IJFS#816_bozza Ital. J. Food Sci., vol. 30, 2018 - 89 PAPER OPTIMISATION OF THE EXTRACTION OF FLAVONOIDS FROM APPLES USING RESPONSE SURFACE METHODOLOGY M. LIAUDANSKAS*1,2, K. ZYMONĖ1, J. VIŠKELIS2 and V. JANULIS1 1Department of Pharmacognosy, Faculty of Pharmacy, Lithuanian University of Health Sciences, Lithuania 2Institute of Horticulture, Lithuanian Research Centre for Agriculture and Forestry, Lithuania *E-mail address: m.liaudanskas@yahoo.com ABSTRACT The ultrasound-assisted extraction of flavonoids from apple samples was modelled using response surface methodology. A three-level-three-factor central composite design using the response surface methodology (RSM) was employed to optimise three extraction variables, including temperature, extraction time and ultrasonic power, for the achievement of the highest extraction yield of the flavonoids from lyophilised apple samples. The optimised extraction conditions were 44.61ºC, an extraction time of 26.90 min, and ultrasonic power 480 W. The experimental yield of flavonoids was 6.58 mg g-1 expressed as rutin equivalent, which was close to the predicted yield (6.69 mg g-1). Optimised extraction conditions were applied for the analysis of apple samples of six cultivars. Keywords: apple, central composite design, flavonoids, HPLC, response surface methodology, ultrasonic extraction Ital. J. Food Sci., vol. 30, 2018 - 90 1. INTRODUCTION Apples play an important role in the human diet. They are one of the most consumed fruits in the whole world (WU et al., 2007, CEYMANN et al., 2012). Based on data from the year 2014, approximately 84.63 million tonnes of apples are grown per annum. Countries that grow the most apples are China (approximately 40.92 million tonnes per annum), the USA (approximately 5.19 million tonnes per annum) and Poland (3.20 million tonnes per annum) (FAO Statistical Database, 2017). Apples are widely used in the food industry to produce various products and drinks (juice, wine, cider); they are also used unprocessed (MARKS et al., 2007, PRICE et al., 1999). Some of the most important biologically active substances in apples are phenolic compounds, which are attributed to natural antioxidants. Oxidative stress causes changes in cell metabolism related to DNA and protein damage as well as lipid peroxidation (COOKE et al., 2003; PIZZIMENTI et al., 2010). It can cause inflammatory processes, cardiac, vascular and other diseases (MADAMANCHI et al., 2010). Phenolic compounds neutralise reactive forms of oxygen and nitrogen (PANDEY and RIZVI, 2009) and therefore are valuable for the treatment and prophylaxis of various diseases. Qualitative and quantitative composition analyses of raw materials that accumulate phenolic compounds are important and relevant. The selection of extraction conditions is an important analytical step in developing the qualitative and quantitative analysis methodologies of phenolic compounds in multi- component matrices. Our developed and validated method of flavonoid and phenolic acid determination in apples is published in the paper by LIAUDANSKAS et al. (2014). In developing this method, the extraction parameter selection was empirical. It is relevant to compare flavonoid extraction yield when the samples are extracted using empirical extraction conditions and when conditions are selected based on statistical modelling. 2. MATERIALS AND METHODS 2.1. Plant material Apple samples of the Ligol cultivar were chosen for the extract optimisation analysis. The Ligol cultivar (winter cultivar, bred in Poland) is one of the main cultivars in commercial apple orchards in Lithuania. Optimised extraction conditions were applied for the analysis of the apple samples of different cultivars. The following apple cultivars were included in the comparable researches: Aldas (early winter cultivar, bred in Lithuania, recommended for ecological orchards), Auksis (early winter cultivar, bred in Lithuania,), Connel Red (winter cultivar, bred in USA), Ligol, Lodel (early winter cultivar, bred in Lithuania) and Rajka (early winter cultivar, bred in Czech Republic). The apple trees were grown in the experimental orchard (block 2, row 4, trees 21-40) of the Institute of Horticulture, Lithuanian Research Centre for Agriculture and Forestry, Babtai, Lithuania (55°60′ N, 23°48′ E). The altitude of Babtai town is 57 m above sea level. Trees were trained as a slender spindle, and pest and disease management was carried out according to the rules of the integrated plant protection. The experimental orchard was not irrigated. Tree fertilisation was performed according to soil and leaf analysis. In addition, nitrogen was applied before flowering at the rate of 80 kg ha-1, and potassium was applied after harvest at the rate of 90 kg ha-1. Soil conditions of the experimental orchard were the following: clay loam, pH - 7.3, humus - 2.8%, P2O5 - 255 mg kg-1 and K2O - 230 mg kg-1. Ital. J. Food Sci., vol. 30, 2018 - 91 2.2. Sample preparation Apples were cut into slices of equal size (up to 1 cm in thickness), and the stalks and the seeds were removed. The apple slices were immediately frozen in a freezer (at -35°C) with air circulation. Apple samples were lyophilised with a ZIRBUS sublimator 3× 4×5/20 (ZIRBUS technology, Bad Grund, Germany) at a pressure of 0.01 mbar (condenser temperature, -85°C). The lyophilised apple slices were ground to a fine powder (about 100 µm) by using the knife mill Grindomix GM 200 (Retsch, Haan, Germany). Loss on drying before analysis was determined by drying the apple lyophilisate in a laboratory drying oven to complete the evaporation of free water and volatile compounds (temperature 105°C) and by calculating the difference in raw material weight before and after drying (European Pharmacopoeia, 2010). The data were recalculated for the absolute dry lyophilisate weight. 2.3. Chemicals All solvents, reagents, and standards used were of analytical grade. Acetonitrile, aluminium trichloride hexahydrate, hexamethylenetetramine and acetic acid were obtained from Sigma-Aldrich GmbH (Buchs, Switzerland), and ethanol from Stumbras AB (Kaunas, Lithuania). Hyperoside, rutin, quercitrin, phloridzin, procyanidin B1 and procyanidin B2, and chlorogenic acid standards were purchased from Extrasynthese (Genay, France); reynoutrin, (+)-catechin and (−)-epicatechin were purchased from Sigma- Aldrich GmbH (Buchs, Switzerland); and avicularin, procyanidin C1 and isoquercitrin were purchased from Chromadex (Santa Ana, USA). In the study, we used deionised water that the Crystal E HPLC (Adrona SIA, Riga, Latvia) water purification system produced. 2.4. Extraction A total of 2.5 g of lyophilised apple powder (exact weight) was weighed, added to 30 mL of ethanol (70%, v/v) and extracted in a Sonorex Digital 10 P ultrasonic bath (Bandelin Electronic GmbH & Co. KG, Berlin, Germany). A total of 480 W is the maximum ultrasonic power that can be achieved by using the Sonorex Digital 10 P ultrasonic bath. The obtained extract was filtered through a paper filter, and the apple lyophilisate on the filter was washed twice with 10 mL of ethanol (70%, v/v) in a 50mL flask. Then, the extract was filtered through a PVDF syringe filter with a pore size of 0.22 µm (Carl Roth GmbH, Karlsruhe, Germany). 2.5. Determination of the total flavonoid content The total flavonoid content in the extracts of lyophilised apple samples was determined by applying the technique that URBONAVIČIŪTĖ et al. (2006) described. It was calculated using the rutin calibration curve and was expressed as its equivalent (mg RE/g) for absolute dry weight (DW). 2.6. High-performance liquid chromatography The qualitative and quantitative analyses of phenolic compounds were performed according to the previously validated and described high-performance liquid chromatography (HPLC) method (LIAUDANSKAS et al., 2014). Ital. J. Food Sci., vol. 30, 2018 - 92 2.7. Experimental design and statistical analyses Before the development of the study through the response surface methodology (RSM), the flavonoid yield from the apples of three extractants (ethanol, methanol, and acetone) of different concentrations were compared. It was determined that the highest yield of flavonoids after four hours of extraction was achieved by macerating apple samples with ethanol 70% (v/v); therefore, this extractant was chosen for further analyses. Selection of extraction method: The efficacy levels of maceration and extraction in ultrasonic bath methods were compared, and it was determined that the yield was higher when sonification method was applied. The results for the extractant and extraction method selections are discussed more in depth in the paper by LIAUDANSKAS et al. (2014). In general, such multiple parameters as liquid/solid ratio, temperature, time, solvent polarity and ultrasonic power influence the efficiency of the extraction of a compound, and these are the primary extraction parameters that many other authors have referred to (TIAN et al., 2013; RADOJKOVIC et al., 2012; CHEN et al., 2012) In this study, three factors (or independent variables) were selected: temperature (20-60°C), extraction time (5-95 min) and ultrasonic power (48-480 W) (Table 1). Table 1. Factors and levels for RSM, and central composite experimental design with the independent variables. Run Coded and non-coded variable levels Total flavonoid content, mg RE g-1 X1 Temperature, °C X2 Extraction time, min X3 Ultrasonic power, W 1 0 40 0 50 0 264 6.889 2 0 40 0 50 -1 48 6.58 3 0 40 -1 5 0 264 4.199 4 0 40 1 95 0 264 6.312 5 -1 20 -1 5 -1 48 3.154 6 1 60 0 50 0 264 6.315 7 0 40 0 50 0 264 7.087 8 1 60 1 95 -1 48 5.618 9 1 60 1 95 1 480 5.963 10 -1 20 1 95 -1 48 4.682 11 1 60 -1 5 -1 48 4.102 12 -1 20 -1 5 1 480 3.555 13 0 40 0 50 0 264 6.794 14 0 40 0 50 0 264 7.175 15 0 40 0 50 0 264 6.821 16 -1 20 0 50 0 264 4.966 17 0 40 0 50 0 264 6.982 18 -1 20 1 95 1 480 5.406 19 1 60 -1 5 1 480 4.665 20 0 40 0 50 1 480 7.465 Ital. J. Food Sci., vol. 30, 2018 - 93 A three-level-three-factor central composite design was employed to determine the optimal combination of flavonoid extraction variables from apple samples. Table 1 represents the coded and non-coded values of the experimental variables and 20 experimental points. Six replicates (1, 7, 13, 14, 15, 17) were used to evaluate the pure error. Experimental data showed that response variables were fitted to a quadratic polynomial model. The general form of the quadratic polynomial model is presented in Fig. 1, where Y is the dependent variable; β0, βi, βii and βij are the regression coefficients for intercept, linearity, square and interaction respectively. Xi and Xj are the independent variables. Figure 1. The general form of the quadratic polynomial model. Design-Expert® 6.0.8 software (Stat-Ease Inc., Minneapolis, Minnesota, USA) was used to analyse the data, develop models and optimise the extraction conditions. The fitness of the quadratic polynomial model was inspected with the regression coefficient of R2. The p- value was used to check the significance of the regression coefficient. All of the experiments (except extraction optimisation researches) were carried out in triplicate. Means and standards errors were calculated with SPSS 20.0 software (Chicago, USA). A single factor analysis of variance (ANOVA) along with the post hoc Tukey's HSD test was employed for statistical analysis. Differences were considered to be significant at the p<0.05 level. 3. RESULTS AND DISCUSSIONS 3.1. Optimisation of extraction conditions of total flavonoids in apples The design matrix and the corresponding results of RSM experiments to determine the effects of the three independent variables, including temperature (X1), extraction time (X2) and ultrasonic power (X3), are shown in Table 2. Through multiple regression analysis of the experimental data, the model for predicted response Y could be expressed with the following quadratic polynomial equation (in the form of coded values), presented in Fig. 2. Figure 2. The model for the predicted response Y expressed by the quadratic polynomial equation (in the form of coded values). Ital. J. Food Sci., vol. 30, 2018 - 94 Statistical testing of the model was performed in the form of analysis of variance (ANOVA). The ANOVA for the fitted quadratic polynomial model of extraction of polysaccharides is shown in Table 2. Table 2. Analysis of variance for fitted quadratic model of extraction of phenolic compounds. Source Sum of squares Degree of freedom p-value Model 32.32 9 <0.0001 Significant Residual 0.60 10 Lack of fit 0.49 5 0.0698 Not significant Pure error 0.11 5 Cor. total 32.92 19 R2 = 0.9818; 𝑅!"# ! = 0.9674; C.V. = 4.27%; adequate precision = 26.232. The results of the analysis of variance for the fitted quadratic model of extraction of phenolic compounds are presented in Table 2. They indicated a high degree of correlation between the observed and predicted values. The lack of fit test determines whether a selected model is adequate for explaining the experimental data or whether another model should be reselected. The value of the lack of fit test indicated that the fitting model was adequate. An adequate precision is a measure of the signal-to-noise ratio, which when greater than 4 is considered to be adequate (CANETTIERI et al., 2007). In addition, the value of adequate precision demonstrates an adequate signal. At the same time, a relatively low value of the coefficient of variation indicates a better precision and reliability of the experimental values. Therefore, the model is adequate for prediction in the range of experimental variables. Residual analysis of the response surface design was performed. A normal probability plot was applied to the residuals. The data points fell along a straight line, indicating that they were distributed normally. In addition, residual runs for analysis were used. It was determined that the order of observations did not influence the results. Residuals versus predicted responses were plotted. The data points fell on both sides of the zero line, so no pattern could be concluded. The relationship between the actual and predicted values was evaluated. The data points fell along a straight line indicating close similarity between the two data points and the adequacy of the model. Table 3. Regression coefficients estimate and their significance test for quadratic model. Source Sum of squares Degree of freedom p-Value X1 2.40 1 <0.0001 X2 6.90 1 <0.0001 X3 0.85 1 0.0037 3.23 1 <0.0001 5.94 1 <0.0001 0.24 1 0.0717 X1X2 0.040 1 0.4337 X1X3 5.886×10 -3 1 0.7605 X2X3 1.378×10 -3 1 0.8825 Ital. J. Food Sci., vol. 30, 2018 - 95 The significance of each coefficient measured using the p-value is listed in Table 3. A smaller p-value means the corresponding variables are more significant. The p-value of the model is less than 0.0001, which indicates that the model is significant and can be used to optimise the extraction variables. The three independent variables (X1, X2, X3) and two quadratic terms (𝑋!! and 𝑋!!) significantly affect the extraction yield of flavonoids. The interaction among temperature (X1), extraction time (X2) and ultrasonic power (X3) did not affect the extraction yield of flavonoids significantly. The three-dimensional response surface is the graphical representation of the regression equation and is very useful for judging the relationship between independent and dependent variables. Different shapes of the contour plots indicate whether the mutual interactions among the variables are significant or not. A circular contour plot means the interactions between the corresponding variables are negligible, whereas an elliptical contour suggests that the interactions between the corresponding variables are significant (MURALIDHAR et al., 2001). The three-dimensional representation of the response surfaces generated via the model is shown in Fig. 3–5. With these three variables, when two variables are depicted in three-dimensional surface plots, the third variable is fixed at the zero level. Figure 3. Response surface plot showing the effect of temperature (X1) and extraction time (X2). X3 (Ultrasonic power) = 264 W. Ital. J. Food Sci., vol. 30, 2018 - 96 Figure 4. Response surface plot showing the effect of temperature (X1) and ultrasonic power (X3). X2 (Extraction time) = 50 min. Figure 5. Response surface plot showing the effect of extraction time (X1) and ultrasonic power (X3). X2 (Temperature) = 40°C. It is found in Figs. 3-5 that all of the three response surfaces are convex in shape, which indicates that the ranges of variables were chosen properly. As shown in Fig. 3, the yield of extraction increases when extraction time and temperature are increased. Based on the chosen model, the projected highest amount of phenolic compounds (6.69 mg g-1) is achieved when samples are extracted at 44.61ºC for 26.90 min. According to the applied model, a further increase of extraction time and temperature decreases the extraction yield. Flavonoid extraction yield dependency on temperature and ultrasonic power is presented in Fig. 4. The highest extraction yield is achieved when apple samples are extracted at highest ultrasound power (480 W), at 44.61ºC. Extraction time and ultrasound power influence on flavonoid extraction yield is presented in Fig. 5. The highest yield is achieved when the ultrasound power is maximal (480 W), and extraction time is 50 min. 3.2. Optimization of extraction parameters and validation of the model Through these three-dimensional plots, the suitability of the model equation for predicting the optimal response values was tested using the selected optimal conditions. The results (Table 3) showed that the optimized conditions were ultrasonic temperature of 44.61ºC, extraction time of 26.90 min, and ultrasonic power 480 W. Under these conditions, the predicted extraction yield of flavonoids was 6.69 mg g-1. However, considering the operability in actual production, the optimal conditions can be modified as follows: temperature of 45ºC, extraction time of 27 min, and ultrasonic power 480 W. Under the modified conditions, the experimental yield of flavonoids was 6.58 mg RE g-1 (n = 3), which was close to the predicted value. Extraction yield was 4.1 mg RE g-1 when extraction conditions were selected empirically (LIAUDANSKAS et al., 2014). When extraction conditions were optimized by statistical modelling method the yield was 37.69% higher than the yield when extraction conditions were selected empirically. Ital. J. Food Sci., vol. 30, 2018 - 97 3.3. Analysis of ethanol extracts of apple samples of different cultivars The chemical composition in fruits of different apple cultivars may vary significantly (CEYMANN et al., 2012; ŁATA et al., 2005; WOJDYŁO et al., 2008), therefore it is very important to determine the qualitative and quantitative composition of individual phenolic compounds in apples that are grown under Lithuanian climatic conditions. Optimized extraction conditions were applied for the HPLC analysis of ethanol extracts of apple samples of six cultivars grown in Lithuania: Aldas, Auksis, Connel Red, Ligol, Lodel and Rajka using previously developed and validated HPLC method (LIAUDANSKAS et al., 2014). These phenolic compounds of various groups were identified and quantified in analysed extracts: procyanidin B1, (+)-catechin, chlorogenic acid, procyanidin B2, (−)- epicatechin, procyanidin C1, rutin, hyperoside, isoquercitrin, reynoutrin, avicularin, quercitrin and phloridzin (synonym phlorizin). Apple sample chromatogram (cultivar Lodel) is presented in Fig. 6. Total amount of phenolic compound in analysed apple sample extracts varied from 2.521 mg g-1 (cultivar Connel Red) to 6.430 mg g-1 (cultivar Aldas). Figure 6. Chromatogram of the ethanol extract of apple sample (λ=280 nm, cultivar Lodel). 1 - Procyanidin B1, 2 - (+)-catechin, 3 - chlorogenic acid, 4 - procyanidin B2, 5 - (−)-epicatechin, 6 - procyanidin C1, 7 - rutin, 8 - hyperoside, 9 - isoquercitrin, 10 - reynoutrin, 11 - avicularin, 12 - quercitrin, 13 - phloridzin. The highest total amount of quercetin glycoside 0.811 mg g-1 was determined in apple samples of cultivar Aldas. It was 1.93 times higher than the lowest total amount of quercetin glycoside (0.420 mg g-1), determined in apple samples of cultivar Auksis. Quantitative composition of quercetin glycoside group compounds determined in apple samples is presented in Table 4. Hyperoside was a predominant quercetin glycoside group compound in apple samples of cultivars Aldas, Auksis, Connel Red, Ligol and Lodel. It amounted to 25.14-35.16% of the total amount of identified and quantified quercetin glycoside group compounds. VAN DER SLUIS et al. indicated similar tendencies of a hyperoside quantitative composition variation. Hyperoside amounted to 23-33% of the total identified quercetin glycosides in samples of cultivars that these authors analysed (VAN DER SLUIS et al., 2001). The composition of apple samples of cultivar Rajka stood out among the analysed cultivars, as the predominant compound in this cultivar was quercitrin. Its amount was 1.66 times higher than the amount of hyperoside determined in the apple samples of this cultivar. Quercetin glycoside rutin-hyperoside-isoquercitrin triplet (where the predominant compound is always hyperoside, and levels of rutin are the lowest) was characteristic in apple sample extracts of all analysed cultivars. This pattern was also established in studies of other scientists (MARKS et al., 2007; PRICE et al., 1999; SCHIEBER et al., 2001). The ratio of rutin-hyperoside-isoquercitrin amount varies in the apple samples of different cultivars. It varied from 1:5.1:1.6 (cultivar Connel Red) to 1:11.9:1.2 (cultivar Lodel). Ital. J. Food Sci., vol. 30, 2018 - 98 Monomeric ((+)-catechin and (−)-epicatechin) and oligomeric (procyanidin B1, procyanidin B2, and procyanidin C1) flavan-3-ols were determined in the apple samples. The highest total amount of identified and quantified flavan-3-ols (2.919 mg g-1) was determined in apple samples of the cultivar Lodel. It was 2.55 times higher than the lowest amount (1.143 mg g-1), determined in the apple samples of cultivar Connel Red. In the scientific literature, it was noted that in apples, the amount of (−)-epicatechin is higher than (+)-catechin (WU et al., 2007; DUDA-CHODAK et al., 2010; KAHLE et al., 2005; PANZELLA et al., 2013). The results of our analysis also confirm this. The ratio of (+)- catechin and (−)-epicatechin in the samples of different cultivars varied from 1:4.3 (cultivar Rajka) to 1:21.9 (cultivar Lodel). In the paper by WOJDYŁO et al., it is specified that in samples of apples grown in Poland, the amount of (+)-catechin varies from 0.010 to 0.720 mg g-1, and (−)-epicatechin − from 0.066 to 2.760 mg g-1 (WOJDYŁO et al., 2008) The amounts of procyanidin B2 and C1 were higher than the amount of procyanidin B1 was in the samples of all analysed cultivars. This predisposition of the quantitative composition variation of these compounds in apples was also presented by other authors (DUDA-CHODAK et al., 2011). The ratio of procyanidin B1, B2 and C1 amounts in samples of different cultivars varied from 1:5.4:3.4 (cultivar Aldas) to 1:13.8:7.9 (cultivar Auksis). The quantitative composition of flavan-3-ol group compounds identified in apple samples is presented in Table 5. Qualitative and quantitative analyses of dihydrochalcone group compounds are extremely important because the compounds of this group may be selected as chemotaxonomic indicators in apple cultivar taxonomy, for apple product identification and for the determination of apple juice and cider quality (SCHIEBER et al., 2001; ALONSO-SALCES et al., 2004; GOSCH et al., 2010). The amount of dihydrochalcone phloridzin in the apple samples of the analysed cultivars varied from 0.101 mg g-1 (cultivar Rajka) to 0.268 mg g-1 (cultivar Lodel) (Table 5). Similar results were presented by other authors as well-the amount of phloridzin in apple fruit comprises 2-6% of the total amount of quantified phenolic compounds (SANONER et al., 1999) The highest amount of chlorogenic acid (3.074 mg g-1) was determined in the apple samples of the cultivar Aldas. It was 4.99 times higher than the lowest determined amount of this acid (0.616 mg g-1), determined in the apple samples of the cultivar Rajka (Table 6). WOJDYŁO et al. analysed the apple samples of apples grown in Poland and indicated similar amounts (0.015-2.960 mg g-1) of this acid (WOJDYŁO et al., 2008). Ital. J. Food Sci., vol. 30, 2018 - 99 Table 4. Quercetin glycoside quantitative variation in apple samples. Substance Quercetin glycoside amount (mg g-1, dry raw material) Coefficient of variation, % Aldas Auksis Connel Red Ligol Lodel Rajka Hyperoside 0.274±0.005a 0.138±0.002c,d 0.152±0.003c 0.129±0.002d 0.154±0.003c 0.184±0.004b 31.10 Isoquercitrin 0.053±0.002a 0.020±0.001c,d 0.047±0.002a,b 0.025±0.001c 0.016±0.001d 0.044±0.002b 45.92 Rutin 0.036±0.001a 0.015±0.001b,c 0.030±0.002a 0.022±0.001b 0.013±0.001c 0.035±0.002a 39.69 Avicularin 0.226±0.005a 0.120±0.002c 0.099±0.001d,e 0.087±0.001e 0.114±0.002c,d 0.139±0.003b 38.15 Reynoutrin 0.080±0.003a 0.053±0.002b 0.039±0.001c 0.037±0.001c 0.043±0.001c 0.025±0.001d 40.92 Quercitrin 0.142±0.003b 0.082±0.001e 0.103±0.002c,d 0.120±0.003c 0.098±0.001d,e 0.305±0.007a 58.30 Different letters in the same row indicate statistically significant differences of individual substance amounts in apple samples of analysed cultivars (p<0.05). Table 5. Flavan-3-ols, chlorogenic acid and phloridzin quantitative variation in apple samples. Substance Flavan-3-ols, chlorogenic acid and phloridzin amount (mg g-1, dry raw material) Coefficient of variation, % Aldas Auksis Connel Red Ligol Lodel Rajka (+)-Catechin 0.092±0.002b 0.077±0.002c 0.034±0.001d 0.033±0.001d 0.042±0.001d 0.121±0.003a 54.26 (–)-Epicatechin 0.720±0.015b 0.448±0.009d 0.315±0.006e 0.311±0.004e 0.928±0.019a 0.525±0.010c 44.86 Procyanidin B1 0.157±0.003a 0.170±0.004a 0.035±0.001d 0.064±0.001c 0.094±0.002b 0.097±0.002b 50.84 Procyanidin B2 0.854±0.017c 0.990±0.021b 0.484±0.007e 0.676±0.014d 1.279±0.023a 0.834±0.015c 31.81 Procyanidin C1 0.534±0.020a 0.366±0.012b 0.275±0.008c 0.258±0.011c 0.576±0.021a 0.249±0.007c 38.56 Phloridzin 0.188±0.004b 0.124±0.002d 0.135±0.002d 0.157±0.003c 0.268±0.005a 0.101±0.002e 36.82 Chlorogenic acid 3.074±0.068a 2.498±0.052b 0.773±0.013e 1.249±0.020d 1.629±0.035c 0.616±0.011e 59.41 Different letters in the same row indicate statistically significant differences of individual substance amounts in apple samples of analysed cultivars (p<0.05). Ital. J. Food Sci., vol. 30, 2018 - 100 The study results confirmed the hypothesis of CEYMANN et al., that apple cultivars can be classified based on what types of compounds phenolic acids or flavan-3-ols are predominant in the apple samples (CEYMANN et al., 2012). The predominant compound in the apple samples of the cultivars Aldas and Auksis was chlorogenic acid, and in the apple samples of other cultivars, it was flavan-3-ol group compounds. The coefficient of variation, which reflects the variation amplitude of every compound, was calculated to evaluate the variation of the quantitative composition of the phenolic compounds in the apple samples of different cultivars. It varied from 31.10% to 59.41% (Tables 5 and 6). The highest calculated coefficient of variation was for chlorogenic acid, and the lowest was for hyperoside. 4. CONCLUSIONS In this study, we applied RSM for the extraction of flavonoids from lyophilised apples. The results showed that the independent variables (temperature, extraction time and ultrasonic power), and the quadratic terms of temperature and extraction time had a statistically significant effect on the efficacy of apple flavonoid extraction. A second-order (quadratic) polynomial model was employed to optimize flavonoid extraction from lyophilised apple samples. The projected optimal extraction conditions following statistical modelling were as follows: temperature 44.61ºC, extraction time 26.90 min and ultrasonic power 480 W. The experimental yield of flavonoids was 6.58 mg RE g-1, which was close to the predicted yield value of flavonoids 6.69 mg g-1. By applying RSM selected via statistical modelling, it was determined that the apple flavonoid extraction yield was 37.12% higher compared with sample extraction when extraction conditions were selected empirically. Optimised extraction conditions were applied for the HPLC analysis of the apple samples of six cultivars grown in Lithuania. The highest total amount of identified phenolic compounds (6.430 mg g-1) was determined in the apple sample extracts of the cultivar 'Aldas'. REFERENCES Alonso-Salces R.M., Herrero C, Barranco A., Berrueta L.A., Gallo B. and Vicente F. 2004. Technological classification of Basque cider apple cultivars according to their polyphenolic profiles by pattern recognition analysis. J. Agric. Food Chem. 52:8006-8016. Canettieri E.V., de Moraes Rocha G.J., de Carvalho J.A. and de Almeida e Silva J.B. 2007. Optimization of acid hydrolysis from the hemicellulosic fraction of Eucalyptus grandis residue using response surface methodology. Bioresour. Technol. 98:422-428. Ceymann M., Arrigoni E., Schärer H., Nising A.B. and Hurrell R.F. 2012. Identification of apples rich in health- promoting flavan-3-ols and phenolic acids by measuring the polyphenol profile. J. Food Comp. Anal. 26:128-135. Chen W., Wang W.P., Zhang H.S. and Huang Q. 2012. Optimization of ultrasonic-assisted extraction of water-soluble polysaccharides from Boletus edulis mycelia using response surface methodology. Carbohydr. Poly. 87:614-619. Cooke M.S., Evans M.D., Dizdaroglu M. and Lunec J. 2003. Oxidative DNA damage: mechanisms, mutation, and disease. FASEB J. 17:1195-1214. Duda-Chodak A., Tarko T. and Tuszyński T. 2011. Antioxidant activity of apples-an impact of maturity stage and fruit part. Acta Sci. Pol. Technol. Aliment. 10:443-454. Duda-Chodak A., Tarko T., Satora P., Sroka P. and Tuszyński T. 2010. The profile of polyphenols and antioxidant properties of selected apple cultivars grown in Poland. J. Fruit Ornam. Plant Res. 18:39-50. Ital. J. Food Sci., vol. 30, 2018 - 101 European Pharmacopoeia. 2010. vol. 1, 7th ed. Council of Europe, Strasbourg, France. FAO. FAO Statistical Database. 2013. Retrieved from http://www.fao.org/faostat/en/#data/QC (Accessed 23-05-2017). Gosch C., Halbwirth H. and Stich K. 2010. Phloridzin: biosynthesis, distribution and physiological relevance in plants. Phytochemistry. 71:838-843. Kahle K., Kraus M. and Richling E. 2005. Polyphenol profiles of apple juices. Mol. Nutr. Food Res. 49:797-806. Łata B., Przeradzka M. and Binkowska M. 2005. Great differences in antioxidant properties exist between 56 apple cultivars and vegetation seasons. J. Agric. Food Chem. 53:8970-8978. Liaudanskas M., Viškelis P., Jakštas V., Raudonis R., Kviklys D., Milašius A. and Janulis V. 2014. Application of an optimized HPLC method for the detection of various phenolic compounds in apples from Lithuanian cultivars. J. Chem., 2014:Article ID 542121, 10 pages. Madamanchi N.R., Vendrov A. and Runge M.S. 2005. Oxidative stress and vascular disease. Arterioscler. Thromb. Vasc. Biol. 25:29-38. Marks S.C., Mullen W. and Crozier A. 2007. Flavonoid and chlorogenic acid profiles of English cider apples. J. Sci. Food Agr. 87:719−728. Muralidhar R.V., Chirumamila R.R., Marchant R. and Nigam P. 2001. A response surface approach for the comparison of lipase production by Candida cylindracea using two different carbon sources. Biochem. Eng. J. 9:17-23. Pandey K.B. and Rizvi S.I. 2009. Plant polyphenols as dietary antioxidants in human health and disease. Oxid. Med. Cell Longev. 2:270-278. Panzella L., Petriccione M., Rega P., Scortichini M., and Napolitano A. 2013. A reappraisal of traditional apple cultivars from Southern Italy as a rich source of phenols with superior antioxidant activity. Food Chem.140:672-679. Pizzimenti S., Toaldo C., Pettazzoni P., Dianzani M.U. and Barrera G. 2010. The “two-faced” effects of reactive oxygen species and the lipid peroxidation product 4-hydroxyno-nenal in the hallmarks of cancer. Cancers. 2:338-363. Price K.R., Prosser T., Richetin A.M. and Rhodes M.J.C. 1999. A comparison of the flavonol content and composition in dessert, cooking and cider-making apples; distribution within the fruit and effect of juicing. Food Chem. 66:489-494. Radojkovic M., Zekovic Z., Jokic S., Vidovic S., Lepojevic Z. and Milosevic S. 2012. Optimization of solid-liquid extraction of antioxidants from black mulberry leaves by response surface methodology. Food Technol. Biotech. 50:167- 176. Reuter S., Gupta S.C., Chaturvedi M.M. and Aggarwal B.B. 2010. Oxidative stress, inflammation, and cancer: how are they linked? Free Radic. Biol. Med. 49:1603-1616. Sanoner P., Guyot S., Marnet N., Molle D. and Drilleau J.F. 1999. Polyphenol profiles of French cider apple varieties (Malus domestica sp.). J. Agric. Food Chem. 47:4847-4853. Schieber A., Keller P. and Carle R. 2001. Determination of phenolic acids and flavonoids of apple and pear by high- performance liquid chromatography. J. Chromatogr. A. 910:265-273. Tian Y., Xu Z., Zheng B., and Lo Y.M. 2013. Optimization of ultrasonic-assisted extraction of pomegranate (Punica granatum L.) seed oil. Ultrason. Sonochem. 20:202-208. Urbonavičiūtė A., Jakštas V., Kornyšova O., Janulis V. and Maruška A. 2006. A. Capillary electrophoretic analysis of flavonoids in single-styled hawthorn (Crataegus monogyna Jacq.) ethanol extracts. J. Chromatogr. A. 1112:339-344. van der Sluis A.A., Dekker M.A., Jager A. and Jongen W.M.F. 2001. Activity and concentration of polyphenolic antioxidants in apple:effect of cultivar, harvest year, and storage conditions. J. Agric. Food Chem. 48:3606-3613. Wojdyło A., Oszmiański J and Laskowski P. 2008. Polyphenolic compounds and antioxidant activity of new and old apple varieties. J. Agric. Food Chem. 56:6520-6530. Wu J., Gao H., Zhao L., Liao X., Chen F., Wang Z. and Hu X. 2007. Chemical compositional characterization of some apple cultivars. Food Chem. 103:88-93. Paper Received March 28, 2017 Accepted September 8, 2017