Content of Total Phenolics , Flavan-3-Ols and Proanthocyanidins , Oxidative Stability and Antioxidant Capacity of Chocolate During Storage

Chocolate represents an important source of antioxidants (AOs). This is especially signifi cant since it contains a superior amount of AOs compared to many other foods and beverages (1–3), as well as taking into account its wide consumption. Thus, 40 g of dark or milk chocolate provides 951 or 394 mg of polyphenolic AOs, respectively (4). In the US diet, chocolate provides an estimated 100– 107 mg per day of antioxidants, which represents more than 20 % of AOs originating from fruits and vegetables ISSN 1330-9862 original scientifi c paper


Introduction
Chocolate represents an important source of antioxidants (AOs).This is especially signifi cant since it contains a superior amount of AOs compared to many other foods and beverages (1)(2)(3), as well as taking into account its (total 488 mg per day) (4).Residents of the European Union consume even more chocolate per capita than the US population.In the Dutch population chocolate contributes up to 20 % of the total fl avonoid intake in adults, and in children the percentage is even higher (5,6).
Flavan-3-ols (37 %), anthocyanins (4 %) and proanthocyanidins (58 %) are the main polyphenols in cocoa (Theobroma cacao L.) and cocoa products (7).According to in vitro and in vivo studies, such phenolics possess biological activity that is refl ected in the AO capacity and regulation of the immune system.They also improve the resistance of cocoa fat towards oxidation processes (3).There has recently been an increasing number of experimental and clinical studies that suggest a benefi cial eff ect of chocolate polyphenols on cardiovascular health, i.e. blood pressure, cholesterol level, vasodilatation, but also on insulin sensitivity, infl ammation and immune function (8)(9)(10)(11)(12).
However, not all chocolates are equally rich in polyphenols.Wide variations in polyphenolic compound content depend on the origin of cocoa beans, post-harvest manipulation (13), the applied manufacturing process and the fi nal chocolate composition.According to Belščak et al. (14), extracts of cocoa products containing the highest content of cocoa solids (cocoa liquor and dark chocolates) have the highest content of total and individual classes of polyphenols and exhibit the highest AO capacity in vitro (ferric reducing ability of plasma (FRAP), 2,2 --diphenyl-1-picrylhydrazyl (DPPH) and 2,2'-azino-bis(3--ethylbenzothiazoline-6-sulphonic acid (ABTS) assays), as opposed to milk chocolates.These results corespond to the AO capacity determined previously by using oxygen radical absorbance capacity (ORAC) and vitamin C equivalent AO capacity, as well as total polyphenols and procyanidins determined by Miller et al. (15).Strong linear correlation between non-fat cocoa solids in diff erent cocoa products including chocolates and ORAC (R 2 =0.9849), total polyphenols (R 2 =0.9793) and procyanidins (R 2 =0.46) was observed (15).According to the fi ndings of Vertuani et al. (16), by increasing the percentage of cocoa, the amount of total polyphenols increases and proportionally with it the antioxidant capacity.Since several studies dealing with AO capacity of diff erent plant products over time came to opposite results (17)(18)(19), it was interesting to examine possible changes in AO capacity and phenolic content of diff erent types of chocolate during twelve months of storage.
In the present study, total content of phenolics, fl avan-3-ols and proanthocyanidins, as well as AO capacity of three common types of chocolate with 27, 44 and 75 % cocoa were determined immediately aft er production and aft er six and twelve months of storage.Two widely accepted spectrophotometric assays (FRAP and ABTS) and DC polarographic assay based on the decrease of anodic current caused by hydroxo-perhydroxyl mercury(II) complex (HPMC) formation were applied.Versatility of HMPC assay developed and optimised recently (20) was confi med through wide application on various food samples (21)(22)(23).Relative antioxidant capacity index (RACI) was calculated by taking the average value of the AO assay of samples with equal mass.Relationship between AO capacity and oxidative stability of the fat extracted from chocolates assessed for the fi rst time using diff erential scanning calorimetry (DSC) was of interest as well.Two parameters of the oxidative stability of cocoa butt er extracted from the chocolate samples were determined: oxidation onset temperature and oxidation induction time.Until now, relationship between AO capacity determined using the assay based on HPMC anodic current decrease in the presence of AOs and oxidative stability has not been studied.Both measured and calculated parameters were correlated using regression analysis.The legitimacy of the applied DC polarographic assay has been examined by correlations to other AO assays, analysis of variance (ANOVA) and F-test.Also, the post hoc Tukey's honestly signifi cant diff erence (HSD) test was conducted to check the diff erences in mean values in each AO assay (statistically signifi cant at p<0.05 level).Principal component analysis (PCA) was utilized for grouping of observed samples in factor space, according to the used AO assays.Possibility to select chocolates with higher resistance to oxidation was demonstrated.

Sample preparation
In the present study three diff erent types of chocolate were tested: milk, semisweet and dark chocolate with 27, 44 and 75 % cocoa, respectively.Chocolate composition was as follows: milk chocolate contained 53 % sugars, 31 % fat and 9 % proteins, semisweet chocolate 54 % sugars, 28 % fat and 5 % proteins, and dark chocolate 25 % sugars, 40 % fat and 10 % proteins.Chocolate production was carried out under industrial conditions in the joint-stock company Štark (Belgrade, Serbia) following standard technological operations.Determination of AO capacity of various types of chocolate was performed immediately aft er production, and aft er six and twelve months of storage at 20 °C.
The chocolate samples were prepared as described by Guyot et al. (24) and Hammerstone et al. (25) with some modifi cations.Chocolates were frozen and manually grated.In order to eliminate lipids from the samples, 2.0 g of each kind of chocolate was extracted three times with 10 mL of n-hexane.Extracted fat was used for determination of oxidative stability by DSC.The defatt ed cocoa was air--dried for 24 h to remove the residual organic solvent (26).In order to extract the phenolic compounds, twofold extraction of defatt ed chocolate (2.0 g) with 70 % methanol (5 mL) was performed for 30 min in an ultrasonic bath.Aft er centrifugation the supernatants of both extracts were combined and fi lled up to 20 mL.Extracts, protected with nitrogen, were kept in a freezer (-18 °C) until analysis.

Determination of total phenols
Total phenolic content (TPC) was determined spectrophotometrically according to a modifi ed method of Lachman et al. (27) with Folin-Ciocalteu reagent.Briefl y, 0.5 mL of the sample was added into a 50-mL volumetric fl ask containing 2.5 mL of Folin-Ciocalteu reagent, 30 mL of distilled water and 7.5 mL of 20 % Na 2 CO 3 , and fi lled up to the mark with distilled water.Aft er two hours, the absorbance of blue colouration was measured at 765 nm against a blank sample.Gallic acid (GA) was used as the standard and the results were expressed in mg of gallic acid equivalents (GAE) per g of defatt ed chocolate.All measurements were performed in triplicate.

Determination of fl avan-3-ol content by vanillin and p-DAC assays
The content of fl avan-3-ols was determined by the vanillin assay using the daily prepared working solution of 4 % vanillin in methanol (28).All measurements were performed in triplicate.The content of fl avan-3-ols was calculated according to the formula: where ΔA is the diff erence obtained by subtracting the absorbance of the blank from the absorbance of the corresponding vanillin-containing sample.The results were ex pressed in mg of (+)-catechin per g of defatt ed chocolate.
Alternatively, due to diff erences in the method-reaction mechanisms and selectivity of the used reagents, the content of fl avan-3-ols was also estimated using the p-dimethylaminocinnamaldehyde (p-DAC) reagent, according to a procedure reported by Di Stefano et al. (28).The content of fl avan-3-ols was calculated according to the formula: where ΔA is the diff erence of the absorbance of the tested extracts and appropriate blanks.The results were expressed in mg of (+)-catechin per g of defatt ed chocolate.

Determination of proanthocyanidins
Proanthocyanidins were analysed by the butanol/ HCl assay, modifi ed from the method of Bate-Smith, described by Porter et al. (29).The quantity of proanthocyanidins, determined from a standard curve of cyanidin chloride treated with butanol-HCl-Fe(III) mixture, was expressed in mg of cyanidin chloride equivalents (CyE) per g of defatt ed chocolate.

Determination of AO capacity using DC polarography
A dropping mercury electrode (DME) with a programmed dropping time of 1 s was used as working electrode .Saturated calomel electrode (SCE) was used as reference and a Pt-foil as auxiliary electrode.The polarographic analyser PAR model 174A (Princeton Applied Research , Artisan Technology Group ® , Champaign, IL, USA), equipped with Omnigraphic ® X-Y recorder 2000 (Houston Instrument, Austin, TX, USA) was used to record the current vs. potential curves.The initial scanning potentials were 0.10 or 0.15 V. Potential scan rate was 10 mV/s.The DME current oscillations were fi ltered with instrumental low pass fi lter positioned at 3 s.Samples were gradually added into 5 mM H 2 O 2 solution obtained by adding 100 μL of 1.0 M H 2 O 2 into 20 mL of Clark and Lubs buff er (pH=9.8)prepared by mixing 25 mL of 0.4 M H 3 BO 3 , 25 mL of 0.4 M KCl and 40.8 mL of 0.2 M NaOH.Aft er the addition of the investigated samples decrease of the initial anodic current of 5 mM H 2 O 2 was recorded.Linear slope of the decrease of the current depending on the volume of the sample, expressed in % per mL, was used to express AO capacity of chocolate extracts.The inert atmosphere during the recording of current vs. potential curve was kept by passing nitrogen above cell solution.A stream of pure nitrogen was passed through the cell so lution for 5 min before the fi rst recording and for 30 s aft er each aliquot addition.All experiments were done in triplicate at room temperature.

Determination of free radical scavenging ability by the use of ABTS˙+
The Trolox equivalent antioxidant capacity (TEAC) of chocolate extracts was estimated by the ABTS˙+ decolourisation assay (30).The results were expressed as mean value of triplicate analyses in Trolox equivalents (TE), derived from a calibration curve (c(Trolox)=100-1000 μmol/L).

FRAP assay
The ferric reducing antioxidant power (FRAP) assay was carried out according to a standard procedure by Benzie and Strain (31).All measurements were performed in triplicate.Aqueous solution of FeSO 4 •7H 2 O was used to prepare the calibration curve and the results of the assay were expressed in mmol of Fe(II) per L.

Determination of oxidative stability by DSC
Oxidation induction time and oxidation onset temperature were determined by DSC, using ASTM methods (32,33) on DSC Q1000 (TA Instruments, New Castle, DE, USA).Lipids obtained from the sample of 3.0-3.3mg of chocolate extract were weighed in the open aluminium pan.For the determination of oxidation induction time, samples were heated at a specifi c constant test temperature in an oxygen environment (50 mL/min).For the determination of oxidation onset temperature, lipid samples were heated at a heating rate of 10.0 °C/min, in temperature interval from 25 to 250 °C, in open aluminium containers, which served as an ethalon, with an oxygen purge fl ow of 50 mL/min.A rapid heat increase was observed with an exothermic heat fl ow manifested during the beginning of the oxidation reaction.Oxidation induction time and onset temperature were determined from the resulting exotherm (Universal Analysis 2000 soft ware v. 4.1D, TA Instruments).

Determination of RACI
The standard scores of a certain sample, calculated aft er the ranking of diff erent assays, gave a single unitless value, called relative antioxidant capacity index (RACI) (21).This value is a particular combination of data obtained using specifi c measuring methods with no unit impediment and variance among them.

Statistical analysis
Descriptive statistical analyses for calculating the mean values and the standard error of the mean were performed using Microsoft Excel soft ware (MS Offi ce 2007, Microsoft Corporation, Redmond, VA, USA).The mean value and standard deviation of triplicate readings were recorded (N=3), and all obtained results were expressed as the mean value±standard deviation (S.D.).The evaluation of correlation matrix, two-way ANOVA and F-test, of the obtained AO assays were performed for comparison of mean values, and signifi cant diff erences were determined using post hoc Tukey's honestly signifi cant diff erence (HSD) test at the p<0.05 level.PCA of the obtained results was also performed using STATISTICA v. 10 software (StatSoft , Dell Soft ware, Tulsa, OK, USA).

Results and Discussion
In the present paper, antioxidant (AO) capacity of three types of chocolate: milk, semisweet and dark chocolate with 27, 44 and 75 % cocoa, respectively, was assessed during twelve months of storage using recently developed direct current (DC) polarographic assay (20).Methanolic extracts of fresh, six-and twelve-month stored samples were gradually added into initial buff ered solution.Signifi cant diff erences were observed between polarograms of 5 mmol/L solution of H 2 O 2 in Clark and Lubs buff er (pH=9.8)recorded before and aft er the addi-tion of chocolate extracts (Fig. 1).The most potent eff ect on the anodic current originating from HPMC formation was ascribed to dark chocolate with the highest percentage of cocoa, as opposed to milk chocolate.Height of initial anodic limiting current was compared with remaining limiting current obtained aft er gradual addition of tested samples.Aft er each addition, the percentage of limiting current decrease was calculated.
A dose-dependent eff ect of the investigated samples on HPMC anodic current was observed.Decrease of limiting current measured aft er each addition of chocolate extract is shown in Fig. 1.It can be seen that linear relationship was observed only for milk chocolates.Deviation from linearity was detected when 100-μL aliquots of both semisweet and dark chocolate extracts were added.Thereaft er, the volume of aliquots decreased.Gradual addition of 50 μL of samples was found optimal (not shown).The slope of the linear part of dose-response curves was used to determinate the AO capacity (Table 1).
In accordance with the recommended multilateral approach to assess AO capacity (34,35), two common spectrophotometric AO assays were applied in parallel with DC polarographic one.Antiradical activity towards ABTS˙+ and FRAP of chocolate samples were measured (Table 1).In order to compare AO activities measured using various assays, unitless RACI was used.Strong relationship between overall AO capacity and cocoa was confi rmed.The highest value of RACI (0.96) was ascribed to dark chocolate with 75 % cocoa, while the lowest to milk chocolate (-1.31;Table 1).Superior AO capacity of chocolates with 75 and 44 % of cocoa in comparison with milk chocolate was confi rmed using RACI (Table 1).Total phenolic, fl avonoid and proanthocyanindin contents were used to characterise samples more comprehensively and also to enable bett er insight into the changes of health-related parameters during storage (Table 2).Time dependence of AO capacity of milk, semisweet and dark choco-  (17).In other words, AO capacity of most analysed samples measured immediately aft er the production was not so diff erent from that found in the samples kept at 20 °C for 180 or 360 days.
Besides phenolic content and AO capacity, oxidative stability of chocolates during storage was measured by DSC.To the best of our knowledge, this is the fi rst time DSC has been used to determine the parameters of the oxidative stability of cocoa butt er extracted from chocolate samples, i.e. oxidation onset temperature and oxidation induction time.The transfer of an oxygen molecule to an unsaturated fatt y acid requires energy easily determined by DSC (33,36).Oxidative stability is represented by oxidation induction time or onset temperature at which chocolate fat resists oxidation.While oxidation induction time was determined by heating the sample (in oxygen) to a constant temperature and measuring the developed heat in DSC experiment, oxidation onset temperature at which oxidation occurs is a relative measure of oxidative stability at the used heating rate.Previously, DSC had been mainly applied to characterise melting profi les of chocolate samples and to determine the degree of crystallisation that infl uences sensory and rheological properties, as well as shelf life (37).Lipids extracted from chocolates with higher total phenolic content and superior AO capacity are less prone to oxidation (Table 3).
In this study, two-way ANOVA was conducted to show the signifi cant eff ects of the independent variables on the responses, which were signifi cantly aff ected by the Determined using Folin-Ciocalteu, vanillin and p-DAC assay, respectively Values with the same lett er in superscript are not statistically diff erent at the p<0.05 level (Tukey's HSD test) various parameter combinations.The cocoa content was the most infl uential variable in AO capacity calculation, statistically signifi cant at p<0.05 level with 95 % confidence limit, while chocolate type was also very important variable in TPC, FRAP, ABTS˙+ and oxidation induction time assays, statistically signifi cant also at p<0.05.The product of these two variables was found statistically signifi cant in TPC, fl avan-3-ol, FRAP, ABTS˙+ and oxidation induction time assays.
Both measured and calculated parameters were correlated using regression analysis, and the obtained DC polarographic assay was validated.Stepwise regression between RACI and the used AO assays (FRAP, ABTS˙+ and HPMC) revealed that each of the three assays matched RACI, and that correlations between RACI and the used assays were highly signifi cant (p<0.01 level).RACI strictly correlated with both FRAP and ABTS˙+ assays (0.99), and HPMC (0.97) (Table 4).Strong positive correlations were established between the AO activities of dark and milk chocolates deduced from cyclic voltammograms with those determined using DPPH, ABTS˙+ and FRAP assays (38).Reliability of HPMC assay as a measuring method with the lowest standard deviation recorded was confi rmed by experimental measurements.The coeffi cients of variation for HPMC assay, calculated as the ratio of the standard deviation divided by the mean, are 0.01-0.05.
Correlations between RACI and total phenolic content (determined by Folin-Ciocalteu reagent), proanthocyanidin and fl avan-3-ol determined using vanillin assay were high (0.99), as well as the correlation with fl avan-3--ol determined using p-DAC assay (0.97).Correlation between RACI and oxidation onset temperature (0.79) was signifi cantly lower than the correlation with oxidation induction time (0.93).Generally, parameters determined using both spectrophotometric and DC polarographic assays correlated well with oxidation induction time (0.86-0.98), while correlations with oxidation onset temperature were considerably lower (0.73-0.85).However, the highest correlation was found between HPMC and oxidation onset temperature (0.85) or oxidation induction time (0.98).Correlations with the results of FRAP (0.732 and 0.865) and ABTS˙+ (0.764 and 0.904) assay were substantially lower.Finding that AO capacity determined by HPMC correlates well with oxidative stability might gain importance.The post hoc Tukey's HSD test was used for comparison between the chocolate samples within each used assay, and statistically insignifi cant diff erence was found in almost all samples during storage at p<0.05 level (Tables 1-3).HPMC and total phenolic content of semisweet chocolate varied signifi cantly during storage.As expected, oxidation onset temperature was statistically signifi cant for all types of chocolate.
The PCA reduced the number of variables and enabled to establish the relationship between the used AO assays and diff erent types of chocolates that give complimentary data.The full autoscaled data matrix consisting of three diff erent types of chocolates, aft er 0, 6 and 12 months of storage, and nine measured parameters, as well as the calculated RACI were submitt ed to PCA.For Table 4. Correlations between antioxidant capacity determined using DC polarographic (HPMC), ABTS˙+ and FRAP assays, content of total phenolics (TPC), fl avan-3-ols (FLA) and proantocyanidins (PAC), oxidative stability (oxidation induction time (OIT) and onset temperature (OOT)) and reactive antioxidant capacity index (RACI) FLA PCA modelling, the 3×3×10 data matrix was divided into a sample code and diff erent data sets.All examples in the distinctive classifi cations obtained diverse scores, as anticipated by PC score plot.The variability of the results in three types of chocolate is large, so the diff erent categories of chocolates are formed according to the content of total phenols, fl avan-3-ol (vanillin and p-DAC assays) and proanthocyanidins, and AO capacity determined using FRAP, ABTS˙+ and HPMC, as well as oxidative stability (oxidation onset temperature and oxidation induction time).
For visualising the data patt erns and the discriminating effi ciency, a scatt er plot of the observed samples, using the fi rst two principal components (PCs), obtained from PCA, was made (Fig. 2).As can be seen, there is a neat separation of the three sorts of chocolates.The fi rst two principal components, accounting for 98.7 % of the total variability, are considered to be suffi cient for such data, also visible from the high correlation coeffi cients obtained between diff erent assays applied 4).

Conclusions
There is a strong relationship between the content of cocoa, total phenols, fl avan-3-ols and proanthocyanidins on the one hand, and total antioxidant capacity (determined using two widely accepted spectrophotometric and a recently developed direct current (DC) polarographic assays) and calculated relative antioxidant index on the other.Also, strong relationship with the oxidative stability determined using diff erential scanning calorimetry has been confi rmed unequivocally.Very good correlation between hydroxo-perhydroxyl mercury(II) complex and both oxidation induction time and oxidation onset temperature suggests that reliable, cheap and easy-to-handle DC polarographic assay, validated using regression analysis, two-way ANOVA and F-test, can be used as an alternative to spectrophotometric assays commonly applied in the analysis of chocolate and cocoa products, and as an indicator of oxidative stability of chocolates as well.Simple and fast procedure with possibility to process opalescent or turbid samples directly can be considered as additional advantage.There is a growing body of data obtained by parallel application of DC polarographic and common spectrophotometric assays on various food and beverage items and this study contributes to its extension.Multilateral approach and introduction of RACI enables more comprehensive comparison between analysed samples.Also, bett er insight into health-related or functional characteristics of chocolates and their changes during storage is allowed.Based on all analysed parameters, quality of chocolate remains unchanged during twelve months of storage.

Fig. 1 .
Fig. 1.Anodic polarographic curves of 5 mmol/L solution of H 2 O 2 in Clark and Lubs buff er (pH=9.8)before (0) and aft er the addition of fi ve equal aliquots (from 1 to 5: V=200-1000 mL) of methanolic extract of chocolate (above) and dose-dependence curves of formed HPMC in relation to volume of chocolate extracts (below) with: a) 27, b) 44 and c) 75 % cocoa

Fig. 2 .
Fig. 2. PCA biplot of the results obtained for three diff erent types of chocolates (M=milk, S=semisweet, D=dark) analysed aft er production, and aft er 6 and 12 months of storage.For abbreviations see Table 4

Table 3 .
Oxidative stability of lipids extracted from milk, semisweet and dark chocolates determined by diff erential scanning calorimetry Determined using vanillin and p-DAC assay, respectively.All shown correlations are signifi cant at p<0.01, N=12, except those marked with asterisk, which are signifi cant at p<0.05 level