Optimization of the extraction of natural antioxidants from Coffea robusta leaves and evaluation of their ability to preserve palm olein from oxidation during accelerated storage

Abstract Response surface methodology (RSM) was used to optimize the extraction of phenolic antioxidants of Coffea robusta leaves and to evaluate the effect of optimized extract and storage time on the stability of palm olein. The optimization of the extraction process was conducted, and the total polyphenol value of 127.06 mg GAE/g and scavenging activity of 90.65% were obtained under optimal extraction conditions. The phenolic antioxidants of the optimized extract and their thermal stability were determined using HPLC‐DAD (high‐performance liquid chromatography‐diode array detector) and Rancimat test, respectively. The effect of concentration of the optimized extract and storage time on the stability of palm olein was also evaluated. Results showed that the optimized extract contains gallic acid, vanillic acid, cafeic acid and was efficient in retarding palm olein oxidation during 32 months at room temperature. Coffea robusta can be recommended as good source of antioxidants for stabilization of palm olein.


| INTRODUC TI ON
Oils or fats undergo many transformations and reactions during processing and storage. These changes are favored by many factors including polyunsaturated fatty acid composition, heat, light, oxygen contact, and moisture (Choe & Min, 2006). Autooxidation is the major cause of oil deterioration during storage. During this process, lipid alkyl radicals are formed and react with oxygen to form hydroperoxides which are broken down into secondary oxidation products (Choe & Min, 2006). These oxidation products have been demonstrated as related to mutations, cancers, and cardiovascular diseases (Kubow, 1992;Mc Clements & Decker, 2000).
Synthetic antioxidants such as butylated hydroxytoluene (BHT), butylated hydroxyanisole (BHA), and tertiary butyl hydroquinone (TBHQ) are usually used as ingredient in the food processing sectors to retard oxidation (Allen & Hamilton, 1994). Their role is to inhibit the development of oxidative rancidity in oils. Health issues about the toxicity of synthetic antioxidants present in foods are forcing the food industry to replace these additives with natural ones that are perceived to be "safer" (Krishnaiah, Sarbatly, & Nithyanandam, 2010). Therefore, investigation of natural products has been a major research interest in screening plant materials for possible antioxidant potential. Studies have shown that antioxidant activity of plant extracts is mainly contributed by their phenolic compounds (Djikeng et al., 2017;Jorge, Heliodoro De La Garza, Alejandro, Ruth, & Noé, 2013;Womeni, Tonfack, Anjaneyulu, et al., 2016). However, extraction conditions might affect the amount of phenolic antioxidant extracted. The choice of extraction solvents is critical for complex matrices because the physicochemical properties of the solvents, particularly its polarity, exert an influence on the yields and types of phenolic compound extracted (Balasundram, Sundram, & Samman, 2006). In addition to the solvents used, other factors such as temperature and extraction time also affect the optimization of extraction yield (Luthria, Mukhopadhyay, & Kwansa, 2006;Tomsone, Kruma, & Galoburda, 2012). Moreover, during storage at ambient temperature, the concentration of the extract and storage time can influence the capacity of the extract to retard lipids oxidation (Maleki, Ariaii, & Fallah, 2016;Womeni, Tonfack, Anjaneyulu, et al., 2016).
The objective of this research was to optimize the extraction process of phenolic antioxidants in coffee leaves and to evaluate the effect of optimized extract and storage time on the oxidative stability of palm olein during storage. Palm olein was used in this study because it is the most produced and consumed refined oil in Cameroon. Response surface methodology (RSM) was used for the process optimization. This method establishes a multivariable mathematic model to obtain the relationship between responses and independent variables (Goupy & Creighton, 2006) with the use of a minimal number of experiments.

| Materials
Refined, bleached, and deodorized palm olein (RBD Palm olein), without additives, was obtained from SCS/RAFCA Palm Oil Industry Company Ltd, Bafoussam, West-Cameroon. Coffea robusta leaves were harvested at Bafang, Haut-Nkam Division, West-Cameroon, in January 2015. All the chemicals and reagents used were of analytical reagent grade.

Experimental design
The optimum extraction conditions of polyphenols from Coffea robusta leaves were evaluated by Central Composite Design (CCD).
The real and coded levels of the various parameters used are shown in Table 1. The intervals of these independent variables were determined on the basis of preliminary studies. A total of 16 experimental runs with two replications at the center point were completed, and the total polyphenols (mgGAE/g) and antioxidant activity (% inhibition) of coffee leaves expressed as the dependent variables were determined (Table 2). This experimental design generates a seconddegree polynomial model (Y) of the form presented in Equation 1: where Y represents the response variables (total polyphenols or antioxidant activity); x 1 , x 2 , and x 3 are the levels of the independent variables; a, b, and c are the linear terms; e, f, and h are the interaction terms; d, g, and i are the quadratic terms and I is a constant.
To confirm or validate the optimum conditions of polyphenol antioxidants extraction, two experimental replicates were performed under optimized conditions. The experimental and predicted values were compared.

Extraction of phenolic compounds from coffee leaves
The extraction procedures were carried out randomly and in accordance with the conditions set by the experimental design. Fresh Coffea robusta leaves were dried in an electric oven (Venticell, MMM, Einrichtungen, Germany) at 50°C for 48 hr. The dried leaves were grounded to pass through a 1-mm sieve. Sample powder (7 g) was blended with 150 ml of solvent (Methanol/water) of concentration specified by the full factorial design ( Table 2). The mixture was placed in an electric oven (Venticell, MMM, Einrichtungen, Germany) and regularly subjected to agitation (400 rpm) at the required temperature and time specified by the experimental design ( Table 2).
The filtrate was concentrated on a rotary evaporator (BUCHI, Pharma and Biotech, Germany) at 45°C before being stored at 4°C for further analysis.

Determination of total phenolic compounds (TPC)
The total phenolic of the extracts was evaluated using the Folin-Ciocalteu colorimetric method as described by Gao, Ohlander, Jeppsson, Björk, and Trajkovski (2000). Briefly, plant extract (20 μl) was added in a test tube containing 2 ml of distilled water and 0.2 ml of Folin-Ciocalteu reagent and incubated for 3 min at room temperature. One (1) ml of 20% sodium carbonate was added to the mixture and re-incubated for 20 min at 40°C. The absorbance of the resulting blue color was determined at 765 nm using a spectrophotometer (HELIOS Epsilone, Dreieich, Germany). The standard curve prepared from Gallic acid solution was used to express the results as gallic acid equivalents (GAE) per gram of extract. (1) TA B L E 1 Coded and real levels of independent variables used in the RSM design for the optimization process of extraction

| Rancimat test
Rancimat test is used for the evaluation of the antioxidant potential of molecules to limit oxidation of oils and fats. Oil samples used for this test were prepared according to the modified method of Iqbal, Haleem, Akhtar, Zia-ul-Haq, and Akbar (2008). The optimized extract was separately added to preheated RBD palm olein (at 50°C for 3 hr) at concentrations of 500, 720, 1250, 1780, and 2000 ppm. The efficacy of natural antioxidants was evaluated by comparing their antioxidation activity with those of butylatedhydroxytoluene (BHT) employed at it legal limit of 200 ppm (Duh & Yen, 1997). Palm olein without additives and prepared under the same conditions served as control.
Induction periods of stabilized (oil containing the optimized extract) and control oil samples were evaluated using an automated Metrohm Rancimat (model 892, Germany) as described by Womeni, Tonfack, Anjaneyulu, et al. (2016). The time elapsed from the beginning until the oil starts to become rancid (induction period) was automatically recorded by the instrument. The protection factor was calculated using the induction time of oil with antioxidant (I) and the induction time of oil without antioxidant (I 0 ).

Experimental design
The effects of process parameters (concentration of extract and storage time) on the oxidative stability of palm olein were evaluated by Central Composite Design (CCD). Real and coded levels of the independent variables used are shown in Table 3, and the intervals were determined on the basis of other studies (Womeni, Tonfack, Anjaneyulu, et al., 2016;Womeni, Tonfack, Iruku, et al., 2016) and preliminary test. The design consisted of 10 runs with two replicates at the center point, and the peroxide, p-Anisidine, and total oxidation (TOTOX) values assays expressed as the dependent variables were determined (Table 4). A full quadratic model was used for fitting of data (Equation 3): where Y ' represents the response variables (peroxide value, p-Anisidine value and TOTOX value); x 1 and x 2 are the levels of the independent variables; a and b are the linear terms; e is the interaction term; c and d are the quadratic terms and I is a constant.

Sample preparation
The optimized extract was dissolved and separately added to 60 g of preheated RBD palm olein (at 50°C for 3 hr) at concentrations indicated by the full factorial design (Table 4). Stabilized oil samples were placed in dark brown glass bottles with narrow necks   Table 2 shows the results of the antioxidant activity (% Inhibition) and total polyphenols (mgGAE/g) of coffee leaves extract. The higher phenolic content registered in this study can be attributed particularly to the solvent used for extraction (Ghumman, Singh, & Kaur, 2017). In fact, methanolic extracts have been reported as exhibiting highest TPC and antioxidant activity compared to ethanol and acetone (Sulaiman, Sajak, Ooi, Supriatno, & Seow, 2011).

| Analysis of variance and regression equations
The experimental design has been formulated to develop an empirical model to investigate the interaction of different associated independent variables responsible for the extraction of phenolic antioxidants present in coffee leaves and also, to identify the optimum conditions of extraction. The analysis of variance (ANOVA) presented in Table 5 showed that solvent (methanol/water mixture) is the only factor in the linear terms that significantly affect (p < 0.05) the total phenolic content and antioxidant activity. Methanol/water in quadratic term had significant (p < 0.05) effect on TPC. Results also showed that the coefficient of determination (R 2 value) of the responses is within the range of a good set (more than 0.75) (Joglekar & May, 1987), indicating that the model could explain adequately up to 89.5% and 84.9% of TPC and antioxidant activity, respectively.
The mathematical models of relationship for total phenolic content (Y 1 ) and antioxidant activity (Y 2 ) with temperature (X 1 ), time (X 2 ) and methanol fraction (X 3 ) are given by the Equations 4 and 5:

| Analysis of response surfaces
According to the equation, the effect of the three independent variables on the total phenolic content ( Figure 1   This shows that the models obtained can be accepted and used to prepare Coffea robusta leaves extract with the best phenolic content and high antioxidant activity.

| HPLC-DAD analysis of Coffea robusta leaves extract
Phenolic compound profile in coffee leaves extract was determined by HPLC-DAD. As shown in Figure 3, the peaks 1, 2, and 3 were found to be matching well with gallic acid (retention time: 7.910), vanillic acid (retention time: 9.591), and cafeic acid (retention time: 9.995), respectively. The presence of these phenolic compounds in Coffea robusta extract has not been reported in other studies.

| Rancimat test for verification of antioxidation activity of Coffea robusta leaves extract
The Rancimat test was performed to evaluate the efficiency of the extract in delaying palm olein oxidation. The effects of different concentrations of extract on the protection factors and induction periods of palm olein in comparison with the oil supplemented with BHT (synthetic antioxidant) and control (palm olein free from additives) are presented in Table 7.

| Effect of extract concentration and storage time on the oxidative stability of palm olein during storage
The peroxide, p-Anisidine, and total oxidation values of palm olein supplemented with coffee leaves extract obtained from the 10 experiments are presented in Table 4. Concentration of extract and storage time was taken as independent variables. The fresh RBD palm olein free from additives (experiment N°10) was of good quality, as shown by its low peroxide value (<10 ppm), low p-Anisidine value (≤ 20), and low TOTOX value (<26) as recommended by homologation (Codex Alimentarius, 1999Alimentarius, , 2015.

| Analysis of variance
The experimental data were used to calculate the coefficients of the second-order polynomial equation, to establish the coefficient of determination (R 2 ) and significant effect of independent variables (Table 8). The coefficient of determinations (R 2 ) for peroxide, p-Anisidine and TOTOX values being 0.985, 0.886, and 0.987, respectively; and falling within a good range (more than 0.75) (Joglekar & May, 1987). This means that the observed model is able to explain 98.5%, 88.6%, and 98.7% of the results in the case of peroxide value,

| Analysis of contour plots
The Peroxide value (PV) is a good indicator of the extent of primary oxidation products in oil (Anwar, Siddiq, Iqbal, & Asi, 2007).
It measures hydroperoxides of oils and according to the Codex Alimentarius (1999), and palm olein is generally recognized as safe if the PV < 10 ppm. As presented Figure  quently increase its oxidative stability (Gordon, 1990). According to the Codex alimentarius (1999), it would be advisable to heat at 70°C palm olein enriched with coffee leaves extract at 2000 ppm for 25 days in order to preserve it quality.
The measurement of p-Anisidine value is intensively used to assess secondary oxidation products like 2-alkenal, 2,4-alkadienal (Anwar, Qayum, Hussain, & Iqbal, 2010). Contour plot (Figure 4b) shows that these responses increase slightly with the storage time of the oil at 70°C. However, the addition of the extract at high concentrations does not stop the production of secondary oxidation products. In fact, hydroperoxides are thermolabile molecules which easily breakdown into secondary oxidation products at high temperature (>120°C). Codex Alimentarius (2015) recommends a p-Anisidine value lower than 20 for good quality fish oil. Considering this standard, it can be observed that this response was lower than 20 during the entire treatment.
Total oxidation value measures both primary and secondary oxidation products and provides a better determination of the progressive oxidative deterioration of oils (Womeni, Tonfack, Iruku, et al., 2016). Figure 4c illustrates  The Schaal oven test (storage at 65-70°C) is a good simulation of normal storage conditions. Evans, List, Moser, and Cowan (1973) showed that heating of oils for 08 hr at 65°C is equivalent to 1-month storage at room temperature. During storage of palm olein at 70°C, Coffea robusta leaves extract at 2000 ppm has efficiency to preserve its quality during 32 days (08 h per day).
Considering assertion of Evans et al. (1973), we can conclude that the supplementation of palm olein with 2000 ppm of Coffea robusta leaves extract can extend its preservation to 32 months at room temperature.

| CON CLUS ION
Based on the experimental design that had been performed using response surface methodology, the optimal conditions for the extraction of antioxidant phenolic compounds from Coffea robusta leaves extract were determined (extraction at 53.70°C with incubation time of 5.60 hr, using methanol fraction of 79.66%). This extract under optimal conditions contains several phenolic compounds and has a good thermal stability. Oxidation of palm olein increases with storage time, and coffee leaves extract have the capacity to delay this alteration reaction and stabilize the oil. This study can be useful for the development of industrial extraction process of coffee leaves and its application as an ingredient to delay lipid oxidation in oils.

CO N FLI C T O F I NTE R E S T
None.

E TH I C A L S TATEM ENT
Humans and animals testing is not applicable to this study.