devel-ped to achieve this separation [22]. The second methodologyas been developed recently by Juaneda [23] and consists ofhe separation of cis and trans-18:1 acid methyl ester isomersy reversed-phase HPLC. Two preparative columns need to beonnected to achieve the separation and a fraction that containsrans-18:1, palmitic and conjugated acid linoleic isomers is iso-ated [23]. In this case, palmitic acid can be used as an internaltandard to calculate absolute content on individual isomers of

ctadecenoic acid. Separation of trans- and cis-18:1 isomersould also been achieved by Ag-HPLC [11]. However, the repro-ucibility of separation using Ag-HPLC is limited and thereforet will be difficult to use this method routinely.

mato

Goamppqetmto1bio1t[ate

FiF

rrachtp

tdp

3

bGTp

F. Destaillats et al. / J. Chro

Quantification of total trans-18:1 can be achieved by directLC analysis without pre-separation steps [14,18]. The method-logy developed by Thompson [27] is based on the fact thatt high temperature (up to 230 ◦C) and under specific chro-atographic conditions (long and very polar capillary column),

ositional isomers of cis- and trans-18:1 elute in two distincteaks. This method is very accurate for trans-18:1 acid isomeruantification in partially hydrogenated vegetable oils. How-ver, the limitation of the procedure for milk fat analysis ishat trans-16 18:1 acid methyl ester elutes with cis-18:1 acidethyl ester. The trans-16 18:1 acid isomer is an important con-

ributor in milk fat [25]. However, an estimation of up to 95%f total trans-18:1 isomers could be directly achieved using a00 m CP-Sil 88 capillary column [24]. As illustrated in Fig. 1,aseline resolution of both oleic (cis-9 18:1) and trans-15 18:1somers remain the critical chromatographic parameters to beptimized. By adding the area corresponding to the trans-168:1 acid isomer to the area corresponding to the trans-4 torans-13/14 18:1 isomers, up to 95% of TFAs can be estimated

24]. The underestimation due to the co-elution of trans-15 18:1nd cis-9 18:1 acid methyl esters is minimized if we considerhat positional cis-6/8 and cis-14 18:1 acid methyl esters co-lute with trans-13/14 18:1 and trans-16 18:1 acid methyl esters,

ig. 1. Resolution of total 18:1 isomers (A), trans-18:1 isomers (B) and cis-18:1somers (C) derived from milk fat with GLC on a 100 m column (CP-Sil 88).ractions have been obtained by Ag-TLC.

aGdGTomdusavGcc

ombmfft

gr. A 1145 (2007) 222–228 225

espectively. Results on 1765 milk samples, have shown that theelative content of these minor cis-18:1 isomers is constant [25]nd therefore, the error associated with the calculation proposedould be considered as negligible [24]. However, this methodas not been compared to the reference method and thereforehe real accuracy of this estimation has not been addressedreviously.

In this study, alternative methodologies suitable to measure,he absolute content of trans-18:1 isomers, vaccenic acid or toetermine isomeric distributions in milk fat by GLC are com-ared to the reference method: Ag-TLC prior to GLC analysis.

.1. Quantification of total trans-18:1 acid in milk fat

The quantification of total trans-18:1 acid in milk fat coulde achieved by three different types of methodologies: directLC [24], pre-separation of trans- and cis-18:1 isomers by Ag-LC or HPLC prior to GLC [23,20] and calculation from TAGrofile [25]. Milk fat was extracted from liquid milk samplesnd analyzed in duplicate by Ag-TLC fractionation followed byLC (reference method), HPLC fractionation followed by GLC,irect GLC and calculation from diagnostic TAGs analyzed byLC. The results, that include replicate values, are reported inable 1. Trueness parameters were estimated from replicate datan replicate and the results clearly show that the three alternativeethodologies give consistent results (R2 > 0.95). In particular,

irect GLC and HPLC fractionation followed by GLC gave val-es very close to the reference method (see Table 1). Resultsuggest that direct GLC is more precise than HPLC fraction-tion followed by GLC since the standard error of predictionalue (0.08) is lower than for HPLC fractionation followed byLC (0.14). The determination of the TAG profile by GLC

ould be achieved using short apolar column [26], and a typicalhromatogram is provided in Fig. 2.

The formulas developed by Precht and Molkentin werebtained from a large set of milk samples [25]. The TAG for-ula gives good estimation of the level of trans-18:1 isomers

ut compared to the other methods is not comparable. For the

easurement of the level of total trans-18:1 isomers in milk

at, the direct GLC method appears to be a very good alternativeor routine analysis. Moreover, experimental conditions suitableo achieve direct transesterification (without prior extraction) of

Fig. 2. Separation of milk fat TAGs by GLC.

226 F. Destaillats et al. / J. Chromatogr. A 1145 (2007) 222–228

Table 1Analysis of total trans-18:1 isomers by different techniques: results of the trueness-regression analysis

Sample Ag-TLC (g/100 g of FAs) Direct GLC (g/100 g of FAs) RP-HPLC (g/100 g of FAs) TAG (g/100 g of FAs)

1 5.10 5.05 5.15 5.302 4.90 5.05 5.19 5.293 4.60 4.57 4.53 4.924 4.70 4.62 4.78 4.975 3.30 3.22 3.34 3.026 3.20 3.19 3.37 2.977 4.10 4.16 4.22 3.978 4.10 4.12 4.33 3.989 3.10 3.15 3.27 3.2610 3.20 3.16 3.25 3.2711 3.10 3.09 3.06 3.1212 3.10 2.95 2.99 3.04

R2 – 0.99 0.98 0.96RSD (%) – 8.69 11.61 18.98SEP – 0.08 0.14 0.22

A nce mw on. RSi

lolbocs

3

ilora

acs

mhcHvTHp

TA

S

123456789111

RRS

Awi

g-TLC fractionation prior to high-resolution GLC was selected as the refereas performed on replicate (n = 12). R2 indicates the coefficient of determinati

ndicated the standard error of prediction.

ipids from liquid or powdered dairy products have been devel-ped by Golay et al. [24]. Therefore, the measurement of theevel of trans-18:1 isomers in dairy products could be performedy direct transesterification followed by GLC analysis underptimal conditions. The use of specific stationary phase andapillary column geometry is mandatory to achieve the requiredeparation (see Fig. 1).

.2. Quantification of vaccenic acid in milk fat

Vaccenic (trans-11 18:1) acid is the main trans-18:1 acidsomer present in milk fat (see Fig. 1). The biosynthesis from

inoleic and �-linolenic acid have been extensively studiedver the past 2 decades [4]. This fatty acid is present in alluminant fats and is partially converted in conjugated linoleiccid in human [7]. Milk is the main dietary source of vaccenic

tgAi

able 2nalysis of vaccenic (trans-11 18:1) acid by different techniques: results of the truen

ample Ag-TLC (g/100 g of FAs) Direct GLC (g/100 g of F

3.00 3.222.90 3.222.70 2.892.50 2.961.10 1.321.10 1.261.80 2.071.80 2.061.20 1.50

0 1.20 1.511 1.20 1.402 1.20 1.35

2 – 0.99SD (%) – 8.04EP – 0.27

g-TLC fractionation prior to high-resolution GLC was selected as the reference mas performed on replicate (n = 12). R2 indicates the coefficient of determination. RS

ndicated the standard error of prediction.

ethod. Milk samples (n = 6) were analyzed in duplicate and trueness analysisD (%) indicates the residual standard deviation expressed in percentage. SEP

cid and the measurement by different techniques of thisompound in liquid milk has been investigated in the presenttudy.

As previously done for total trans-18:1 isomers, alternativeethodologies to Ag-TLC followed by high-resolution GLC

ave been analyzed. The data, reported in Table 2, indicateslearly that TAG formula is less accurate than direct GLC andPLC fractionation prior to GLC analysis. Determination ofaccenic acid by direct measurement is accurate (R2 = 0.99).he standard error of prediction value (0.27) is higher than forPLC fractionation followed by GLC (0.09), but the truenessarameters indicate that this method is valid to measure

he level of vaccenic acid without prior pre-separation ofeometrical isomers of octadecenoic acid. Compared to theg-TLC technique, the pre-separation of trans- and cis-18:1

somers by HPLC is a good alternative (see Table 2). The

ess-regression analysis

As) RP-HPLC (g/100 g of FAs) TAG (g/100 g of FAs)

3.16 2.652.89 2.662.60 2.322.77 2.221.04 1.251.06 0.931.76 2.271.77 2.651.21 1.441.20 1.331.11 1.581.09 1.52

0.99 0.749.51 33.200.09 0.39

ethod. Milk samples (n = 6) were analyzed in duplicate and trueness analysisD (%) indicates the residual standard deviation expressed in percentage. SEP

F. Destaillats et al. / J. Chromatogr. A 1145 (2007) 222–228 227

Table 3trans-18:1 Acid isomeric distribution obtained by analyzing FAMEs fractionated by Ag-TLC or HPLC prior to high-resolution GLC analysis (mean value of duplicate)

trans-18:1 Sample A Sample B Sample C Sample D Sample E Sample F

HPLC Ag-TLC HPLC Ag-TLC HPLC Ag-TLC HPLC Ag-TLC HPLC Ag-TLC HPLC Ag-TLC

�4 0.00 0.01 0.00 0.02 0.00 0.01 0.00 0.02 0.00 0.02 0.00 0.01�5 0.00 0.01 0.00 0.03 0.00 0.01 0.00 0.02 0.00 0.02 0.00 0.01�6/8 0.20 0.12 0.20 0.13 0.20 0.13 0.22 0.06 0.21 0.13 0.19 0.08�9 0.17 0.17 0.17 0.15 0.18 0.18 0.21 0.15 0.19 0.18 0.18 0.16�10 0.15 0.17 0.17 0.22 0.22 0.23 0.24 0.26 0.27 0.28 0.22 0.25�11 3.02 2.99 2.68 2.62 1.05 1.08 1.77 1.77 1.21 1.20 1.10 1.20�12 0.27 0.25 0.26 0.26 0.30 0.28 0.33 0.33 0.28 0.26 0.27 0.27�13/14 0.79 0.69 0.65 0.65 0.80 0.73 0.85 0.80 0.62 0.55 0.60 0.58�15 0.24 0.26 0.21 0.27 0.27 0.28 0.29 0.31 0.21 0.22 0.20 0.23�16 0.33 0.32 0.30 0.31 0.34 0.34 0.37 0.38 0.27 0.26 0.27 0.29

T .27 4.27 4.10 3.26 3.11 3.03 3.08

A

dmitfiae

3

tTdaficmt

itpibop(mttmcorlaS

Fig. 3. Average trans-18:1 acid isomeric distributions in milk samples (n = 6,results of duplicate analysis). Pre-fractionation by Ag-TLC (grey bars) or HPLC(black bars) was used to separate cis- and trans-18:1 isomers prior to high-rca

4

cmbTrpfuaftr

otal 5.17 5.00 4.65 4.66 3.36 3

ll the results given are expressed as g/100 g of fatty acids.

ifference in the quantification procedure is that stearic acidethyl ester is selected as an internal reference when Ag-TLC

s used, and palmitic acid methyl is used for calculation whenhe pre-separation is achieved by HPLC. For each case arst GLC analysis of total FAMEs should be performed tossess the absolute level of palmitic or stearic acid methylsters.

.3. Determination of trans-18:1 acid isomeric distribution

Assessing the isomeric distribution in milk fat is interestingo understand the biological process involved in formation ofFAs. Dietary polyunsaturated fatty acids present in ruminantiets are converted by anaerobic bacteria in trans-monoenoicnd saturated fatty acids [4]. Ruminant health status like milkat depression can lead to an alteration of the metabolism TFAsn the rumen [4]. The isomeric distribution of trans-18:1 acidould also be used to assess the quality (authenticity) of theilk fat. Milk fat contains a wide range of positional isomers of

rans-18:1 acid (see Fig. 1).To assess distribution of the positional isomers, trans-18:1

somers should be isolated before GLC analysis. The twoechniques suitable to obtain this pre-separation have been com-ared. The results of the absolute level of individual trans-18:1somers are reported in Table 3. The pre-fractionation of FAMEsy HPLC is easy to achieve, could be automatized, and onef the main advantages is that CLA isomers are consequentlyurified. However, trans-4/5 are not collected in the fractionsee Table 3 and Ref. [23]) and the values for trans-6/8 iso-ers are superior compared to Ag-TLC fractionation. Therefore,

he separation is not suitable to achieve the separation of allhe positional isomers. However, these isomers (trans-4/8) are

inor in milk fat (less than 10% of total trans-18:1 acid isomerontent). The comparison of the average relative distributionf trans-18:1 isomers in the liquid milk samples (see Fig. 3)

eflects the slight difference previously mentioned. Neverthe-ess, the HPLC fractionation of cis- and trans-18:1 isomers isvaluable alternative to Ag-TLC (R2 = 0.99, RSD% = 4.49 andEP = 0.045).

otba

esolution gas-liquid chromatography analysis (see Section 2 for analyticalonditions). Error bars are ±standard deviations calculated on six samplesnalyzed in duplicate.

. Conclusion

The analysis of trans-isomers of octadecenoic acid in milk fatan be achieved by several methodologies. Accurate measure-ent of total trans-18:1 isomers in milk fat can be performed

y direct GLC under optimal chromatographic conditions [24].his methodology could be used on a routine basis to analyzed

aw materials or lipid extracted from food products. In addition,revious studies have demonstrated that the methylation of theatty acids can be performed directly on liquid or dairy prod-cts without prior fat extraction [24]. The analysis of vacceniccid, the main trans-18:1 isomers in milk fat could also be per-ormed by direct measurement [24]. Comparison of this methodo a reference method and HPLC fractionation confirmed theseesults. The TAG formulae could be used to assess the content

f trans-18:1 isomers but the accuracy is lower than the otherested methodologies [25,26]. The analysis of isomeric distri-ution of trans-18:1 acids could not be achieved by direct GLCnalysis since a fractionation step is required. HPLC could be

2 mato

uwmt

A

l

R

[

[

[

[

[[[

[[[[[

[[[

[

28 F. Destaillats et al. / J. Chro

sed to perform this pre-separation [23] and the obtained resultsere comparable to results obtained with the Ag-TLC (referenceethod). However, HPLC fractionation does not allow minor

rans-18:1 isomers, such as trans-4/5 18:1 isomers to be isolated.

cknowledgement

The authors would like to acknowledge Dr. Sebastien Popu-aire for his assistance in data interpretation and statistics.

eferences

[1] T. Leth, H.G. Jensen, A.Æ. Mikkelsen, A. Bysted, Atherosclerosis Suppl.7 (2006) 53.

[2] J. Moss, Atherosclerosis Suppl. 7 (2006) 57.[3] M.C. Craig-Schmidt, Atherosclerosis Suppl. 7 (2006) 1.[4] D.E. Bauman, J.M. Griinari, Annu. Rev. Nutr. 23 (2003) 203.[5] A.L. Lock, P.W. Parodi, D.E. Bauman, Aus. J. Dairy Technol. 60 (2005)

134.[6] F. Destaillats, E. Buyukpamukcu, P.A. Golay, F. Dionisi, F. Giuffrida, J.

Dairy Sci. 88 (2005) 449.

[7] R.O. Adlof, S.M. Duval, E.A. Emken, Lipids 35 (2000) 131.[8] F. Ulberth, M. Henninger, J. Dairy Res. 61 (1994) 517.[9] C. Cruz-Hernandez, Z. Deng, J. Zhou, A.R. Hill, M.P. Yurawecz, P. Del-

monte, M.M. Mossoba, M.E. Dugan, J.K. Kramer, J. AOAC Int. 87 (2004)545.

[

[[

gr. A 1145 (2007) 222–228

10] J.K.G. Kramer, C. Cruz-Hernandez, J. Zhou, Eur. J. Lipid Sci. Technol.103 (2001) 600.

11] S. Momchilova, B. Nikolova-Damyanova, W.W. Christie, J. Chromatogr.A 793 (1998) 275.

12] I. Marekov, R. Tarandjiiska, S. Panayotova, N. Nikolova, J. Planar Chro-matogr. 14 (2001) 384.

13] J. Molkentin, D. Precht, Kieler Milchwirtsch. Forschungsber. 56 (2004)53.

14] D. Precht, J. Molkentin, Eur. J. Lipid Sci. Technol. 102 (2000) 635.15] W.M.N. Ratnayake, J. AOAC Int. 87 (2004) 523.16] S. Satchithanandam, C.J. Oles, C.J. Spease, M.M. Brand, M.P. Yurawecz,

J.I. Rader, Lipids 39 (2004) 11.17] H. Steinhart, K. Winkler, R. Rickert, Ol. Corps Gras Lipides 8 (2001) 29.18] D. Precht, J. Molkentin, Kieler Milchwirtsch. Forschungsber. 51 (1999) 63.19] D. Precht, J. Molkentin, F. Destaillats, R.L. Wolff, Lipids 36 (2001) 827.20] R.L. Wolff, J. Am. Oil Chem. Soc. 72 (1995) 259.21] R.L. Wolff, N.A. Combe, F. Destaillats, C. Boue, D. Precht, J. Molkentin,

B. Entressangles, Lipids 35 (2000) 815.22] W.W. Christie, J. Lipid Res. 30 (1989) 1471.23] P. Juaneda, J. Chromatogr. A 954 (2002) 285.24] P.A. Golay, F. Dionisi, B. Hug, F. Giuffrida, F. Destaillats, Food Chem. 101

(2006) 1115.25] D. Precht, J. Molkentin, Int. Dairy J. 6 (1996) 791.

26] F. Destaillats, M. de Wispelaere, F. Joffre, P.A. Golay, B. Hug, F. Giuffrida,

F. Dionisi, J. Chromatogr. A 1131 (2006) 227.27] R.H. Thompson, J. Chromatogr. Sci. 35 (1997) 536.28] B. Nikolova-Damyanova, in: W.W. Christie (Ed.), Advances in Lipid

Methodology, Oily Press, Ayr, UK, 1992, p. 181.