Qelc

PD

a

ARR1AA

KHDRRF

1

fgsdtlRrfAdod

0h

Journal of Chromatography A, 1275 (2013) 1 8

Contents lists available at SciVerse ScienceDirect

Journal of Chromatography A

jou rn al h om epage: www.elsev ier .com/ locat e/chroma

uantification of -carotene, retinol, retinyl acetate and retinyl palmitate innriched fruit juices using dispersive liquidliquid microextraction coupled toiquid chromatography with fluorescence detection and atmospheric pressurehemical ionization-mass spectrometry

ilar Vinas, Mara Bravo-Bravo, Ignacio Lpez-Garca, Manuel Hernndez-Crdoba

epartment of Analytical Chemistry, Faculty of Chemistry, Regional Campus of International Excellence Campus Mare Nostrum, University of Murcia, E-30071 Murcia, Spain

r t i c l e i n f o

rticle history:eceived 2 November 2012eceived in revised form0 December 2012ccepted 11 December 2012vailable online 19 December 2012

eywords:PLCAPCI-MSispersive liquidliquid microextraction-Caroteneetinol

a b s t r a c t

A detailed optimization of dispersive liquidliquid microextraction (DLLME) was carried out for devel-oping liquid chromatographic (HPLC) techniques, using both fluorescence and atmospheric pressurechemical ionization mass spectrometric (APCI-MS) detection, for the simultaneous analysis of preformsof vitamin A: retinol (R), retinyl acetate (RA), retinyl palmitate (RP) and -carotene (-C). The HPLC anal-yses were carried out using a mobile phase composed of methanol and water, with gradient elution. TheAPCI-MS and fluorescence spectra permitted the correct identification of compounds in the analyzedsamples. Parameters affecting DLLME were optimized using 2 mL of methanol (disperser solvent) con-taining 150 L carbon tetrachloride (extraction solvent). The precision ranged from 6% to 8% (RSD) andthe limits of detection were between 0.03 and 1.4 ng mL1, depending on the compound. The enrichmentfactor values were in the 2144 range. Juice samples were analyzed without saponification and no matrixeffect was found when using fluorescence detection, so calibration was possible with aqueous standards.etinol estersruit juices

However, a matrix effect appeared with APCI-MS, in which case it was necessary to apply matrix-matchedcalibration. There was great variability in the forms of vitamin A present in the juices, the most abundantester being retinyl acetate (0.04 to 3.4 g mL1), followed by the amount of retinol (0.01 to 0.16 g mL1),while retinyl palmitate was not detected, except in the milk-containing juice, in which RP was the mainform. The representative carotenoid -carotene was present in the orange, peach, mango and multifruit

he mejuices in high amounts. T

. Introduction

Vitamin A includes a group of compounds which are requiredor vision, bone growth, reproduction and cell differentiation. Ineneral, there are two categories, depending on whether the foodource is an animal or a vegetal [1]. The vitamin A found in animal-erived foods is called preformed vitamin A, which is absorbed inhe form of retinol, a fat-soluble active substance. Sources includeiver, whole milk, eggs, meat and some fortified food products.etinol can be transformed into other active forms (retinal andetinoic acid) in the body. Since retinol is unstable, the vitamin isound in tissues as retinyl acetate or retinyl palmitate. The vitamin

found in plant-derived foods comes in the form of carotenoids,

ark-colored dyes (pigments) that may be transformed into a formf vitamin A [2]. One such carotenoid is -carotene, an antioxi-ant that protects cells from damage caused by free radicals [3].

Corresponding author. Tel.: +34 868887406; fax: +34 868887682.E-mail address: hcordoba@um.es (M. Hernndez-Crdoba).

021-9673/$ see front matter 2012 Elsevier B.V. All rights reserved.ttp://dx.doi.org/10.1016/j.chroma.2012.12.022thod was validated using two certified reference materials. 2012 Elsevier B.V. All rights reserved.

The more intense the color of a fruit or vegetable, the higherthe -carotene content. Moreover, vegetable sources are fat- andcholesterol-free [4]. -Carotene is less easily absorbed than retinoland must be converted into retinal and retinol by the body. Manysupplements provide a combination of retinol and -carotene.

In recent years, many nutrients have been added to foods anddrinks to cover dietetic recommendations and to solve deficien-cies in a specific nutrient [5,6]. Although such functional products,including enriched foods, are designed to improve human health[7], many problems due to possible interactions or imbalanceshave been described [8]. Enrichment with vitamins involves fewerprocessing problems that relate to enrichment with minerals orproteins because lower amounts are required. Thus, functionaldrinks, known as ACE juices, are soft drinks based on fruit and veg-etable juices to which the antioxidant vitamins A, C and E havebeen added. However, vitamin loss during manufacture and storage

means that substantial amounts of the nutrient need to be addedif the consumer is to receive the required levels. The EU Commis-sion [9] has established that vitamin A in vegetable juices must bepresent in a final content of no lower than 25 g RE/100 kJ (100 g

dx.doi.org/10.1016/j.chroma.2012.12.022http://www.sciencedirect.com/science/journal/00219673http://www.elsevier.com/locate/chromamailto:hcordoba@um.esdx.doi.org/10.1016/j.chroma.2012.12.022

2 matog

RIrma

acclfihwmfad

ttsii

iattdlesaindtsu

dwaracofigaomc

2

2

cwBtts

P. Vinas et al. / J. Chro

E/100 kcal), RE being the equivalents of all-trans-retinol (annexI). The compounds permitted to be added as vitamin A are retinol,etinyl acetate, retinyl palmitate and -carotene (annex IV). Theaximum permitted content of vitamin A in cereal-based productsnd baby foods is 180 g RE/100 kcal (annex VI).Classical methods of vitamin A analysis include saponification

nd/or solvent extraction in low polarity organic solvents, as it is thease with the standardized analysis method for retinol [10] and -arotene [11]. However, these methods are tedious and also requirearge amounts of organic solvents, as the samples must be saponi-ed using 50150 mL methanol, and then extracted 34 times withigh volumes of organic solvents. The combined extracts were thenashed, evaporated and analyzed by normal-phase liquid chro-atography (NP-HPLC) for retinol [10] and by reversed-phase HPLC

or -carotene [11]. Moreover, the saponification process convertsll the esters into the free form of retinol, so that it is unsuitable forifferentiating the different forms of vitamin A.On the other hand, several studies have tackled the determina-

ion of vitamin A in foods or biological samples by HPLC coupledo mass spectrometry (MS) using particle beam [12], electro-pray ionization (ESI) [1319] and atmospheric pressure chemicalonization (APCI) [2030] in this way allowing selectivity to bemproved.

To determine the different forms of vitamin A in food samples,t is necessary to include isolation and/or preconcentration steps tovoid interference from the complex food matrix. Microextractionechniques [31,32] are a way to miniaturize the analytical labora-ory [33], the selective extraction of compounds being based onifferences in their physicalchemical characteristics. Dispersiveiquidliquid microextraction (DLLME) is a very simple and rapidxtraction method, based on the use of a ternary component solventystem [34]. Speed and the low consumption of organic solventsre two of the main advantages of this technique, which can bencluded in the group of clean chemistry procedures. As regardsew green sample preparation techniques, only one study [35] hasescribed the determination of cis- and trans-retinol isomers usinghe coupling of DLLME-LC-fluorescence after saponification of theample to convert all forms of the vitamin A into retinol beforesing NP-HPLC.In the present study, two miniaturized sample treatment proce-

ures based on DLLME coupled to a reversed-phase HPLC techniqueith two different detection systems, fluorescence and APCI-MS,re proposed for determining retinol (R), retinyl acetate (RA),etinyl palmitate (RP) and -carotene (-C). The procedures werepplied to the determination of preforms of vitamin A and -arotene in different types of enriched ACE fruit juices. The noveltyf the present approach is based on the use of DLLME for therst time for the determination of these forms of vitamin A usingreen chemistry principles coupled to liquid chromatography,voiding the use of high amounts of solvents and the generationf residues. The sensitivity together with the selectivity of theass spectra using APCI-MS permitted the correct identification ofompounds.

. Experimental

.1. Reagents

Chromatographic quality methanol, ethanol and carbon tetra-hloride were obtained from Sigma (St. Louis, MO, USA). Theater used was previously purified in a Milli-Q system (Millipore,

edford, MA, USA). All-trans-retinol, retinyl acetate, retinyl palmi-ate and -carotene were obtained from SigmaAldrich. Stock solu-ions (1000 g mL1) of R, RA and RP were prepared in ethanol andtored in amber vials at 20 C, while the stock solution of -C wasr. A 1275 (2013) 1 8

prepared in chloroform. Working standard solutions were prepareddaily in ethanol.

2.2. Instrumentation for HPLC-fluorescence

The HPLC-fluorescence system consisted of an Agilent 1100(Agilent, Waldbronn, Germany) quaternary pump (G1311A) oper-ating at a flow-rate of 0.9 mL min1. The solvents were degassedusing an on-line membrane system (G1379A). The fluorescencedetector (G1321A) was operating at an excitation wavelength of325 nm and an emission wavelength of 480 nm for R, RA andRP, and at an excitation wavelength of 450 nm and an emis-sion wavelength of 530 nm for -C. The analytical column usedfor the reversed-phase technique was a Zorbax Eclipse XDB-C8 (15 cm 46 mm 5 m). The mobile phase was a mixtureof methanol and water, operating under gradient elution withthe following optimized program: linear gradient from 90:10methanol:water (v/v) to 100% methanol in 15 min and held for8 min. Finally, the initial conditions were re-established in 1 minand held for 10 min. Aliquots of 20 L were injected manually usinga Model 7125-075 Rheodyne injection valve (Rheodyne, Berkeley,CA, USA). Solutions were stored in 2 or 10 mL amber glass vials. Tofilter the samples, PVDF filters (0.45 m) (Teknokroma, Barcelona,Spain) were used. An EBA 20 (Hettich, Tuttlingen, Germany) cen-trifuge was used at a speed near to the maximum supported by theconical glass tubes, 3000 rpm.

2.3. HPLCAPCI-MS system

The HPLC system consisted of an Agilent 1200 binary pump(G1312A) operating at a flow-rate of 0.9 mL min1. The solventswere degassed using an on-line membrane system (G1379A). Thecolumn was maintained in a thermostated compartment at roomtemperature (G1316A). The injection (20 L) was performed usingan autosampler (G1329A). The column and the gradient programfor the mobile phase were the same as those optimized for fluo-rescence detection. The HPLC system was coupled to an ion-trapmass spectrometer (1036 model) equipped with an APCI inter-face operating in positive ion mode. Selected ion monitoring (SIM)mode using the ion m/z 269 from 0 to 21.5 min (R, RA and RP), andthe ion m/z 539 from 21.5 to 26 min (-C) was applied, accord-ing to previous studies [18,20,2527]. The instrument parameterswere drying temperature, 350 C; APCI temperature, 400 C; dry-ing gas flow, 5 L/min and nebulizer gas pressure, 60 psi. Calibrationwas performed using matrix-matched standards for APCI-MS as themost effective way to compensate for the influence of the matrix onmethod performance. Statistical analysis was carried out by apply-ing the ANOVA test and a multiple comparisons test versus controlgroup.

2.4. Samples and certified reference materials

The samples were different commercial types of juicesenriched with vitamin A and E (ACE juices) containing orange,grapefruit, peach, pineapple, banana, pear, applemango,applebananaorange, multifruit and milk-containing fruitjuices. Samples were analyzed in triplicate. The method wasvalidated using two reference materials: infant/adult nutritional

formula SRM 1849a supplied by the National Institute of Standardsand Technology (NIST) and whole milk powder ERM-BD600 sup-plied by the Institute for Reference Materials and Measurements(IRMM).

matogr. A 1275 (2013) 1 8 3

2

wTctgcte3twTu

3

3

wopcpEf(n

fCwmatwweaafh1ta(

3

distgftf(ToR

Fig. 1. Effect of the extractant solvent on the peak area of the vitamin A formsin a diluted juice sample by DLLME-HPLC-fluorescence and on the volume of theorganic drop sedimented after centrifugation of the dispersion. Conditions: 200 L

ion monitoring (SIM) mode using the ion m/z 269 from 0 to 21.5 min(R, RA and RP), and the ion m/z 539 from 21.5 to 26 min (-C) wasapplied.P. Vinas et al. / J. Chro

.5. DLLME procedure

For DLLME, a 0.12 mL aliquot of juice was diluted to 10 mLith water in a 15 mL screw cap glass tube with conical bottom.hen, 2 mL of methanol (dispersive solvent) containing 150 L ofarbon tetrachloride (extractant solvent) was rapidly injected intohe water solution using a micropipette, and the mixture was againently shaken manually for several seconds. A cloudy solutiononsisting of very fine droplets of carbon tetrachloride dispersedhrough the sample solution was formed, and the analytes werextracted into the fine droplets. After centrifugation for 2 min at000 rpm, the extraction solvent was sedimented at the bottom ofhe conical tube (volume about 50 10 L). The sedimented phaseas collected and evaporated to dryness with a nitrogen stream.he residue was reconstituted with 50 L of methanol, and a vol-me of 20 L was injected into the HPLC.

. Results and discussion

.1. Chromatographic separation

The use of both normal-phase and reversed phase techniquesas compared for improving the resolution of the different formsf vitamin A using fluorescence detection in both cases. The normal-hase technique was tried using a LiChrospher Si 60 analyticalolumn and a mobile phase consisting of a 98:2 (v/v) n-hexane:2-ropanol mixture at a flow-rate of 2 mL min1, according to theuropean Standard [10]. Under these conditions, the retention timeor all-trans-retinol was 5.8 min. However, the other compoundsthe esters RA and RP, and -carotene) were not retained, elutingear the void time.The reversed phase technique was assayed using a C8 and dif-

erent C18 columns. Retention of the esters was too high with the18 packing, and consequently the Zorbax Eclipse XDB-C8 columnas selected. Several mobile phases corresponding to mixtures ofethanol or acetonitrile and water in different percentages weressayed. Isocratic elution using a 90:10 (v/v) methanol:water mix-ure led to retinol being eluted with a retention factor of near 3,hile the esters eluted with very high retention factors and -Cas retained. When the methanol proportion was increased, retinolluted at the void time. Consequently, gradient elution was triednd, after several experiments with different solvent proportionsnd ramps, the optimized gradient program was linear gradientrom 90:10 methanol:water (v/v) to 100% methanol in 15 min, andeld for 8 min. Finally, the initial conditions were re-established in

min and held for 10 min. The flow-rate was 0.9 mL min1. Withhis program, the order of elution and the retention factors for thenalytes were: R (tR = 5.2 min, k = 2.7); RA (tR = 7.0 min, k = 4.0); RPtR = 20.3 min, k = 13.5) and -C (tR = 22.4 min, k = 15).

.2. APCI-MS detection

The ionization mode selected in HPLCMS measurementsepends on the polarity of the target compounds. APCI, a soft ion-zation technique used for small compounds of low polarity, waselected. Moreover, signal intensity is dependent on the volatility ofhe compounds, and vitamin A forms are neutral and non-volatile,iving high signals in APCI. A full scan of each analyte was per-ormed using the above-described chromatographic conditions andhe experiments in APCI(+) showed that the protonated positive ionragment m/z 269 was the most abundant for the three compounds

R, RA and RP), while no protonated molecular ions were observed.his ion reflects the loss of water from the protonated moleculef retinol, loss of acetic acid from RA and loss of palmitic acid fromP. However, the protonated molecule [MH]+ (m/z 539) was moreextraction solvent, 2 mL methanol disperser solvent, 20 ng mL1 concentration ofanalytes.

intense and had a higher signal-to-noise (S/N) ratio for -C. SelectedFig. 2. Influence of the extractant carbon tetrachloride volume (A) and the dispersermethanol volume (B) on the peak area of vitamin A forms in a diluted juice sampleby DLLME-HPLC-fluorescence and on the volume of the organic drop sedimentedafter centrifugation of the dispersion. Concentration of analytes 20 ng mL1.

4 P. Vinas et al. / J. Chromatogr. A 1275 (2013) 1 8

F LCAi ach o

3

itagdc

advovatw2p

ig. 3. Elution profile obtained using DLLME-HPLC-fluorescence (A) and DLLME-HPn the selected conditions. Positive-ion APCI mass spectra of the extracted ions for e

.3. DLLME parameters

The microextraction step was optimized by studying the exper-mental variables which affect the extraction efficiency, such ashe type and volume of both extraction and disperser solvents, theddition of salt, pH and both the speed and time of the centrifu-ation step. For this purpose, an aqueous standard solution or ailuted juice sample (1 mL juice and 9 mL water) containing analyteoncentrations of about 20 ng mL1 was used.

The extraction solvent must have high affinity for the analytes, higher density than water and low solubility in water, while alsoisplaying good chromatographic behavior. In addition, the sol-ent must have a low boiling temperature because the sedimentedrganic phase must be evaporated and reconstituted using a sol-ent compatible with reversed-phase HPLC. Thus, chlorobenzenend tetrachloroethane, both of which have a high boiling tempera-

ure, were discarded. Well-defined settled volumes were recoveredith dichloroethane, chloroform and carbon tetrachloride, using00 L of the extraction solvent and 2 mL of methanol as the dis-erser solvent. Fig. 1 depicts the peak area values for the differentPCI-MS in SIM mode (B) for a standard solution of the different forms of vitamin Af the analytes (C).

compounds (mean of triplicate experiments). As it can be seen, thebest extraction efficiency was obtained using both chloroform andcarbon tetrachloride. On the other hand, Fig. 1 also shows the vol-ume obtained for the organic drop sedimented after centrifugation.As this volume was considerably higher when carbon tetrachloridewas used, this was selected as the extraction solvent.

Acetone, methanol and acetonitrile are miscible in the extrac-tion solvent and in the aqueous solution and so were assayed asdisperser solvents by rapidly injecting 2 mL of each disperser con-taining 200 L of CCl4 into 10 mL of the diluted juice sample. Theextraction efficiency was greatest with methanol, which had theadditional advantage of low toxicity, and was therefore selected.

The influence of the carbon tetrachloride volume was studiedbetween 50 and 250 L for a diluted juice sample, carrying outexperiments in triplicate. Volumes of between 50 and 60 L ledto very small volumes for the sedimented drop. Fig. 2A shows that

peak areas increased with increasing extraction solvent volumesup to 150200 L for the different analytes (values are mean oftriplicate experiments). On further increasing the volume of thesolvent, the sensitivity decreased as a consequence of dilution, and

P. Vinas et al. / J. Chromatogr. A 1275 (2013) 1 8 5

Table 1Calibration graphs by DLLME-HPLC-fluorescence.

Compound Slope SD (mL ng1) r2 Linearity (ng mL1) LOD (ng mL1) LOQ (ng mL1)All-trans-retinol 0.522 0.011 0.9956 1100 0.82 2.7Retinyl acetate 0.148 0.003 0.9949 1100 0.76 2.5Retinyl palmitate 0.084 0.003 0.9992 1100 0.96 3.2-Carotene 0.013 0.001 0.9956 1100 0.98 3.3

Table 2Calibration graphs by DLLME-HPLCAPCI-MS.

Compound Slope SD (mL ng1) r2 Linearity (ng mL1) LOD (ng mL1) LOQ (ng mL1)All-trans-retinol 8.33 104 0.61 104 0.9947 0.550 0.07 0.25Retinyl acetate 7.62 104 0.50 104 0.9935 0.550 0.08 0.26Retinyl palmitate 6.95 103 0.08 103 0.9997 2.550 1.4 4.8-Carotene 3.43 106 0.21 106 0.9943 0.150 0.03 0.09

Table 3Calibration slopes using DLLME-HPLCAPCI-MS for juice samples (mL ng1).

Juice sample R RA RP B-C

Aqueous standards 8.33 104 0.61 104 7.62 104 0.50 104 6.95 103 0.10 103 3.43 105 0.21 105Orange 5.16 104 0.50 104 5.30 104 0.62 104 2.98 103 0.16 103 1.51 105 0.10 105Multifruits 4.51 104 0.19 104 5.28 104 0.52 104 2.96 103 0.37 103 1.43 105 0.10 105

4 4 4 0 4 3 3 5 54 04 0

sofuaao

cjtstp1miw

itwTu

TV

V

Pear 4.82 10 0.29 10 4.33 10Orangebananaapple 4.86 104 0.10 104 4.68 10Fruit + milk 4.72 104 0.28 104 4.35 10

o a 150 L extractant volume was selected. Fig. 2B shows the effectf varying the volume of the disperser solvent on the sensitivityor a diluted juice sample (triplicate experiments). When the vol-mes assayed were below 1 mL, the dispersion was very deficientnd peak areas were very low. For higher volumes of between 1nd 2 mL, the sensitivity continuously increased and so a volumef 2 mL was selected.To study the effect of the ionic strength, the experiments were

arried out at different sodium chloride concentrations in diluteduice solutions, ranging from 0% to 30% (m/v). The results showedhat extraction efficiency decreased for all the compounds when thealt concentration was increased. Therefore, the addition of NaCl tohe extraction solution was discarded. To evaluate the influence ofH, different samples were adjusted to pH values of between 4 and0 with 0.01 M phosphate buffer solutions. The volume of the sedi-ented organic solvent did not vary and no significant differences

n the analytical signals were appreciated. Consequently, samplesere analyzed without pH adjustment.Extraction time in DLLME, defined as the interval between

njecting the mixture of disperser and extraction solvents and

he centrifugation step, had no influence on extraction efficiency,hich is one of the most important advantages of this technique.he mixture of extractant, disperser and the diluted juice was man-ally shaken for different times of between a few seconds and

able 4itamin A content in juice samples by DLLME-HPLCAPCI-MS (g mL1).

Juice sample All-trans-retinol Retinyl acetate -Carotene

Orange 0.048 0.005 0.73 0.08 1.70 0.20Pineapple 0.050 0.007 0.79 0.01 0.11 0.01Pear 0.056 0.006 1.50 0.15 0.17 0.014Banana 0.070 0.004 0.52 0.05 NDPeach 0.086 0.010 2.30 0.27 0.89 0.16Grapefruit ND 0.69 0.01 NDAppleorangebanana 0.021 0.003 0.77 0.13 0.087 0.005Applemango 0.080 0.007 2.90 0.12 2.30 0.27Multivitamins ND 0.074 0.006 11 1Multifruits ND 0.078 0.008 3.90 0.29Milkmultifruit 0.049 0.009 0.047 0.005 3.30 0.54alues are mean SD (n = 6)..36 10 2.78 10 0.15 10 1.51 10 0.20 10

.65 104 2.55 103 0.15 103 1.31 105 0.16 105

.57 104 2.83 103 0.10 103 1.25 105 0.10 105

5 min, before submitting the mixture to centrifugation. Similar peakareas were obtained in all cases, indicating that the DLLME proce-dure was practically time-independent. Consequently, only a fewseconds were needed to extract the analytes. The centrifugationtime and speed necessary to disrupt the cloudy solution and col-lect the sedimented phase were evaluated in the ranges 15 minand 5003000 rpm, respectively. Best results were attained bycentrifuging the mixture for 2 min at the maximum speed recom-mended for the conical glass tubes used, 3000 rpm. The volume ofthe sedimented phase was 50 10 L after extraction and centrifu-gation.

Fig. 3A shows the elution profile obtained using DLLME-HPLC-fluorescence and Fig. 3B shows the corresponding chromatogramobtained by DLLME-HPLCAPCI-MS in SIM mode for a standardsolution of the different forms of vitamin A in the selected con-ditions, as well as the mass spectra of the extracted ions for eachone of the analytes (Fig. 3C).

3.4. Analytical characteristics of the method

The method was validated for linearity, detection and quantifi-cation limits, selectivity, accuracy and precision. The calibrationgraphs for aqueous standards using DLLME-HPLC-fluorescencewere obtained by least-squares linear regression analysis ofthe peak area versus analyte concentration using ten levels(1100 ng mL1) in duplicate experiments. Quantification was per-formed by the external standard procedure. The results obtainedare summarized in Table 1. The r2 values were good (r2 > 0.994),demonstrating the excellent linearity for the range studied. Thelimits of detection (LOD, calculated for a signal-to-noise ratio of3) and the limits of quantification (LOQ, calculated for a signal-to-noise ratio of 10) are also shown in Table 1. The enrichmentfactor (EF) was calculated as the ratio between the slope obtainedby DLLME and the slope obtained by direct HPLC-fluorescence with-

out preconcentration. The slope values (mL ng1) for direct HPLCwere 0.0118, 0.00464, 0.00385 and 0.000516 for R, RA, RP and -C,respectively. Consequently, the EF values were 44.2, 31.9, 21.8 and25.6 for R, RA, RP and -C, respectively.

6 P. Vinas et al. / J. Chromatogr. A 1275 (2013) 1 8

I-MS d

urqo

TV

Fig. 4. Elution profiles obtained using DLLME-HPLC with APC

Calibration graphs for aqueous standards were also obtained bysing DLLME-HPLCAPCI-MS and the results obtained are summa-ized in Table 2. SIM of the fragment ion of m/z 269 was used for the

uantitative analysis of R, RA and RP, because it was the base peakf the positive ion APCI mass spectra of the three compounds. For-C, the APCI mass spectra show higher abundance of the intact

able 5itamin A content in certified reference materials.

Reference material DLLME-LCAPCI-MS (mg kg1)

All-trans-retinol Retin

Infant/adult nutritionalformula SRM 1849a

7.98

Whole milk powderERM-BD600

3.72 0.27 etection for peach, mangoapple and milkmultifruit juices.

protonated molecule with m/z 539. Table 2 shows the values ofr2, which again demonstrated good linearity for the range studied,as well as the LOD and the LOQ values, which were calculated as

previously.

The selectivity of the method was judged from the absenceof interfering peaks at the elution times of the vitamin A forms

Certified value (mg kg1)

yl palmitate Vitamin A

0.58 7.68 0.23 mg kg1 retinol equivalents, total(trans + cis)-retinol, added as retinyl palmitate3.8 mg kg1 all-trans-retinol and 4.1 mg kg1

all-trans-retinol and 13-cis-retinol

matog

fpsoo

dsRts

3

Fspavbaps

bilsacwastgetooapsmcp

3r

sMmwcAobtwuflfcp

P. Vinas et al. / J. Chro

or chromatograms of different non-spiked juices. No matrix com-ounds existed that might give a false positive signal in theamples. The selectivity was also improved by the high resolutionf HPLCAPCI-MS because no peaks at m/z 269 and m/z 539 werebserved in the non-spiked juices.The repeatability was calculated by using the relative standard

eviation (RSD) from a series of ten consecutive DLLME-LC analy-es of a juice sample spiked with all the analytes at 20 ng mL1. TheSD values were 7.6, 6.8, 6.1 and 8.6 for R, RA, RP and -C, respec-ively, values that indicate that the precision of the method wasatisfactory for control analysis.

.5. Study of the matrix effect

The matrix effect was first evaluated for the DLLME-HPLC-l procedure by comparing the slopes of aqueous standards andtandard additions calibration graphs for the different juice sam-les, obtained by plotting concentration (at six levels) against peakrea. A statistical study was carried out using one way analysis ofariance (ANOVA). The presence of a matrix effect was discardedecause the P values obtained were higher than 0.05 for all thenalytes. Consequently, calibration and analysis of all the juice sam-les using fluorescence detection were performed using aqueoustandards.

Nevertheless, quantitative analysis using APCI can be affectedy the occurrence of signal suppression due to unknown matrixnterferences [36]. This phenomenon affects the reproducibility,inearity and accuracy of the method and must be carefully con-idered. Matrix effects were studied in the different types of juicesnalyzed to evaluate the ion suppression produced by co-elutingompounds from the juice, which can affect analyte ionization. Thisas again evaluated by comparing the slopes of aqueous standardsnd standard additions calibration graphs for the different juiceamples (Table 3). The application of the ANOVA test showed thathere was no statistically significant difference between the juiceroups. However, when the juice groups were compared by ref-rence to the aqueous standard slopes, the ANOVA tests indicatedhe presence of statistically significant differences. The applicationf the multiple comparisons test versus control group using aque-us standards (HolmSidak method) revealed that P < 0.001 forll the comparisons. Consequently, for the DLLME-HPLCAPCI-MSrocedure, calibration must be performed using matrix-matchedtandards, as there were no differences between the different juiceatrices, and this technique proved to be the most effective way toompensate for the adverse influence of the matrix on the methodserformance.

.6. Analysis of juice samples and validation using certifiedeference materials

Juice samples containing different fruits were analyzed. Fig. 4hows the elution profiles obtained using DLLME-HPLC with APCI-S detection for three juice samples corresponding to peach,angoapple and milkmultifruits juices. Similar chromatogramsere obtained for the other samples. The absence of interferingompounds eluting at the retention times of the different vitamin

forms was checked by peak purity analysis and a comparisonf MS spectra. The criteria for confirmation were a concordanceetween the retention times, fluorescence spectra and MS spec-ra for the standards, non-spiked samples and the samples spikedith the standards. Table 4 shows the results obtained whensing APCI-MS detection. The contents were similar when the

uorescence detector was used. The values reported by the manu-acturers for the vitamin A content in juices were 1.2 g mL1. Theontents found in the juices analyzed decreased with time, com-ared with the contents found in freshly prepared juices. Similar

[

r. A 1275 (2013) 1 8 7

values were obtained for the vitamin A forms when using the pro-posed DLLME-HPLC procedure in both detection modes. The formsof vitamin A present in the juices varied greatly. The most abun-dant ester was found to be retinyl acetate, whose content rangedfrom 0.04 to 2.9 g mL1, while the amount of all-trans-retinolranged from 0.02 to 0.09 g mL1, and retinyl palmitate was notdetected, except in the juice containing milk, in which RP wasthe main form (7.90 1.3 g mL1). Carotenoid -carotene waspresent in the orange, peach, mango and multifruit juices at highconcentrations.

Finally, the reliability of the method was further checked byusing two certified reference materials, infant/adult nutritional for-mula SRM 1849a and whole milk powder ERM-BD600. Table 5shows the results obtained. The values obtained by the proposedDLLME-LCAPCI-MS method were in excellent agreement withthe certified contents. The statistical study using the paired t-testshowed that there was no significant difference (95% confidenceinterval) between the results obtained and the certified values (theP value obtained was 0.666). Such data also confirm the efficacyof the extraction procedure for recovering both free supplementedand endogenous vitamin A in juices.

4. Conclusion

The use of a miniaturized preconcentration procedure based onDLLME coupled to HPLC permits low detection limits because ofthe high enrichment power of DLLME, with the additional advan-tage of using very low quantities of solvents. The combination ofMS data with retention times, the fragmentation patterns of R,RA, RP and -C obtained by APCI, and the fluorescence spectramade peak identification very reliable. The practical applicabilityof the method was demonstrated by the quantitative analysis ofcommercially available fruit juices to provide reliable chromato-graphic fingerprints of their vitamin A forms, avoiding the tedioussaponification step. The validation procedure indicated that thismethod affords reliable analysis and is appropriate for vitamin Aquality control of complex matrices using the new green samplepreparation techniques.

Acknowledgments

The authors acknowledge the financial support of the Comu-nidad Autnoma de la Regin de Murcia (CARM, Fundacin Sneca,Project 15217/PI/10), the Spanish MICINN (Project CTQ2009-08267/BQU) and Hero Espana, S.A. M. Bravo-Bravo acknowledges afellowship from Fundacin Sneca, CARM.

References

[1] P. Sauvant, M. Cansell, A. Abdessattar, H. Sassi, C. Atgi, Food Res. Int. 46 (2012)469.

[2] E. Fernndez-Garca, I. Carvajal-Lrida, M. Jarn-Galn, J. Garrido-Fernndez, A.Prez-Glvez, D. Hornero-Mndez, Food Res. Int. 46 (2012) 438.

[3] P.T. Gardner, T.A.C. White, D.B. McPhail, G.G. Duthie, Food Chem. 68 (2000) 471.[4] F. Granado, B. Olmedilla, I. Blanco, E. Rojas-Hidalgo, J. Agric. Food Chem. 40

(1992) 2135.[5] K. Brandt, L.P. Christensen, J. Hansen-Moller, S.L. Hansen, J. Haraldsdottir, L.

Jespersen, S. Purup, A. Kharazmi, V. Barkholt, H. Frokiaer, M. Kobaek-Larse,Trends Food Sci. Technol. 15 (2004) 384.

[6] C. Snchez-Moreno, L. Plaza, B. de Ancos, P. Cano, J. Sci. Food Agric. 83 (2003)430.

[7] F.A. Toms-Barbern, Aliment. Nutr. Salud 10 (2003) 41.[8] M.M. Donma, O. Donma, Food Res. Int. 38 (2005) 681.[9] Commision Directive 1996/5/EC, of 16 February 1996 on processed cereal-based foods and baby foods for infants and young children (1996).10] European Standard CSN EN 12823-1, Foodstuffs Determination of Vitamin

A by High Performance Liquid Chromatography Part 1: Measurement ofAll-trans-retinol and 13-Cis-Retinol, European Committee for Standardization,Brussels, 2000.

8 matog

[

[

[[[

[

[

[[[

[

[

[

[

[[

[

[

[

[[[

[

[

P. Vinas et al. / J. Chro

11] European Standard CSN EN 12823-2, Foodstuffs Determination of VitaminA by High Performance Liquid Chromatography Part 2: Measurement of -Carotene, European Committee for Standardization, Brussels, 2000.

12] R. Andreoli, M. Careri, P. Manini, G. Mori, M. Musci, Chromatographia 44 (1997)605.

13] R.B. van Breemen, C.R. Huang, FASEB J. 10 (1996) 1098.14] C. Rentel, S. Strohschein, K. Albert, E. Bayer, Anal. Chem. 70 (1998) 4394.15] A. Lienau, T. Glaser, M. Krucker, D. Zeeb, F. Ley, F. Curro, K. Albert, Anal. Chem.

74 (2002) 5192.16] F. Priego Capote, J. Ruiz Jimnez, J.M. Mata Granados, M.D. Luque de Castro,

Rapid Commun. Mass Spectrom. 21 (2007) 1745.17] T. Huang, W. Zhang, J. Liu, Y. Tian, G. Yang, J. Liq. Chromatogr. Related Technol.

32 (2009) 712.18] Y.O. Lim, B. Kim, S. Ahn, J. Kim, Food Chem. 126 (2011) 1393.19] O. Midttun, P.M. Ueland, Rapid Commun. Mass Spectrom. 25 (2011) 1942.20] R.B. van Breemen, D. Nikolic, X. Xu, Y. Xiong, M. van Lieshout, C.E. West, A.B.

Schilling, J. Chromatogr. A 794 (1998) 245.21] Y. Wang, X. Xu, M. Van Lieshout, C.E. West, J. Lugtenburg, A.F. Verhoeven-Creemers, D.P. Muhilal, R.B. Van Breemen, Anal. Chem. 72 (2000) 4999.22] Y. Wang, W.Y. Chang, G.S. Prins, R.B.J. van Breemen, J. Mass. Spectrom. 36 (2001)

882.23] P. McCaffery, J. Evans, O. Koul, A. Volpert, K. Reid, M.D.J. Ullman, Lipid Res. 43

(2002) 1143.

[

[

r. A 1275 (2013) 1 8

24] C.W. Huck, M. Popp, H. Scherz, G.K. Bonn, J. Chromatogr. Sci. 38(2000) 441.

25] O. Heudi, M.J. Trisconi, C.J. Blake, J. Chromatogr. A 1022 (2004) 115.26] R. Andreoli, P. Manini, D. Poli, E. Bergamaschi, A. Mutti, W.M.A. Niessen, Anal.

Bioanal. Chem. 378 (2004) 987.27] D. Zhu, Y. Wang, Y. Pang, A. Liu, J. Guo, C.A. Bouwman, C.E. West, R.B. van

Breemen, Rapid Commun. Mass Spectrom. 20 (2006) 2427.28] M. Kamao, N. Tsugawa, Y. Suhara, A. Wada, T. Mori, K. Murata, R. Nishino, T.

Ukita, K. Uenishi, K. Tanaka, T. Okano, J. Chromatogr. B 859 (2007) 192.29] T.E. Gundersen, N.E. Bastani, R. Blomhoff, Rapid Commun. Mass Spectrom. 21

(2007) 1176.30] A. Gentili, F. Caretti, J. Chromatogr. A 1218 (2011) 684.31] L. Ramos, J.J. Ramos, U.A.Th. Brinkman, Anal. Bioanal. Chem. 381 (2005) 119.32] C. Nern, J. Salafranca, M. Aznar, R. Batlle, Anal. Bioanal. Chem. 393

(2009) 809.33] S. Mitra (Ed.), Sample Preparation Techniques in Analytical Chemistry, Wiley-

Interscience, New Jersey, 2003.34] M. Rezaee, Y. Assadi, M.R.M. Hosseini, E. Aghaee, F. Ahmadi, S. Berijani, J. Chro-matogr. A 1116 (2006) 1.35] P. Vinas, M. Bravo-Bravo, I. Lpez-Garca, M. Hernndez-Crdoba, Talanta 103

(2013) 166.36] H. Trufelli, P. Palma, G. Famiglini, A. Cappiello, Mass Spectrom. Rev. 30 (2011)

491.

Quantification of -carotene, retinol, retinyl acetate and retinyl palmitate in enriched fruit juices using dispersive liq...1 Introduction2 Experimental2.1 Reagents2.2 Instrumentation for HPLC-fluorescence2.3 HPLCAPCI-MS system2.4 Samples and certified reference materials2.5 DLLME procedure

3 Results and discussion3.1 Chromatographic separation3.2 APCI-MS detection3.3 DLLME parameters3.4 Analytical characteristics of the method3.5 Study of the matrix effect3.6 Analysis of juice samples and validation using certified reference materials

4 ConclusionAcknowledgmentsReferences