v86n2p367.pdf

FOOD COMPOSITION AND ADDITIVES

Determination of Total Vitamin C in Fruit Juices and RelatedProducts by Liquid Chromatography: Interlaboratory StudyALLAN R. BRAUSE

Analytical Chemical Services of Columbia Inc., 9110 Red Branch Rd, Columbia, MD 21045DAVID C. WOOLLARD

1

AgriQuality NZ Ltd., PO Box 41, Auckland, New ZealandHARVEY E. INDYK

NZMP-Fonterra, PO Box 7, Waitoa, New Zealand

Collaborators: J. Acar; K. Adadevoh; G. Cherix; B. Durst; T. Eisele; E. Elkins; J. Foos; S. Hammack; D. Hammond;F. Hesford; C. Hischenhuber; V. Hong; C.J. Huang; S. Kirksey; L. Kline; D. Kruger; M.J. Lawson; A. Lea; G. Martin;A. Parkih; J. Weiss; E. Wilhelmsen; B. Woodward; R. Wrolstad; L. Zygmunt

A interlaboratory study was conducted to evaluatea liquid chromatographic (LC) procedure for thedetermination of total vitamin C in foods at levelsof 5–60 mg/100 g. Emphasis was placed on fruitjuices, although selected foods were also includedin the study. Following dissolution of sample in wa-ter, endogenous dehydroascorbic acid was con-verted to ascorbic acid by precolumn reduction withdithiothreitol at neutral pH. Total ascorbate was de-termined by C18 reversed-phase LC with a phos-phate eluent at pH 2.5, incorporating dithiothreitol tomaintain vitamin C in the reduced form, and UV de-tection at 254 nm. Seven types of fruit juices andfoods were tested by 19 collaborators in 7 coun-tries. Three duplicate juices and foods met the crite-ria for Youden pairs and yielded repeatability rela-tive standard deviation of 5.80–14.66%.Reproducibility relative standard deviation rangedfrom 6.36 to 35.54% (n = 10) with HORRAT values of0.82–4.04. The LC method is suitable for routine usein fruit products and foods containing �5 mg/100 gvitamin C and is recommended for further validationby AOAC INTERNATIONAL and International FruitJuice Union.

The importance of vitamin C in human health is well un-derstood, particularly as an antioxidant and in collagensynthesis (1, 2). The recommended daily allowance

(RDA) for vitamin C is determined as 60 mg/day, sufficient toprevent scurvy and maintain a stable body pool of1500 mg (3), much of which comes from fruits and vegeta-bles (4). Vitamin C action is supplied by L-ascorbic acid and

its oxidized form, dehydroascorbic acid, each with equivalentmolar activity, with total vitamin C defined as the sum of bothforms.

To support physiological and pharmacological studies, it isimportant to have reliable analytical data, which can be prob-lematic in biological materials (5), and methods for the deter-mination of vitamin C have been well-reviewed (3, 6–8). Foodmatrixes are equally difficult to determine, particularly vege-tables, where the analyte is in a bound form known asascorbigen. The lability of L-ascorbic acid is a critical factor inanalytical procedures with dehydroascorbate, which is partic-ularly unstable, and can lead to irreversible conversion to in-active diketogulonic acid and other carboxylic acids (9). Insome nutritionally supplemented food and feed products, vita-min C esters, notably ascorbyl 5-palmitate or ascorbyl2-phosphate, can be used to enhance stability (10), although innatural products and supplemented fruit juices, these com-pounds are absent.

A test method must be able to assay ascorbic acid anddehydroascorbic acid without interference from their syn-thetic diastereoisomers isoascorbic acid anddehydroisoascorbic acid or other organic acids. Isoascorbicacid (also called D-ascorbic acid or erythorbic acid) is legallyused as an antioxidant food additive but has poor antiscorbuticactivity (<5%); therefore, its differentiation is a prerequisite toany reliable vitamin C method (8, 11). With only 2 activeforms, the determination of vitamin C is seemingly a simpleanalytical challenge compared with other vitamins. However,poor correlation between laboratories is commonly attribut-able to differences in method specificity for the vitamin Ccongeners, analyte instability to elevated pH, autocatalysisduring extraction, and detection sensitivity limitations (3).

The emergence of liquid chromatographic (LC) methodshas been rapid, and many are preferred for water-soluble vita-min analysis (12), although recent developments in capillaryelectrophoresis indicate potential advantages for vitamin Ctesting (3, 13, 14). Prior to chromatography, vitamin C is usu-

BRAUSE ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 86, NO. 2, 2003 367

Received July 9, 2002. Accepted by SG September 18, 2002.1 Author to whom correspondence should be addressed; e-mail:

[email protected].

368 BRAUSE ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 86, NO. 2, 2003

Tab

le1.

Raw

dat

asu

bm

itted

by

par

ticip

atin

gla

bo

rato

ries

Syn

thet

icor

ange

juic

e

Syn

thet

icor

ange

juic

e

Veg

etab

le-

base

dba

byfo

rmul

a

Veg

etab

le-

base

dba

byfo

rmul

aS

ingl

e-st

reng

thpi

neap

ple

juic

eS

ingl

e-st

reng

thor

ange

juic

eC

ranb

erry

juic

eco

ncen

trat

eC

ranb

erry

juic

eco

ncen

trat

eP

eas

Pea

sP

ears

Pea

rsF

ortif

ied

appl

eju

ice

Unf

ortif

ied

appl

eju

ice

Lab

AB

CD

EF

GH

IJ

KL

MN

123

.523

.55.

7a7.

0a51

.520

.536

.725

.76.

46.

211

.817

.242

.4N

Db

224

.825

.14.

311

.453

.618

.032

.0c

64.1

c6.

15.

212

.512

.762

.4d

ND

325

.126

.06.

212

.858

.621

.534

.835

.87.

37.

913

.419

.544

.7N

D

432

.0c

47.6

c11

.4a

17.2

a62

.7a

23.9

a40

.029

.610

.611

.114

.421

.363

.8d

ND

526

.426

.48.

212

.659

.321

.036

.536

.97.

98.

813

.716

.146

.3N

D

623

.624

.27.

511

.369

.2c

18.7

c33

.334

.36.

78.

212

.017

.940

.6N

D

724

.823

.5N

R11

.0e

49.1

12.3

15.7

f37

.5f

7.8

9.4

14.5

10.3

49.5

ND

824

.521

.35.

716

.350

.6f

8.4f

24.3

37.7

10.8

8.7

13.5

9.7

46.5

ND

928

.227

.4N

R11

.9e

50.4

f7.

5f25

.634

.43.

8a4.

2a12

.410

.938

.4N

D

1024

.024

.0N

R11

.0e

51.0

16.0

27.0

36.0

5.7c

13.0

c11

.017

.046

.0N

D

1125

.226

.62.

111

.556

.218

.134

.838

.56.

16.

114

.317

.845

.1N

D

1223

.423

.14.

911

.849

.315

.6N

R36

.9e

7.9

8.5

13.2

eN

R41

.0N

D

1325

.126

.37.

111

.656

.418

.134

.934

.79.

110

.411

.320

.346

.8N

D

1425

.220

.05.

512

.557

.519

.633

.834

.48.

07.

713

.314

.645

.5N

D

1518

.818

.62.

410

.345

.011

.820

.926

.03.

5a3.

4a4.

4a11

.3a

32.0

dN

D

1623

.623

.15.

012

.851

.017

.832

.233

.78.

89.

59.

017

.941

.9N

D

1726

.525

.65.

5a6.

0a48

.210

.124

.222

.48.

810

.49.

311

.638

.3N

D

1825

.519

.66.

516

.842

.113

.728

.0f

48.6

f5.

16.

2N

R14

.2e

51.1

ND

1931

.230

.33.

412

.155

.622

.330

.726

.79.

78.

316

.018

.555

.6N

D

aIn

valid

data

,poo

rch

rom

atog

raph

y.b

ND

=N

otde

tect

ed;N

R=

notr

ecor

ded.

cC

ochr

anou

tlier

.d

Gru

bbs

outli

er.

eIn

valid

data

,no

mat

ched

pair.

fIn

valid

data

,sam

ple

spoi

lage

.

ally extracted into metaphosphoric or trichloroacetic acid, ei-ther of which functions to stabilize ascorbate (15). Most LCseparations are based on ion-exchange (16) or ion-pair re-versed-phase chromatography (17). Although many methodsexploit the weak-anion exchange capability of amine-bondedsilica (10, 18), reversed-phase techniques with C18 functional-ity have inherent advantages and are frequently used with bothsilica and polymeric styrene-divinyl benzene supports (3, 11).

Despite the widespread use of LC for determination ofvitamin C in foods, there is no official regulatory LC methodcurrently available in the 17th Edition of Official Methods ofAnalysis (19). Of the 4 current AOAC Methods, 967.21 and985.33 are based on visual end point indophenol titration andrequire no sophisticated equipment. However, such methodscannot estimate dehydroascorbic acid, lack sensitivity, and aresubject to interferences from coreductants and colored pig-ments. Methods 967.22 and 984.26 use manual andsemiautomated techniques to assay total vitamin C asdehydroascorbic acid following catalytic oxidation with acti-vated charcoal and reaction with o-phenylenediamine to thefluorescent quinoxaline derivative. These latter methods takeadvantage of the sensitivity and selectivity of fluorescence de-tection, and this principle has also been used in a number ofLC approaches for total vitamin C with precolumn oxida-tion (20, 21). With various combinations of pre- orpostcolumn oxidation schemes, the contributions of ascorbicacid and dehydroascorbic acid have been individually deter-mined (16, 22–29), with dehydroascorbate measuredfluorimetrically, and ascorbate by either UV or electrochemi-cal detection.

Unlike most organic acids, ascorbate absorbs well in themid-UV range and may be detected without interference. Theuse of direct UV detection for total vitamin C in LC-basedmethods, therefore, offers a simple alternative to electrochem-istry or fluorescence. Although inherently less sensitive, thehigh vitamin C content of fruit juices, vegetable products, andsupplemented foods mitigates against this limitation of UVdetection. Because dehydroascorbic acid has no usefulchromophore, UV detection schemes accordingly requirevitamin C to be in the ascorbate form. Depending on mobilephase pH, the optimum detection wavelength can be assignedby using photodiode array detectors (30), although a compro-mise wavelength of 254 nm is commonly used (10, 25).

Much of the early LC development work for vitamin Ccontent in foods was performed to establish a reliable nutri-tional database (4, 10, 11, 16, 26, 31). There have been a num-ber of LC-UV studies for the determination of vitamin C infruits and vegetables (22, 25, 30, 32–35), often concurrentwith other organic acids. Metaphosphoric acid andtrichloroacetic acid are regularly used both to precipitate pro-tein and to stabilize ascorbic acid. However, the presence ofthe weakly retained metaphosphoric acid has occasionallybeen reported to interfere with ascorbate if retention is inade-quate. Dithiothreitol is frequently used to convertdehydroascorbate to the reduced form, although the efficiencyof this step is pH-dependent. Other reductants, includingmercaptoethanol (36), homocysteine (11), and cysteine (37),

have been used, but dithiothreitol remains the most commonfor LC and capillary electrophoresis applications (3, 14).

In 1994, the National Food Processors Associationconducted an unpublished study for vitamin C in fruit juices,based on AOAC Official Method 986.13 for organic acids,which had been developed previously (38). It involved theseparation of organic acids on a C18 column by usingion-suppression techniques at pH 2.4 with UV detection at214 nm. This method has now been optimized for vitamin Canalysis, primarily through use of dithiothreitol reduction andstabilization of the reduced ascorbate form during both samplepreparation and chromatographic analysis. Because processedjuices commonly contain significant dehydroascorbate,dithiothreitol is incorporated with standards, sample extracts,and mobile phase. This study reports the results of the methodperformance trial, primarily focused on processed fruit juicesand selected foods.

Interlaboratory Study

Fourteen samples with vitamin C concentrations rangingfrom 1 to 60 mg/100 g were dispatched to 22 collaborators inthe United States, Canada, Switzerland, the United Kingdom,France, Austria, and Turkey. Frozen samples were dispatchedin dry ice, with random 5-digit numeric coding, by overnightair courier in closed light-proof containers under dry ice.Upon receipt, participants were instructed to store the samplesin a freezer (–20�C) prior to analysis. Participants were alsorequested to perform the determinations (once only) at the ear-liest convenience, but no specific date was given. All laborato-ries tested within 4 weeks of receipt. Only data >2 mg/100 g(20 ppm) were to be recorded.

A practice solution of vitamin C (containing 100 mg ascor-bic acid) in nitrogen-flushed amber vials was also included foreach participant to optimize the chromatography. In particu-lar, the use of tandem columns was permitted if retention wasinsufficient.

A Youden pair design was intended to give estimates ofwithin-laboratory precision without compromising data bymatrix identification. However, because it proved difficult toobtain pairs of samples with concentration differences <5% (arequirement of the Youden pair statistical model), most of thesamples were subsequently treated statistically as individualmatrixes. Reproducibility is the major parameter obtainedfrom interlaboratory studies; therefore, loss of repeatabilitydata was not deemed a disadvantage. The first matrix pair (A& B) consisted of laboratory-prepared orange juices withknown concentrations of vitamin C to determine method re-coveries. All other samples were commercially obtained andwell-mixed before subsampling into nitrogen-flushed poly-propylene bottles with sealed tamper-evident caps. Thesewere stored under dry ice prior to dispatch. Although focusedon fruit juices, namely single-strength orange/pineapple juices(E & F), concentrated cranberry juices (G & H), and supple-mented/natural apple juices (M & N), vegetable-based babyfoods (C & D), canned pears (K & L), and samples of pack-


aged peas (I & J) were included in the study to demonstrate theacceptability of the method to related food samples.

Following removal of invalid data, outliers identified bythe Cochran test for extremes of repeatability (where possible)and the Grubbs test for extremes of reproducibility were omit-ted from estimates of precision. The means, within (sr) and be-tween (sR) laboratory standard deviations, repeatability (r) andreproducibility (R) values, relative standard deviations (RSDr

and RSDR), and HORRAT values were determined accordingto AOAC guidelines (39).

METHOD

Determination of Vitamin C in Fruit Juices andSelected Foods Using LC

(Applicable to the determination of ascorbic acid in fruitjuices and selected foods at 5–60 mg/100 g.)

Principle

Vitamin C is maintained in its reduced form withdithiothreitol, which also converts endogenousdehydroascorbic acid to ascorbic acid. Fruit juices are assayeddirectly or diluted with water; semisolid matrixes are dis-persed in water and filtered. Total vitamin C is determined asascorbic acid with LC–UV. Separation is on a C18 re-versed-phase column with a phosphate-buffered mobile phase(pH 2.5) and detection at 254 nm.

Apparatus

(a) LC system.—Automated or manual equipped withpump for continuous delivery at 0.5–1.0 mL/min and accurateinjection device for 50 �L. A UV detector with stable baselineis set at 254 nm. The use of a photodiode array detector is rec-ommended. Data collection is by integrator or PC.

(b) Chromatography column.—Any suitable 5 �mmonomeric or polymeric silica-based C18 reversed-phase col-umn (nominally 250 � 4 mm). Tandem columns may be usedif required.

(c) Solvent and sample clarification apparatus.—With0.45 �m hydrophilic membrane.

Reagents

Water should be analytical LC grade with resistivity �18 M�.(a) Dithiothreitol.—Aldrich Chemical Co. (Milwaukee,

WI; 15,046.0), or equivalent.(b) Mobile phase.—KH2PO4 (0.5%, w/v), pH 2.5, with

dithiothreitol (0.1%, w/v): Weigh 5.0 g KH2PO4 in 1 L volu-metric flask and add ca 950 mL water. Add 1.0 gdithiothreitol, stir until dissolved, and adjust pH to 2.5 withconcentrated phosphoric acid. Dilute to 1 L with water, andfilter through 0.45 �m membrane.

(c) Vitamin C standards (10–50 �g/mL).—Dissolve5.0 mg ascorbic acid in 100 mL water to give 50 �g/mL. Di-lute 2, 4, 6, and 8 mL to 10 mL with water to give10–40 �g/mL, respectively. Add ca 10 mg dithiothreitol toeach standard. Prepare fresh each day.


Tab

le2.

Sta

tistic

alan

alys

iso

fvita

min

Cin

fru

itju

ices

and

rela

ted

pro

du

cts

by

liqu

idch

rom

ato

gra

ph

y

IDM

ater

ial

No.

ofla

bsa,

bN

o.of

test

sa,b

Mea

n,m

g/10

0g

Src

RS

Dr,

%c

SR

RS

DR,%

rR

HO

RR

AT

Aan

dB

Syn

thet

icor

ange

juic

e18

(1)

38(2

)24

.56

1.42

5.80

2.72

11.0

93.

997.

631.

59

CIn

fant

form

ula

13(0

)13

(0)

5.29

——

1.88

35.5

4—

5.27

4.04

DIn

fant

form

ula

14(2

)14

(2)

11.7

6—

—0.

756.

36—

2.09

0.82

EO

rang

e/pi

neap

ple

juic

e,si

ngle

-str

engt

h16

(0)

16(0

)53

.35

——

6.46

12.1

1—

18.0

31.

95

FO

rang

eju

ice,

sing

le-s

tren

gth

16(0

)16

(0)

17.1

9—

3.65

21.2

5—

10.2

32.

88

Gan

dH

Cra

nber

ryju

ice

conc

entr

ate

15(1

)30

(2)

31.8

84.

6714

.66

5.33

16.7

013

.08

14.9

12.

49

Iand

JP

eas

16(1

)32

(2)

8.12

0.7

9.53

1.67

20.6

12.

174.

682.

50

KP

ears

17(0

)17

(0)

12.6

8—

—1.

8314

.46

—5.

141.

87

LP

ears

17(0

)17

(0)

15.7

4—

—3.

6323

.05

—10

.15

3.08

MF

ortif

ied

appl

eju

ice

16(3

)16

(3)

44.9

8—

—4.

6210

.27

—12

.94

1.61

aN

umbe

rof

labo

rato

riess

orte

sts

reta

ined

afte

rin

valid

data

and

outli

ers

wer

eel

imin

ated

.b

Num

ber

ofla

bora

torie

ssor

test

sre

mov

edas

outli

ers

(inpa

rent

hese

s).

cR

epor

ted

for

You

den

pairs

only

(diff

eren

cein

mea

ns<5

%).

Preparation of Test Solutions

(a) Fruit juices.—Dilute juice samples so that ascorbic acidconcentration is in response range of standards (e.g., becauseorange juice usually contains >30 mg/100 mL ascorbic acid, di-lute 1 mL to 10 mL with water). Record dilution factor and im-mediately add ca 1 mg dithiothreitol for each 1 mL filtrate.

(b) Fruit and processed foods.—In a mixer, blend 10 gfood with sufficient water so that ascorbic acid concentrationis in response range of standards. Record dilution factor andimmediately add ca 1 mg dithiothreitol for each 1 mL filtrate.Filter through fluted 18.5 cm cellulose filter paper (Fisher Sci-entific Co., Pittsburgh, PA; 09-790-14F, or equivalent). Ifsample contains proteinaceous material, mix equal parts of ex-tract with 5% trichloroacetic acid solution before filtration.Extracts difficult to filter may be clarified by centrifugation.

Allow extracts to stand for at least 2 h, filter through0.45 �m membrane, and proceed to determination step.

Determination

Set up instrumentation and software according to manufac-turer’s instructions. Set detector to 254 nm with appropriatesensitivity (usually 0.1 aufs). Set flow rate to 0.5 mL/min, orother, suitable to avoid excessive system pressure, and runwater through the column for ca 30 min, followed by mobilephase for 1 h to equilibrate column.

Inject 50 �L of each of the 5 calibration standards and en-sure stable retention times. On a 25 cm column, ascorbic acidelution time is typically ca 10 min. Measure peak response(height or area), plot manually or electronically against con-centration, and ensure a linear response.

Inject 50 �L of each test solution. Repeat samples that donot lie within the calibration range with appropriate dilution.If photodiode array detector is available, confirm spectralidentity and peak purity of putative ascorbic acid peak.

Calculations

The concentration of vitamin C (mg/100 g or mg/100 mL)can be interpolated directly from the calibration regression us-ing automated data reduction software, from a manually con-structed calibration curve, or from the following equation us-

ing the single-level standard corresponding most closely to theunknown:

Vitamin Csample response standard concentrati

�� on( g / mL) dilution factor

standard response weigh

� �

� t(or volume) used 10�

Collaborators’ Comments

Four laboratories reported that the samples were at roomtemperature or warm upon arrival. Laboratory 12 requested anew set of samples. Where there was any suggestion of spoil-age, the data were removed as invalid. Possible fermentation of2 orange juice and 2 cranberry juice samples was consideredsufficient by the participants to adversely affect their results,and were therefore not subjected to statistical evaluation.

Laboratory 8 reported initial chromatographic difficultiesand repeated 3 samples. Laboratory 10 suggested that asolid-phase extraction cleanup was required to removenonpolar compounds and to prevent carryover of late-elutingpeaks. Laboratories 10, 11, and 16 reported the use of 20 �Linjection volumes rather than 50 �L. Laboratory 11 clarifiedextracts by using centrifugation at 40 000 � g rather than fil-tration, while Laboratory 12 commented on the slow filtrationof some samples. Laboratories 13 and 15 suggestedpredissolution of dithiothreitol rather than adding as a solid.Laboratory 16 suggested using a closed container during re-duction of vitamin C with dithiothreitol, possibly with N2

flushing, and suggested that the column be selected to give anadequate separation of ascorbic acid and isoascorbic acid.

Only one collaborator used a dual-column configuration,because most single-column systems provided sufficient reten-tion and resolution of target analyte. The use of dual columns,however, was considered an advantage in the separation ofascorbic acid from the potential adulterant, isoascorbic acid.

Advice from a collaborator stated that the weight of ascor-bic acid used in the standard solution was too low for accuratemeasurement.

Results and Discussion

The interlaboratory trial protocol aims to assess theinterlaboratory capability of a previously intralaboratory vali-dated analytical method, primarily involving an estimation ofprecision parameters following removal of discordant values.A total of 22 participants were invited to participate in thestudy, 19 of which submitted data. Although laboratory rank-ing analysis indicated that Laboratory 4 had significant posi-tive bias (p = 0.05), the total data set was not removed as in-valid. However, poor chromatography reported by thisparticipant for 2 sets of samples (baby food and orange juice)justified their removal from statistical analysis. Similarly,Laboratory 15 showed a borderline negative bias, andone sample (peas) was removed from further consideration.Of the total raw data set listed in Table 1, 6 failed to report forpaired samples and 11 were compromised either by prior sam-ple spoilage or poor chromatography and were therefore elim-inated as invalid data prior to statistical analysis.


Figure 1. Typical chromatogram of supplementedfood product.

Because of its low vitamin C content (<0.5 mg/100 g), rawdata for sample N (unfortified apple juice) are not reported(only 3 laboratories reported estimates), and this material wasused to define the lower limits of detection of the method.

All collaborators returned acceptable standard calibration pa-rameters with linear regression coefficients (r) typically >0.998.

All valid data sets were submitted to the AOAC harmo-nized statistical protocol. Cochran (p = 0.025, 1-tail) andGrubbs (single and double, p = 0.025, 2-tail) tests were used todetermine outliers. Three Cochran outliers and 2 Grubbs outli-ers were observed among the Youden pairs. Five Grubbs outli-ers were removed from 7 unpaired matrixes, where only an esti-mate of reproducibility precision was available. The totalrejection rate was therefore 10 of 209 data points, which is be-low the rejection level considered acceptable in interlaboratorystudies (39). Table 2 lists the results of the statistical analysis formeans, precision, and HORRAT values following outlier re-moval.

RSDR values estimated for all samples containing vita-min C at 5–53 mg/100 g ranged overall from 6.4 to 35.5%(mean, 17.1%). Because observed variance was reasonablyconsistent across most samples and concentrations, the meanRSDR may be considered to represent overall method preci-sion. Limiting HORRAT values of 0.5–2.0 demonstrates ac-ceptable method precision (39), and in this study, a meanHORRAT value of 2.28 (range, 0.82–4.04) exceeded theseguidelines. However, known exceptions to these generalguideline values include unstable analytes, and the lability ofascorbate is likely to contribute to poorer precision values. In-deed, a similar recently approved European standard methodfor foods, EN 14130 (40), also yielded a comparably elevatedmean HORRAT value of 2.97 (range, 2.86–3.48).

RSDr values were available for only 3 paired matrixes, butthe overall mean RSDr:RSDR value measured 0.57, whichcomplies with the accepted guideline of 0.5–0.7 for the rela-tionship between repeatability and reproducibility precision.

The method protocol allowed for a wide range of LC sys-tem configurations with respect to analytical column and de-tector, and a typical chromatogram is illustrated in Figure 1.

Although most laboratories reported using a single250 mm column, 3 participants used a 300 mm column, andone laboratory used a 150 mm column in tandem with a250 mm column to increase resolution. Five laboratories usedthe Supelcosil (Supelco, Bellefonte, PA) LC-18 column.

Samples A and B were laboratory-made “synthetic” or-ange juices supplemented to 25.0 mg/100 mL and25.5 mg/mL, respectively. Overall means for these sampleswere 25.34 and 25.38 mg/100 mL, which indicate recoveriesof 101.4 and 99.5%, respectively.

Despite the availability of electrochemical or fluorescencedetection techniques, UV detection is the most facile detectionstrategy, with �max of ascorbate at 243 nm, although spectralcharacteristics are pH-dependent (pKa of ascorbate, 4.17). Themid-UV detection of ascorbate minimizes the potential for in-terference from several organic acids, ubiquitous in fruit juicesand other foods, which absorb in the low-UV region (<220 nm).Nevertheless, use of photodiode array detection is considered

expedient to confirm both the identity and peak purity of puta-tive ascorbic acid in poorly characterized food samples.

The polar nature of reduced vitamin C results in relativelypoor retention on hydrophobic reversed-phase functionalities,which is partly moderated through use of a low-pH eluent. Al-though ion-pair techniques may be used to further improve re-tention, the opportunity for interference remains. In reality, al-though samples with low concentrations of vitamin C cannot beunequivocally assayed with UV detection, for fruit, juices, andselected foods with high (>5 mg/100 mL) natural and/or addedascorbate content, the described method provided reliable data.

Although none of the samples in this study containedisoascorbic acid (erythorbic acid), a biologically inactivestereoisomer of ascorbic acid, its use in foods can cause poten-tial overestimation of the vitamin C content of supplementedfood products. These isomers cannot be distinguished by tra-ditional nonchromatographic methods because they haveidentical chemical properties. Separation of these isomers,however, is a desirable attribute, and although resolution is notcomplete with all C18 columns, the current method will iden-tify the presence of isoascorbic acid (Figure 2).

The reduction of endogenous dehydroascorbate and stabi-lization of reduced ascorbic acid with dithiothreitol is a popu-lar approach in vitamin C analysis techniques, and was used inthe method reported here. Because ascorbic acid is also vul-nerable to on-column oxidation during chromatography,dithiothreitol was included in the mobile phase at a concentra-tion of 0.1%, thereby reversing potential auto-oxidation, atechnique previously used for fruit juices (30). It has also beenreported that, in contrast to the typically acidic ascorbate ex-traction conditions frequently described, maintenance of anear neutral solution pH during precolumn reduction is signif-icant in optimizing the reduction efficiency of dithiothreitol,under which conditions ascorbate exists predominantly in themonoanion form. Stoichiometry and kinetic considerationsrequire both excess reductant and adequate reaction time dur-ing precolumn reduction, thereby resulting in quantitativeconversion. Further, the reduced ascorbate is stable for manyweeks under refrigeration in the dark.

It is commonly assumed that ascorbic acid anddehydroascorbic acid have equal equimolar antiascorbutic ac-


Figure 2. Chromatogram of ascorbic and isoascorbicacid standards.

tivities, so that the concentrations of each can be summed toassess the total vitamin C status of foods (3, 8). In reality,dehydroascorbic acid has a somewhat lower activity(approximately 80%), but adjustments for this difference arenot necessary. However, if the recent estimations ofdehydroascorbic acid activities in rats (approximately 10%)can be verified in humans, then there are implications for fu-ture nutritional assessments (41).

Conclusions

The described LC method is applicable over a range ofvitamin C concentrations; however, a study of reproducibilityindicates that it is moderately less precise at the lower concen-tration range. Nevertheless, despite a mean HORRAT value>2.0, the method is considered fit-for-purpose for nutritionallabeling and compositional tables. Although unsuitable forfoods containing low concentration of vitamin C, levels<5 mg/100 mL are nutritionally insignificant relative to theRDA of 60 mg.

On the basis of this interlaboratory study, it is recom-mended that the LC method for the determination of total vita-min C in fruit juices and selected foods at 5–60 mg/100 g issuitable for routine industrial use. The method indicates that itis suitable for further interlaboratory study by AOAC, particu-larly with high concentration materials such as fruit juice andfortified infant formula. It is suggested that alternative proce-dures be used for general food items.

Acknowledgments

The assistance of David Copestake (Fonterra Research Centre,Palmerston North, New Zealand) is acknowledged. We thank thefollowing collaborators for their participation in this study:

AOAC Collaborators

GenevieveCherix,NestleQualityAssuranceLab,Dublin,OHTom Eisele, TreeTop Inc., Selah, WAEd Elkins and C.J. Huang, NFPA East, Washington, DCJane Foos, ABC Labs, Gainesville, FLStacie Hammack and Betsy Woodward, Florida Dept. of

Agriculture, Tallahassee, FLSanford Kirksey, Proctor & Gamble, Cincinnati, OHLinda Kline, Gerber Products, Fremont, MIDana Kruger, Krueger Food Labs, Cambridge, MAMary Jane Lawson, FDA, Atlanta, GAEric Wilhelmsen and Victor Hong, Dole Food Co., San

Jose, CARon Wrolstad and Bob Durst, Oregon State University,

Corvallis, ORL. Zygmunt, Quaker Oats, Barrington, IL

IFU Collaborators

Jale Acar, Hacettepe University, Ankara, TurkeyKodjo Adadevoh and Archana Parkih, Analytical Chemis-

try Services of Columbia, MDDavid Hammond and Andrew Lea, RSSL, Reading, UK

Frank Hesford, Swiss Federation Research Station,Wadenswil, Switzerland

Claudia Hischenhuber, Nestle, Geneva, SwitzerlandGilles Martin, Eurofins Labs, Nantes, FranceJosef Weiss, Hohere Bunidesleh und Versuchanstalt fur

Wein Klosterneuberg, Austria

References

(1) Sauberlich, H.E. (1991) Ann. Rev. Nutr. 14, 371–391

(2) Gershoff, S.N. (1993) Nutr. Rev. 51, 313–326

(3) Eitenmillar, R.R., & Landen, W.O. (2000) in Vitamin Analy-sis for the Health and Food Sciences, CRC Press, BocaRaton, FL, pp 223–270

(4) Vanderslice, J.T., Higgs, D.J., Hayes, J.M., & Block, G.(1990) J. Food Comp. Anal. 3, 105–118

(5) Liau, L.S., Lee, B.L., New, A.L., & Ong, C.N. (1993) J.Chromatogr. 612, 63–40

(6) Nyyssonen, K., Salonen, J.T., & Parviainen, M.T. (2000) inModern Chromatographic Analysis of Vitamins, 3rd Ed.,Chromatographic Science Series, 84, A.P. De Leenheer,W.E. Lambert, & J.F. Van Bocxlaer (Eds), Marcel Dekker,New York, NY, pp 271–300

(7) Moser, U., & Bendich, A. (2000) in Handbook of Vitamins,2nd Ed., L.J. Machlin (Ed.), Marcel Dekker, New York, NY,pp 195–232

(8) Ball, G.F.M. (1998) Bioavailability and Analysis of Vitaminsin Foods, Chapman & Hall, London, UK, pp 517–560

(9) Niemela, K. (1987) J. Chromatogr. 399, 235–243

(10) Sapers, G.M., Douglas, F.W., Ziolkowski, M.A., Miller,R.L., & Hicks, K.B. (1990) J. Chromatogr. 503, 431–436

(11) Hidiroglou, N., Madere, R., & Behrens, W. (1998) J. FoodComp. Anal. 11, 89–96

(12) Ball, G.F.M. (1994) Water-Soluble Vitamin Assays in HumanNutrition, Chapman & Hall, London, UK, pp 274–295

(13) Thompson, C.O., & Trenerry, V.C. (1996) Food Chem. 53,43–50

(14) Cancalon, P.F. (2001) J. AOAC Int. 84, 987–991

(15) Kim, S.R., Lane, R.H., A-Ghany, M., & Stitt, K.R. (1987) J.Food Quality 10, 1–7

(16) Vanderslice, J.T., & Higgs, D.J. (1984) J. Chromatogr. Sci.22, 485–489

(17) Sood, S.P., Sartori, L.E., Wittmer, D.P., & Haney, W.G.(1976) Anal. Chem. 49, 796–798

(18) Dennison, D.B., Brawley, T.G., & Hunter, G.L. (1981) J.Agric. Food Chem. 29, 927–929

(19) Official Methods of Analysis (2000) 17th Ed., AOAC IN-TERNATIONAL, Gaithersburg, MD

(20) Dodson, K.Y., Young, E.R., & Soliman, A-G.M. (1992) J.Assoc. Off. Anal. Chem. 75, 887–891

(21) Speek, A.J., Schriver, J., & Schreurs, W.H.P. (1984) J. Agric.Food Chem. 32, 352–355

(22) Kall, M.A., & Anderson, C. (1999) J. Chromatogr. B 730,101–111

(23) Bognar, A., & Daood, H.G. (2000) J. Chromatogr. Sci. 38,162–168

(24) Kneifel, W., & Sommer, R. (1985) Z. Lebensm. Unters.Forsch. 181, 107–110


(25) Kacem, B., Marshall, M.R., Matthews, R.F., & Gregory, J.F.(1986) J. Agric. Food Chem. 34, 271–274

(26) Ali, M.S., & Phillippo, E.T. (1996) J. AOAC Int. 79, 803–808(27) Haddad, P.R., & Lau, J. (1984) Food Technol. Aust. 36,

46–48(28) Ziegler, S.J., Meier, B., & Sticher, O. (1987) J. Chromatogr.

391, 419–426(29) Finley, J.W., & Duang, E. (1981) J. Chromatogr. 207,

449–453(30) Furusawa, N. (2001) Food Control 12, 27–29(31) Doner, L., & Hicks, K. (1981) Anal. Biochem. 115, 225–230(32) Rodriguez, M.A.R., Oderiz, M.L.V., Hernandez, J.L., &

Lozano, J.S. (1992) J. Chromatogr. Sci. 30, 433–437(33) Daood, H.G., Biacs, P.A., Dakar, M.A., & Hajdu, F. (1994)

J. Chromatogr. Sci. 32, 481–487(34) Gennaro, M.C., & Bertolo, P.L. (1990) J. Liq. Chromatogr.

13, 1419–1494

(35) Bushway, R.J., King, J.M., Perkins, B., & Krishnan, M.(1988) J. Liq. Chromatogr. 11, 3415–3423

(36) Margolis, S.A., & Black, I. (1987) J. Assoc. Off. Anal. Chem.70, 806–809

(37) Iwase, H., & Ono, I. (1993) J. Chromatogr. A 654, 215–220(38) Coppola, E.D., & Duang, M.S. (1986) J. Assoc. Off. Anal.

Chem. 69, 594–597(39) Official Methods of Analysis (2000) 17th Ed., AOAC IN-

TERNATIONAL, Gaithersburg, MD, Appendix D(40) European Committee for Standardization (CEN),

TC275/WG9 (2001) Method EN 14130, Brussels, Belgium(41) Ogiri, Y., Sun, F., Hayami, S., Fujimura, A., Yamamoto, K.,

Yaita, M., & Kojo, S. (2002) J. Agric. Food Chem. 50,227–229


v86n2p367.pdf

Documents