^{1}

^{2}

^{2}

^{3}

^{3}

^{2}

^{2}

The current work concerns the optimization process of phenolic compounds solid liquid extraction from grape byproducts at high temperatures and short incubation times. The effect of five experimental parameters (solidliquid ratio, particle size, time, temperature and solvent mixture) mostly believed to affect the extraction process was undertaken. A first response surface methodology experimental design was used to optimize the solid-liquid ratio and milling time parameters. A second design was used for the optimization of the quantitative and qualitative parameters. The quantitative parameters studied are: total phenolic compounds, flavonoid content, total monomeric anthocyanin composition and tannin concentration. The qualitative parameters analyzed are: antiradical activity and antioxidant capacity. The second design was based on the use of time, temperature and solvent mixture as optimization parameters. The assays were first conducted separately revealing the best experimental conditions for the maximization of each response variable alone. A simultaneous response surface methodology of all the responses taken together was then conducted, showing the optimal extraction conditions to be: 93 minutes at 94?C and in 66% ethanol/water solvent. The maximal response values obtained for each parameter are: Total Phenolic Compounds yield (5.5 g GAE/100g DM), Flavonoid Content (5.4 g GAE/100g DM), Total Monomeric Anthocyanin yield (70.3 mg/100g DM), Tannin Concentration (12.3 g/L), Antiradical Activity (67.3%) and Total Antioxidant Capacity (393 mgAAE/L). All of the optimal values were acquired at 3 mL/g solid-liquid ratio and 6.8 min milling time. The obtained extracts could be used as natural bioactive compounds in several industrial applications.

The annual production of large waste quantities by the food processing industry creates serious environmental problems as a consequence of the absence of efficient policies regarding their disposal. Many processes are being established, targeting the conversion of waste materials into bio-fuels, food ingredients and other added value bio-products [

All reagents were of analytical grade. The Folin’s phenol reagent (SCOTT SCIENCE UK) and sodium carbonate (Fluka, Buchs, Switzerland) were utilized to measure the total phenolic compounds concentrations using the FolinCiocalteu method; the calibration curve was built with gallic acid (Sigma Chemical Co., St. Louis, MO, USA).

The Cabernet Sauvignon grape byproducts were provided by château KSARA (Beqaa Valley, Lebanon). On arrival the raw material was stored at −20˚C. Defrosted at room temperature, the grape byproducts were milled to fit the required particle size. After the solid-liquid extraction process with the heated solvent and under agitation, solids were separated by filtration [

According to the Folin-Ciocalteu method previously described by Slinkard and Singleton [

The indirect method of flavonoid determination was performed as described by Ough and Amerine [

Monomeric anthocyanins were measured by the pH-differential method, which relies on the structural transformation of the anthocyanin chromophore as a function of pH, which can be measured using optical spectroscopy [_{vis-max} and at 700 nm against a blank cell filled with distilled water. The absorbance (A) of the diluted sample was calculated as follows:

A = (A_{λ}_{vis-max} − A_{700})_{pH1} − (A_{λ}_{vis-max} − A_{700})_{pH4.5}(1)

The monomeric anthocyanin pigment (MAP) concentration in the original sample was calculated using the following formula:

MAP_{(mg/L)} = (A × MW × DF × 1000)/(molA × L)(2)

MW and molA are the molecular weight and the molar absorptivity, respectively of the pigment cyanidin-3-glucoside used as reference; MW = 449.2 g/mole, molA = 26,900 mg^{−1}·L^{−1}·cm^{−1} and DF is the dilution factor. Milligrams of Monomeric Anthocyanin per liter of extract (mg/L) were then transformed into Total Monomeric Anthocyanin yield (TMA) which is milligrams per 100 grams of grape dry matter (mg/100g DM).

Total tannin content (g/L) was determined according to Ribérau-Gayon et al. [

The total antioxidant activity of the extracts was determined by the phosphomolybdenum reduction assay [

According to Kallithraka et al. [

The quantity and quality of the phenolic compounds extracts are affected by several factors. Regarding the incapacity of identifying all parameters effects at the same time, it was necessary to group the parameters into two experimental designs.

A central composite design (2^{2} + star) was established to assess the main effect of two factors in 12 runs. The influence of particle size and solid-liquid ratio on the phenolic compounds extraction from grape pomace was studied. Particle size was estimated by the duration of the milling process going from 2 to 6 minutes. Solid-liquid ratio (L/S) varied from 2.31 mL/g to 8.69 mL/g. Time, temperature and solvent mixture were fixed to 24 hours, room temperature, and 70% ethanol/water solvent. The two independent variables were coded at five levels (−α_{1}, −1, 0, 1, α_{1}) resulting in an experimental design of twelve experimental points including four repetitions at the central points. The optimization process by response surface methodology, took into consideration particle size and solid-liquid ratio as two independent variables. Considering two parameters and a response, experimental data were fitted to obtain a second-degree regression equation of the form:

Y = β_{0} + β_{1}T + β_{2}R + β_{12}T·R + β_{11}T^{2} + β_{22}R^{2}(3)

where Y is the predicted response parameter, T is the time of the milling process representing the particle size, R is the solid-liquid ratio, β_{0} is the mean value of response at the central point of the experiment; β_{1} and β_{2 }are the linear coefficients, β_{11} and β_{22} the quadratic coefficients and β_{12} the interaction coefficient. The values of independent variables where the response TPC is the highest enables the identification of the optimal extraction conditions for the maximization of the response. Experimental design and statistical treatment of the results were performed using STATGRAPHICS Plus 4.0 for Windows.

The choice of the time and temperature intervals was the result of a preliminary study in which phenolic compounds extraction from milled grape pomace was conducted at high temperatures and short periods of time. The total phenolic content and the free radical scavenging activity were determined after 30, 60, and 90 minutes. Based on the results, the lower and upper levels of both variables were chosen for the Response Surface Methodology. For the optimization process of time, temperature and solvent mixture, a Central composite design (2^{3} + star) was created to study the effects of 3 factors in 20 runs. Time varied from 40 to 99 minutes, temperature from 40˚C to 80˚C and solvent mixture from 30% to 80% ethanol/water (v/v). The three independent variables were coded at five levels (−α_{2}, −1, 0, 1, α_{2}) resulting in an experimental design of twenty experimental points including six central points. Six responses where studied: Total phenolic compounds yield (TPC), flavonoid content (FC), total monomeric anthocyanin content (TMA), tannin concentration (TC), antiradical activity (AA), and total antioxidant capacity (AC). The values of independent variables where the response variables are the highest enables the identification of the optimal extraction conditions for the maximization of the responses. Considering two parameters and six responses, experimental data were fitted to obtain a second-degree regression equation of the form:

where Y is the predicted response X_{n} and X_{m} are the coded values for the factors, b_{0} is the mean value of the responses at the central point of the experiment; b_{n}, b_{nn}, and b_{nm} are respectively the linear, quadratic, and interaction coefficients. A multi-response surface optimization was effected to maximize all the responses at the same time. Experimental design and statistical treatment of the results were performed using STATGRAPHICS Plus 4.0 for Windows.

A first response surface methodology study was conducted in the aim of determining the adequate solid-liquid ratio and particle size for the optimization of phenolic compounds extraction from grape pomace. The response values (TPC) at different variable combinations and statistical analyses showed that all response values fitted best the second order polynomial model, which the correspondent equation is shown as follows:

where T is the milling time and R the solid-liquid ratio.

The model had a satisfactory level of adequacy (R^{2} = 84%), indicating a reasonable agreement of the corresponding model with the experimental results. Statistically considered as significant, the solid-liquid ratio has a negative linear and quadratic effect on the PCY. Pinelo et al. [

Meyer [

In the purpose of determining the experimental conditions for the optimization of TPC, FC, TMA and TC so as for the maximization of the bioactivity represented by the Antiradical Activity (AA) and the Antioxidant Capacity (AC), a response surface methodology study was performed using a rotatable central composite design. Using ethanol/water mixtures, the TPC ranged from 2.1 to 5.8 g GAE/100g DM, the FC from 2.1 to 5.8 g GAE/100g DM, the TMA from 28.6 to 72.5 mg/100g DM, the TC from 4.8 to 13.7 g/L, the AA from 47.9 to 67.3%, and the AC from 310.7 to 404.5 mg AAE/L. Considering the variability of the extraction parameters (time, temperature, solvent mixture), the starting material (seeds, skins, pomace) and its pre-treatment, a comparison of our study with all other published results could not be entirely done. Nevertheless a general evaluation with

other extraction processes from grapes, grape parts and byproducts could give a clear idea about the extraction efficiency. Concerning the TPC extracted from grape pomace, Spigno and De Faveri [^{−2} and 5% ethanol as modifier, for the extraction of phenolic compounds from grape (Vitis labrusca B.) peel. Thus, our results taken all together were found to be in agreement with several previous published works.

All response values were demonstrated by statistical analyses to fit best the second order polynomial equations expressing the relation between the experimental parameters and the response variables (Data not shown). The regression models permitted the calculation of the predicted values, analyzed for the calculation of the coefficients of determination (R^{2}). The models had high and satisfactory levels of adequacy shown by the closeness to 1 of the (R^{2}) values (0.918, 0.92, 0.813, 0.863, 0.86 and 0.866 for TPC, FC, TMA, TC, AA and AC, respectively), indicating a high degree of correlation between all observed and predicted values. This means that a reasonable agreement of the corresponding model with the experimental results was found.

The response surface plots shown in Figures 2, 3 and 4 give by their shapes, information about the significance of each experimental parameter. It can be noticed from Figures 2(a) and (b) that temperature had a positive linear effect on TPC since it increased with temperature increase to reach an optimum of 94˚C. The enhancing capacity of the temperature parameter on the extraction efficiency of phenolic compounds was reported by many authors [5,8,21]. It ameliorates the mass transfer, improves the solubilization of the solutes in the solvent and reduces the surface tension and viscosity [

the temperature enhanced the extraction process but with relatively long periods the effect is inverted, and the phenolic compounds are threatened by oxidation or degradation [

optimal solvent mixture was found to be 63% ethanol/ water. Spigno et al. [

Flavonoids extraction was reported to be affected by many parameters such as time, temperature, ethanol concentration, solid-liquid ratio and extraction cycles [39- 41]. Herein, temperature had a positive linear effect on FC (Figures 2(d) and (e)). Temperature increase leads to FC increase to reach an optimum of 94˚C, identically to the TPC. Many authors showed the effect of temperature on flavonoids extraction. Sheng et al. [

Time was shown in this study to have linear and quadratic positive effects on TMA (

As shown in Figures 3(d), (e) and (f), time, temperature and solvent mixtures have negative quadratic effects on the tannin content. TC increases with time increase to reach its optimal value after 77 minutes, after which a decrease was obtained.

The same tendency of tannin augmentation was observed with temperature and ethanol concentration increase, until they reached 94˚C and 67% ethanol/water mixture, respectively. Tannin extraction from bark was patented to be preferably conducted at high temperatures, between 90˚C and 100˚C [

In this study a medium polarity of the solvent was found to be the most appropriate (64% ethanol/water) for tannin content maximization. This comes in agreement with the diverse solubilities of the components; procyanidins were shown to be soluble in the aqueous phase and catechins in the organic part of the solvent [

In order to emphasize on the biological properties of the extracts, the optimal extraction parameters for the maximization of the AA and the AC were defined.

Temperature had a quadratic positive effect on both properties (Figures 4(a), (b), (d) and (e)) and the optimal temperature for the maximization of the bioactivity was 94˚C. Chamorro et al. [

Wijngaard and Brunton [

Water extracts were reported by many authors to have lower DPPH inhibition percentages than alcoholic extracts from several natural products [

The simultaneous optimization of multiple responses is a main concern for industrial applications [

if short periods of times were adopted. Moreover, relatively high quantities of phenolics were extracted with important bioactive properties. Respecting these optimal conditions, we could save several hours, up to 18.5, for each extraction process. Nevertheless, an accurate economical evaluation of the extraction energy cost on the overall production fee is required to confirm the choice of the parameters. Furthermore, a semi-pilot followed by a pilot scale studies are necessary for the scaling up of our results on an industrial level.

RSM was revealed accurate in predicting models and optimizing several extraction conditions for the simultaneous maximization of many parameters such as temperature, thus minimizing the degradation process. A potential alternative was proposed for an industrial solid-liquid extraction process of phenolic compounds from grape pomace. We increased the incubation temperature up to 93˚C, reduced the time of the process to 93 minutes, and used 66% ethanol/water as solvent mixture. These extraction conditions reduce the energy cost and maximize simultaneously the extraction of total phenolic compounds (5.5 g GAE/100g DM), flavonoids (5.4 g GAE/ 100 g DM), anthocyanins (70.3 mg/100g DM) and tannins (12.3 g/L), retaining both the antiradical (67%) and antioxidant activities (393 mgAAE/L).

The Research Council of Saint Joseph University of Beirut funded this work (Project FS54). We are grateful to Joseph Yaghi for technical assistance and to the society Château KSARA for providing the grape pomace.