Introduction

Polyphenols, found in fruits and vegetables, are significant for manufacturing high-value natural products with antioxidant properties applicable in food, cosmetic, or pharmaceutical applications. There is a growing interest in using polyphenols as target molecules to reutilize them in food processing industries, aligning with circular economy principles. In this context, recycling biomass from food processing waste to obtain high-value polyphenols is a current and relevant topic.

The United Nations (UN) promotes research to obtain high-value extracts, infusions, potions, and oils from agri-food by-products, aligning with a new biorefinery concept. This approach, highlighted by the UN Environment Program and recent studies, represents a novel perspective on utilizing agri-food waste for obtaining industrially relevant chemical components (United Nations Environment Programme, 2009; Vuong, 2017; da Silva et al., 2023).

Extraction is pivotal in biorefinery, and the technique choice is crucial for successful outcomes (Álvarez-Sánchez et al., 2010; Kim & Verpoorte, 2010; Luque de Castro & Priego-Capote 2018). Extraction starts with sample pretreatments, where drying methods are to decrease the water activity in solid matrices and stop enzymatic reactions, maintaining chemical composition and making easy the release of metabolites (Ledesma-Escobar & Luque de Castro, 2014).

Zalewska and co-workers (2022) tested oven protocols at different temperatures and times to pretreat tomato samples, noticing that vitamin C and lycopene content, total polyphenols content, and antioxidant capacity decreased or increased in each protocol because of reactive oxidation of molecules caused by oxygen species. Also, Borchani and co-workers (2012) observed a similar tendency for functional date fiber concentrates, calling to use low temperatures to obtain those food ingredients.

Considerable progress has been made in advancing green and efficient extraction methods for the recovery of natural products. Techniques such as ultrasound-assisted extraction (UAE), microwave-assisted extraction (MAE), pressurized liquid extraction (PLE), supercritical fluid extraction (SFE), and ohmic extraction (OE) aim to reduce environmental impact and analysis time. Despite these advancements, conventional methods like Soxhlet extraction (SOX) and stirring extraction (SE) maintain their significance. Noteworthy studies, including those by Alara et al. (2018), Lama-Muñoz et al. (2020), and Karami et al. (2015), have utilized SOX to extract polyphenols from various sources, obtaining glucoside polyphenols such as 1-galloyl-β-d-glucose, 5-O-dicaffeoylquinic acid, quercetin-7-O-rutinoside, luteolin-7-O-glucoside, apigenin-7-O-glucoside, and verbascoside.

In parallel, SE has been employed by Gómez-Mejía et al. (2019), Rosas Ulloa et al. (2023) and Safdar et al. (2017) for recovering phenolic acids and flavonoids from Citrus peels, detecting aglycon phenols like gallic acid, chlorogenic acid, caffeic acid, trans-ferulic acid, coumaric acid, catechin, epicatechin, hesperidin, naringenin, mangiferin, myricetin, rutin, quercetin, and kaempferol. Meanwhile, UAE and MAE, in both conventional and focused modes, serve as alternatives to SE in a wide range of agri-food matrices. For instance, Ledesma-Escobar and Priego-Capote (2015) extracted polar and mid-polar components, including polyphenols, from Citrus fruits. The technique-dependent variations in phenolic number, type, and concentration highlight the importance of selecting the appropriate extraction method for optimal results.

Experimental conditions play a crucial role in each extraction technique, influencing efficiency and ultimately determining the optimal recovery of phenolic compounds. Extraction mainly depends on the nature of the solvent or mixture of solvents used as extractants. Methanol, ethanol, and their aqueous mixtures are the most used extractants in polyphenol extraction because of the wide range of polarities from high hydrophilic compounds like phenolic acids to hydrophobic ones as flavonoids (Delgado-Torre et al., 2012; Haminiuk et al., 2014). Sometimes, the equipment requirements, which could be unhandy or expensive, must be considered.

Mass transfer resistance is another parameter that needs control whereby the sample-extractant ratio is studied to optimize it. Overall, all extraction techniques have advantages and limitations, which could favor or hinder the extraction of phenolic compounds according to their own goals. So, according to Costa et al. (2023), proper management, development, optimization, and evaluation of promising technologies contribute to improving and selecting extraction processes of polyphenols from agri-food biomass.

Tomato (Lycopersicon esculentum P. Mill.) ranks as the second most globally distributed horticultural industry, with an annual production of approximately 180 million tons (FAO, 2019). In Mexico, where 30% of tomato production is wasted, underexploited by-products, like seed and peel, are utilized for colorant recovery and cattle feed (Kemper et al., 2019). In particular, seed and peel are rich sources of bioactive compounds. Lyu et al. (2023) demonstrated the potential of incorporating industrial whole tomato and peel powder into extruded high-moisture meat analogs, resulting in improved color, total polyphenols, antioxidant capacity, and texture—a high-value product with beneficial chemical components.

Lycopene, a carotenoid strongly associated with tomato color and antioxidant properties, is a crucial bioactive compound extracted from tomato samples. Zuorro et al. (2013) employed an enzyme-assisted stirring extraction process, using Peclyve LI multi-enzyme preparation, to recover essential oils enriched with lycopene, yielding an oleoresin with a 6.8% (w/w) lycopene content. In another innovative study, Szabo et al. (2019) enriched poly(vinyl)alcohol films with a carotenoid-rich extract of tomato pulp obtained through sonication, showcasing the tomato by-products’ value in diverse applications.

On the other hand, a limited number of investigations have shown the relevance of polyphenols, which play a relevant role in the antioxidant potential of tomato and its by-products and show good bioavailability by oral ingestion in drugs and foods (Kalogeropoulos et al., 2012; Masuoka et. al., 2012; Nour et al., 2018, Upadhyay and Mohan Rao, 2013; Wink, 2011). However, they did not explain the extraction mechanism of polyphenols from tomato by-products, where authors poorly assessed the effect of chemical factors on extraction, and there is no exploitation of the main differences between the proposed method and other ones (Andreou et al., 2020; Azabou et al., 2020; Ćetković et al., 2012; di Donato et al., 2018; Fernández et al., 2018; Kalogeropoulos et al., 2012; Perea-Domínguez et al., 2018; Szabo et al., 2018; Tranfić Bakić et al., 2019).

Despite the widespread use of conventional methods, particularly stirring extraction, numerous studies advocate for green extraction techniques such as ultrasound and microwaves. These methods are preferred for recovering bioactive compounds, such as polyphenols, due to their cost-effectiveness, eco-friendliness, speed, lower sample consumption, and minimal compound degradation. However, these studies often focus on a single extraction procedure, lacking comprehensive comparisons with other extraction methods (Moraes et al., 2022; Boateng, 2023; Dutta et al., 2023; Messinese et al., 2023).

This study focuses on evaluating the effectiveness of four novel extraction methods in obtaining antioxidant-rich polyphenol extracts from tomato by-products, specifically peels and seeds. The research aims to establish a correlation between the composition of the matrix and the extraction process by comparing total polyphenols and antioxidant capacity indices through UV–Vis spectrophotometry. In addition, it analyzes the individual concentrations of a few phenolic compounds—gallic acid (GAL), caffeic acid (CAF), chlorogenic acid (CHL), quercetin (QUE), and kaempferol (KAE)—using liquid chromatography (LC-UV and LC–MS). These compounds represent the majority of phenolic families in tomato by-products, known for their antioxidant capacity and bioavailability.

The study incorporates two drying methods for sample pretreatment (oven and commercial dehydrator) and utilizes multivariate experimental designs to assess the impact of chemical factors on polyphenol extraction. The optimal experimental conditions for the proposed Soxhlet extraction (SOX), stirring extraction (SE), ultrasound-assisted extraction (UAE), and microwave-assisted extraction (MAE) methods are determined. Principal component analysis and the Pearson correlation test are employed to discriminate among extraction methods and correlate phenolic metabolite concentrations with total polyphenols and antioxidant capacity indices. This comprehensive investigation provides valuable insights for establishing criteria to exploit tomato by-products in diverse ways.

Materials and Methods

Chemicals and Reagents

All solvents for extractions and chromatographic analysis, such as ethanol, methanol, deionized water, and acetic acid, were LC-grade and purchased by Sigma-Aldrich. Reagents for antioxidant capacity assays, F–C reagent (F–C, 2 N), sodium carbonate, and DPPH reagent were acquired from Sigma-Aldrich. Individual polyphenol standards of GAL, CAF, CHL, QUE, and KAE were obtained from Sigma-Aldrich. The standard solutions for spectrophotometric methods were prepared using analytical-grade ethanol or methanol.

Apparatus and Instrumentation

An oven (RIOSA) and a commercial dehydrator (Hamilton-Beach) were used to dry the tomato by-product samples. A stirring/heating plate from Barnstead International was used for the SOX and SE methods. Ultrasound equipment with a cylindrical titanium probe (13-mm diameter and 400 W) from Branson was employed for the UAE method, while for MAE, microwave equipment Maxidigest MX 350 was used. A portable UV–Visible spectrophotometer (USB4000) with optical fibers of 200 μm (FIBER 200-UV) and a Multi-LED light source (BluLoop) from OceanOptics were used for the antioxidant capacity assays.

The individual separation and quantification of polyphenols were carried out by liquid chromatography (LC), Agilent 1260 Infinity equipped with a quaternary pump, online degasser, autosampler, and UV detector, and data processing was done by EZChrom from Agilent Technologies. The confirmatory analysis was in an HPLC series 1200 with DAD, ALS Therm, autosampler, binary pump, column compartment, and degasser coupled to a triple quadrupole mass detector (QqQ MS) with electrospray source, and data processing was done in MassHunter software from Agilent Technologies.

Samples and their Pretreatment

Five kilograms of fresh tomato (L. esculentum P. Mill.) were acquired in markets from Mexico City. All samples had to comply with the characteristics of a “Big Tomato” according to Mexican regulation NMX-FF-031–1997-SCFI “Non-industrialized food products for human consumption – Fresh vegetable – Tomato – (Lycopersicun esculentum Mill.) – Specifications” (1997): whole samples with fresh appearance, red color, reasonably well-built, 59–71 mm in diameter, without damage, compact, and firm consistency.

Once in the laboratory, samples were washed using deionized water, dried with absorbent paper, and separated into peel and seeds. Then, peels and seeds were dried in the oven at 45 ± 2 °C for 60 h or by the dehydrator at 45 ± 2 °C for 15 h and were powdered (i.d. = 0.40 mm). Later, a peel/seeds pool was prepared to simulate a matrix where the pulp was squeezed entirely, and only the waste was composed of these by-products like in factory treatment. This pool was used for all experiments, storing it at − 20 °C in darkness until analysis.

Spectrophotometric Analysis to Determine Total Polyphenols and Antioxidant Capacity

The F–C assay was employed to quantify the total concentration of polyphenols in all extracts, as follows: 100-µL adequate diluted aliquot of the extract solution, 100 µL of F–C reagent (1 N), and 800 µL of carbonate buffer solution (pH 10, 500 mmol L−1) were mixed and rested for 15 min measuring the absorbance at 760 nm. The total polyphenol content was expressed as milligrams of gallic acid equivalents per gram of sample (mg GAE g−1).

The antioxidant capacity estimation in extracts was determined by DPPH assay. Briefly, 80 µL of the extract solution was mixed with 800 µL of DPPH reagent (500 mmol L−1 in methanol/water 80:20, v/v) and rested for 10 min, measuring the absorbance at 517 nm. The reduction percentage is calculated according to Eq. (1):

$$\%Reduction=\left(1-\frac{{Abs}_{sample}}{{Abs}_{blank}}\right)\times 100$$
(1)

The %reduction is directly proportional to antioxidant capacity in terms of micromoles of gallic acid equivalents per gram of sample (µmol GAE g−1).

Extraction Methods

The operational conditions for the SOX, SE, UAE, and MAE extraction methods were determined using univariate analysis for SOX and using a combination of Plackett–Burman (P-B) and Box–Behnken (B-B) multivariate designs for SE, UAE, and MAE. Each method was evaluated separately because each one has different parameters to affect the extraction. All experiments were carried out using the pooled samples in triplicate.

Univariate Optimization of the SOX Method

Three previously reported SOX methods for polyphenols extraction from different agri-food matrices were tested, and total polyphenols and antioxidant capacity indices were determined.

(1) The procedure described by Alara and co-workers (2018) (SOX I) was carried out: 10 g of sample was extracted with a 200-mL ethanol/water mixture (60:40, v/v) for 2 h at 70 °C into a Soxhlet extractor thimble, constructed with filter paper, and placed in the extraction apparatus. (2) Also, the operational setting of Karami and co-workers (2015) (SOX II) was done: 25 g of sample was extracted in reflux for 6 h using a 250-mL ethanol/water mixture (80:20, v/v) at 70 °C into Soxhlet extractor. (3) According to Lama-Muñoz and co-workers (2020) (SOX III), 10 g of sample was extracted by reflux over 4 h using a 250-mL ethanol/water (60:40, v/v) conditions.

To develop the proposed Soxhlet method, the effect of extractant composition for mixtures of ethanol/water (v/v) at 50:50, 60:40, 70:30, 80:20, and 90:10, the extractant volume such as 100, 150, and 200 mL, and an extraction time of 6 h were evaluated using the total polyphenols and antioxidant capacity indices.

Experimental Designs for Multivariate Optimization of the Proposed SE, UAE, and MAE Methods

As the number of chemical factors tested in SE, UAE, and MAE protocols was high, multivariate optimization was carried out using a combination of Plackett–Burman (P-B) and Box–Behnken (B-B) designs. The P-B designs are fractionated factorial strategies to identify the chemical factors with significant effects on the extraction process, where the number of experiments is multiple of 4. Each variable had an upper ( +) and a bottom ( −) value to build the design matrix with a total of 12 experiments in duplicate. The variables and their upper and bottom values are shown in Table 1.

Table 1 Bottom and upper values for the Plackett–Burman and Box–Behnken designs to optimize the chemical factors of the proposed extraction methods
Full size table

Total polyphenolic content and antioxidant capacity indices were the signals measured for each experiment and for all extraction methods. Data were analyzed using STATGRAPHICS Centurion XVI (Stat point Technologies, USA) to describe the behavior of each independent factor by a first-order model Eq. (2):

$$Y={\beta }_{0}+\sum {\beta }_{i}{X}_{i}+\varepsilon$$
(2)

where Y is the predicted target response, β0 is the model intercept, βi is the ith regression coefficient, Xi is the independent parameter, and ε is the random error (Box et al., 2005). The Pareto chart was plotted using information from the matrix to confirm the influence of factors.

The B-B design was carried out to determine the optimal operating conditions for the detected significant factors. It is a rotatable or almost rotatable design that stands out because it does not include the vertices of the experimental region as treatments (Box et al., 2005). Each factor was assigned with a bottom (− 1), a central (0), and an upper (+ 1) value with a total of 15 experiments, which were carried out in duplicate, estimating total polyphenolic content and antioxidant capacity indices as a percentage of recovered solid residues on extract, also known as extraction yield, according to Eq. (3):

$$\%Extraction\:yield=\frac{{S}_{ae}}{{S}_{be}}\times 100$$
(3)

where Sbe is the amount of dried sample before extraction and Sae is the amount of dried sample after extraction. This percentage describes the amount of mass transferred from soluble solid grounds to extractant where, if the value increases, it represents an exhaustive process as much as possible. The factors and their upper and bottom values are described in Table 1.

Data was also analyzed, and the results of this design were adjusted to a quadratic model or response surface regression Eq. (4):

$$Y={\beta }_{0}+\sum {\beta }_{i}{X}_{i}+\sum {\beta }_{ii}{X}_{i}^{2}+\sum {\beta }_{ij}{X}_{i}{X}_{j}+\varepsilon$$
(4)

where Y is the predicted response, Xi and Xj represent independent coded variables, β0 is the intercept, βi is the ith linear coefficient, βii is the ith quadratic coefficient, βij is the ijth interaction coefficient, and ε is the random error. A desirability function (Eq. (5)) was used to select the conditions that maximized the total polyphenolic content, antioxidant capacity, and extraction yield:

$$D={\left({d}_{1}\left({Y}_{1}\right){d}_{2}\left({Y}_{2}\right)\dots {d}_{n}\left({Y}_{n}\right)\right)}^{1/n}$$
(5)

For each response Yi(X), a desirability function di(Yi) = 1 represents an entirely desirable or the best value obtained by the surface response model (Vera Candioti et al., 2014).

Precision and Recovery Studies of Proposed Extraction Methods

To evaluate the reliability of data obtained from proposed methods, intra-day and inter-day precision and the recovery, expressed as a percentage, were estimated as follows. Under the suited extraction conditions, samples were extracted by each proposed method in duplicate for 7 days, and the extracts were then analyzed by F–C and DPPH methods. After that, a one-way ANOVA test was carried out to calculate the percentage of relative standard deviation (%RSD) (Miller & Miller, 2000). For recovery, samples were spiked with 5 mg of GA standard under slow stirring over 24 h and then extracted using suited operation conditions in triplicate.

The extracts were analyzed by F–C and DPPH methods, and recoveries were calculated as indicated in Eq. (6):

$$\%Recovery=100\left(\frac{{C}_{1}}{{C}_{0}}\right)$$
(6)

where C1 and C0 are the concentrations after and before the analysis.

According to AOAC guidelines (2019), the acceptance criterion is < 20% of RSD for precision and between 80 and 120% for recovery.

Liquid Chromatography for the Determination of a Polyphenol Panel

In this investigation, five polyphenols were tested representing the main classes of polyphenols detected in these by-products: hydroxybenzoic acids (GAL), hydroxycinnamic acids (CAF and CHL), and flavonoids (QUE and KAE) with the objective of correlating such composition to the antioxidant capacity estimated by spectrophotometric methods.

The quantification of analytes was carried out by LC-UV using a ZORBAX Extend-C18 (4.6 × 150 mm, 5 μm) column (Agilent Technologies). This column increases hydrocarbon chains on the surface of particles, improving the separation of neutral compounds; and, by its way, mobile phase pH (~ 3.5) assures the presence of non-ionized polyphenols, resulting in their good partitioning between stationary and mobile phase.

The LC-UV conditions were 5 μL of injection volume and a flow rate of 0.5 mL min−1, a Mobile Phase A that comprised 0.2% acetic acid aqueous solution, and a Mobile Phase B of methanol. The gradient program was as follows: from 0 to 3 min at 30% B, increase to 50% B to 5 min and held to 6 min, increase to 70% B to 8 min and held to 12 min, decreased to 30% to 14 min, and maintained to 17 min. The analytes were tentatively identified by comparing their retention times and UV spectra with the corresponding standards and quantified by interpolating the calibration curves. For GA, QUE, and KAE, the absorption wavelength was 260 nm, and 324 nm was for CAF and CHL.

Polyphenols’ identities were confirmed by LC-QqQ MS. The ESI source was employed in the negative mode in scan mode between 50 and 1650 m/z range and to set the nebulization pressure to 30 psi. The capillary voltage was set at 5 kV, the cone voltage at 167 V, and the temperature of the additional nitrogen at 350 °C.

Statistical Analysis

Spectrophotometric and chromatographic data with different statistical tools according to specific goals were evaluated. The F-Fisher test was used to determine the homo- or heteroscedasticity of variances and compare the two groups. The t-student test contrasted the media at two tails (95% confidence). A one-way ANOVA test determined significant differences between groups, and the Fisher LSD test determined the group(s) that showed a significant difference.

Also, a data matrix 24 × 7 (groups × variables) differentiated the methods based on the response variables. The groups were SOX, SE, UAE, and MAE, and the response variables were total polyphenol content, antioxidant capacity, and the concentration of the five polyphenols by LC-UV. Data was autoscaled and subjected to principal component analysis (PCA). Microsoft Excel 2019 and STATGRAPHICS Centurion XVI software were used to treat and analyze the data.

Results and Discussion

The present work is divided into two main sections: (i) to determine the influence of drying techniques (oven and dehydrator) and (ii) extraction techniques (SOX, SE, UAE, and MAE) on the extraction process of phenolic compounds from tomato peel and seed. For this purpose, total polyphenols and antioxidant capacity indices were estimated, and LC-UV analysis of the five phenols was done. A deep study of the different effects of extraction conditions in each method is also presented to optimize such processing. Besides, obtained spectrometric and chromatographic data was used (a) to determine matrix phenols/extraction technique relation, (b) to correlate antioxidant indices and polyphenol concentrations, (c) to compare information with literature review, and (d) to propose potential applicability of given methods.

Identification and Quantification of the Selected Phenolic Compounds by LC-UV and LC–MS

The HPLC–UV chromatogram to detect hydroxybenzoic acids and flavonoids is shown in Supplementary Information Fig. S1. The highly polar components are attributed to hydroxybenzoic acids confirmed with the GAL standard analyzed. At the end of the chromatogram, a mixture of compounds was also detected, but with lower intensity than hydroxybenzoic acids, assigned to flavonoids and demonstrated injecting QUE and KAE. On the other hand, for hydroxycinnamic acids, an LC-UV chromatogram was acquired at 324 nm, detecting CAF and CHL (Supplementary Information Fig. S1).

LC-QqQ MS analyses were carried out to corroborate phenols’ identity and avoid the wrong assignation due to isomers’ presence. MS allowed to identify phenolics in extracts, using the characteristic m/z for each phenol, comparing fragmentation patterns to those obtained by standards. Mass fragments corresponding to [M-H]– adduct for the polyphenols were GAL at m/z 169, CAF at m/z 179, CHL at m/z 353, QUE at m/z 301, and KAE at m/z 285. External standard calibration graphs were plotted in the range of 5–50 μg mL−1 in MeOH to quantify the concentration of the polyphenols. Values of regression equations, linearity, LOD and LOQ, and run-to-run repeatability are described in Supporting Information Table S1. For all cases, r2 values were higher than 0.9940. LOD was from 0.97 to 1.94 μg mL−1, and LOQ was from 2.96 to 5.88 μg mL−1. Good run-to-run repeatability (< 20%) was also achieved with RSD in the range of 3.70–7.42% for all standards (AOAC, 2019) at three concentrations.

Estimation of Total Polyphenols and Antioxidant Capacity Indices by UV/Vis Spectrophotometry

Over the years, many experimental conditions have been proposed for estimating total polyphenols and antioxidant capacity indices using F–C and DPPH spectrophotometric assays in fruits and vegetables. Nevertheless, they must be controlled to generate reproducible information (Granados-Guzmán et al., 2014). In this study, experimental conditions for performing F–C and DPPH assays were studied, and calibration parameters were set.

For F–C assay, suited conditions were 1.0 N F–C reagent concentration, carbonated buffer solution (500 mmol L−1, pH 10), and 15 min of reaction time. Data from these experiments are shown in Supplementary Information Fig. S2. To sum up, a direct correlation between F–C reagent concentration and absorbance was found; it was avoided NaOH solutions as alkaline media because of the rapid destruction of blue complex (Chen et al., 2015; Magalhães et al., 2010); and high concentration of salts in buffer enhanced the complex formation (Turkmen et al., 2006). In the case of the DPPH method, experimental conditions were 500 µmol L−1 DPPH reagent and 10-min reaction time (Supplementary Information Fig. S3).

The calibration graphs, using GAL standard solutions in water/ethanol solutions (80:20 v/v), were prepared in working ranges from 20 to 80 mmol L−1 for the F–C method and from 10 to 40 µmol L−1 for antioxidant capacity by DPPH. Good linearity was achieved with r2 of 0.9914 and 0.9937 for F–C and DPPH, respectively. Besides, LOD and LOQ were lower than 2.53 and 7.96 mg GAE mL−1 for the F–C and 1.08 and 3.29 µmol GAE L−1 for the DPPH, respectively. Precision was acceptable with RSD, at three levels, under 11.32% for both assays, in concordance with AOAC guidelines (2019) (Supporting Information Table S2).

Characteristic of the Drying Processes for Tomato By-products

Drying is the first step for the extraction of components from solid matrices. For instance, the convective drying process is used to remove water from foods through heat transfer, where hot air is allowed to pass through the samples in a manner to transfer the heat to them, and moisture is osmotically removed effectively, and these combined osmotic/convective methods have been primarily studied for many fruits and vegetables. In this regard, the oven is one of the methods of choice because it is one the most used method to dry vegetables because it is simple and cheap, whereas a dehydrator is an attractive, easy-to-use, and rapid alternative with high potential to be applied in laboratories, additionally being both low-cost processes.

The peel and seeds tomato pool were prepared and subjected to dryness in the oven and dehydrator up to constant weight. The samples were weighted every 6 h. The found values of percent moistures were 95.86 and 89.45% for the oven and dehydrator processes, respectively, at drying times for the oven and dehydrator of 60 and 15 h, respectively.

From the samples dried by oven and dehydrator for each drying process, pools were prepared. Then, 10 g of each pool was extracted using Soxhlet extraction conditions reported by Alara and co-workers (2018). Afterward, extracts were analyzed using three approaches: (i) plotting calibration graphs for F–C and DPPH methods using GAL standards at described working ranges and spiking such solutions with extracts, (ii) estimating total polyphenols and antioxidant capacity indices, and (iii) determining the concentration of target phenols in extracts.

The first approach was to compare slope as intercept values in searching for possible interferences by matrix effects. A summary of these results is shown in Table 2. Data were then subjected to comparison analysis. Herein, regression parameters of two groups were compared at the time: (i) standard calibration versus spiked calibration with oven-dried extracts, and (ii) standard calibration versus spiked calibration with dehydrator-dried extracts, applying given analysis for F–C and DPPH assays. First, homoscedasticity analysis was carried out, where results indicated the similarity of variances (p-values > 0.05); consequently, a paired data t-student analysis was done at two tails and 95% confidence. The null hypothesis was accepted in slope (tc = 11.61 and t0 = 4.30) and intercept (tc = 50.34 and t0 = 4.30) comparisons for the F–C method, indicating no significant differences among values and discarding possible matrix effects by drying type.

Table 2 Regression equation of standard and spiked calibration graphs for Folin–Ciocalteu (F–C) and DPPH spectrophotometric methods
Full size table

In the case of the DPPH method, the slope did not show significant differences in either oven (tc = 7.75 and t0 = 4.30) or dehydrator (tc = 36.63 and t0 = 4.30); nevertheless, a notorious increase in the intercept was noticed when spiked calibration with oven-dried extract was tested (tc = 0.05 and t0 = 4.30). If intercept increases in a spiked calibration model in comparison to standard calibration one, it is indicative that components apart from analytes (polyphenols) in the matrix have been detected; however, this tendency is noticed along all analyzed concentrations (no changes on the slope) which means that there are some components in extracts with a high capacity to exchange a proton from their internal structure to stabilize DPPH radical. However, they do not represent interferences to this quantification step. In addition, all models were validated with good linearity, showing r2 values higher than 0.991.

After that, total polyphenol and antioxidant capacity indices were estimated in extracts, and data was analyzed by F-Fisher and t-student tests as described before (Table 3). No significant differences were found in polyphenols index (tc = 5.33 and t0 = 4.30), but the antioxidant capacity index of oven-dried extracts was 31.72% greater than dehydrator-dried ones (tc = 0.23 and t0 = 4.30). These results concord with our previous tests, showing that many antioxidant components are extracted when the oven is used for sample drying.

Table 3 Total polyphenol index (expressed as mg of gallic acid equivalent per g of dried sample), antioxidant capacity index (expressed as µmol of gallic acid equivalent per g of dried sample), and concentration of some polyphenols by HPLC-DAD in extracts from tomato by-products obtained by oven and dehydrator processes (n = 6)
Full size table

Determination of Polyphenols in the Extracts Treated by the Two Drying Processes

LC-UV and LC–MS then analyzed extracts to determine the concentration of GAL, CAF, CHL, QUE, and KAE, and Table 3 summarizes the results. It is essential to mention that KAE was not detected in any extract, but it is attributed to the influence of the employed extraction process, as described in the following sections.

Data was analyzed by F-Fisher and t-student tests as previously described. As can be seen, oven-dried extracts significantly increased the concentrations of GAL, CHL, and QUE by 4.16%, 49.05%, and 32.35%, respectively, more than dehydrator-dried extracts (tc = 7.45 and t0 = 4.30). It shows that hydroxybenzoic acids, high molecular mass hydroxycinnamic acids, and flavonoids are suited to be extracted from tomato by-product samples when the oven is used as a pre-processing step at studied conditions, hypothesizing that a low amount of water in the sample promotes the releasing of these components to hydro-alcoholic extractants. By its way, CAF did not show significant concentration changes between both extracts (tc = 1.25 and t0 = 4.30), indicating that water content does not significantly affect extraction of low molecular mass hydroxycinnamic acids.

Principal Component Analysis for the Drying Methods

To find patterns between phenolic compounds, antioxidant indices, and drying process, PCA was used to analyze data from Table 3. The principal component number was set at 3. These components explain 98.95% of cumulative variance. Figure 1 shows the 3D biplot polyphenol composition under the two drying processes. The samples treated with both drying processes separated along component 1. On the left of the plot, marked with a red circle, the dehydrator process is completely separated from the oven process.

Fig. 1
figure 1

Principal component analysis plot for comparison of the effect of drying processes on the determination of polyphenols from tomato by-products

Full size image

The 3D signals indicated for the analytical parameters like total polyphenols, antioxidant capacity, and concentration of individual polyphenols show that the scores of the oven process fall near all analyzed polyphenols but CAF, which is close at the two drying processes because the difference is not significant (see result of CAF in Table 3). These results mean that significant chemical changes occur in the extracts when the oven process is used. The graph also shows the positive effect of the oven on total polyphenol concentration and antioxidant capacity of extracts.

Corresponding to the direction of loadings, the CAF is opposite to total polyphenol concentration and antioxidant capacity by the DPPH method, indicating that the oven promotes transformations between CAF and spectrophotometric data: an increase of total polyphenols and antioxidant capacity and a decrease of CAF was observed. It also suggests a lack of correlation between these parameters. Therefore, the oven was selected as a sample drying pretreatment method for further experiments.

Univariate Optimization of SOX Extraction

SOX is the method of choice for extracting organic compounds from solid matrices due to its demonstrated exhaustiveness. So, a SOX method to extract polyphenols from tomato by-products was developed. For this purpose, three previously proposed SOX methods to recover polyphenols from vegetable matrices were first assessed. Data was subjected to a one-way ANOVA test to compare the variances of results and to find statistical differences (Fig. S5 in Supplementary Information).

The total polyphenol index increased when the methods proposed by Alara et al. (2018) (SOX I) and Lama-Muñoz (2020) (SOX III) were applied. The main differences in chemical factors of both protocols tested are extractant composition, sample-extractant ratio, and extraction time, which indicate their importance in the extraction phenomena. Thus, further experiments to evaluate the effect of such factors were conducted to develop the new SOX protocol.

Extracts were obtained using a 5-g oven-dried pooled sample at 70 °C heating extractant in reflux. The extractant was composed of hydro-alcoholic mixtures using ethanol as an organic solvent. Five EtOH/water mixtures with different proportions from 50:50 to 90:10 (v/v) were tested as extractants. It was observed that an increase in the content of organic solvent in the extractant enhanced around 30% of the total polyphenol index; however, no significant differences were observed above 70% EtOH. In this case, for the extraction process to be done, the extractant needed to migrate from the still pot to the thimble and reach the sample. At 70 °C (condition used), the EtOH boiling point (78 °C) is almost reached, so an efficient migration of extractant could be assumed. Nevertheless, although EtOH% in the extractant increased, the solvent semi-polar nature did not allow extraction of polar compounds (main components in tomato), causing extractant saturation by those semi-polar components.

Furthermore, the effect of the sample/extractant ratio was tested using 1:20, 1:30, and 1:40 ratios. Polyphenol extraction was enhanced by 17% when a 1:40 ratio was used, being in concordance with Luque de Castro and Priego-Capote (2010), who indicate that a high surface area of contact of the sample with extractant enhances the permeability of the extractant into cells to release polyphenols. On the other hand, the extraction time was tested over 6 h; phenolics increased by 80% content in extracts from the beginning up to 4 h, but a 12% decrease was shown after that, principally attributed to oxidation of some polyphenols at high temperatures along time (Antony & Farid, 2022). These conditions were used for the SOX proposed method, as indicated in Table 4.

Table 4 Optimal conditions obtained by experimental designs of the evaluated factors for each extraction method
Full size table

Multivariable Optimization of the Polyphenol Extraction by SE, UAE, and MAE

Alternative techniques to SOX are stirring extraction and those green approaches assisted by auxiliary energies. This study optimized three alternative extraction methods: SE, UAE, and MAE, the last two in focused mode. The suited extraction conditions for each process were determined by first studying the effect of some chemical factors on total polyphenols and antioxidant capacity indices of the extracts, finding those with significant impact on the extraction process, and subsequently using such information to precisely establish optimal values for given factors.

Extractant composition, sample/extractant ratio, and extraction time were optimized for the three methods: heating temperature and stirring speed were for SE, sonication and relaxation periods were for UAE, and microwave exposure time and power were for MAE. The experimental values for such factors and the results of matrices are described in Supporting Information Tables S3–S9. In addition, Pareto charts and surface response graphs from this multivariate study are shown in Supplementary Information Fig. S6. Oven-dried pooled samples were employed as a model for these experiments, and results are described in the following sub-sections.

Effect of Extractant Composition

The extractant was evaluated according to the water percentage used in hydro-alcoholic mixtures and the type of organic solvent used. MeOH and EtOH were tested because both have demonstrated a high capacity for extracting polyphenols in different food matrices. In all experiments, Pareto charts did not show significant differences between both solvents; however, an enhancement of up to 15.62% in total polyphenols and antioxidant capacity indices was observed when EtOH was used for UAE and MAE. Moreover, it is proven that EtOH is more compatible with biological systems than MeOH (Azabou et al., 2018). Thus, EtOH was used for subsequent experiments.

As follows, ethanol/water mixtures were studied in the range from 50:50 to 90:10. During P-B tests, the percentage of water in extractant preparation was significant, but with different effects according to each extraction process. Polyphenols and antioxidant capacity indices enhanced around 5% for SE when low water percentage was used, but they showed a 12% increase with high %water for UAE and MAE. This is attributed to the interaction of this variable with others in each extraction system. For instance, in SE, the extractant was influenced by heating temperature and sample/extractant ratio, where a slow extraction process is carried out; hence, long periods of contact of the sample with the extractant contributed to the leaching of all compounds from polar to semi-polar.

In contrast, rapid kinetics in UAE and MAE processes make necessary high %water to extract as many polar compounds (major in tomato samples) as possible in a short time. The extraction of polyphenols is favored by using a water content in the extractant because of the enhancement of mass transfer of high polar compounds, such as phenolic acids, from the cellular matrix to the extractant (Ross et al., 2009). So, %water was optimized in the three techniques.

Effect of Sample/Extractant Ratio

Ratios from 1:10 to 1:50 of sample/extractant were tested by modifying the extractant volumes. For UAE and MAE, Pareto charts did not show a significant influence of this factor, neither total polyphenols nor antioxidant capacity, so, in both cases, a 1:40 sample/extractant ratio, 0.5 g in 20 mL of extractant, was chosen. However, on the SE method, total polyphenol and antioxidant capacity indices from the by-product extracts increased by 9.5% using a 1:50 sample/extractant ratio. The SE extraction process needs a long period of sample/extractant contact; this period is exceptionally beneficial whether the surface area of such sample is exposed to high amounts of extractant transporting a higher number of compounds from matrix to extractant.

Effect of Extraction Time and Assistance of Physical Energies

Extraction time is a parameter related to extraction kinetics, indicating the amount of compounds released from matrices in a specific time.

Stirring Extraction

SE kinetics is slow, where stirring and heating only promote temperature elevation and contact of extractant with the sample, requiring an extended period. In this work, extraction time was evaluated from 1 to 6 h, without significant differences, selecting 1 h for short processing. Stirring also did not show a significant effect, setting minimum stirring speed. Nevertheless, heating was statistically influential in the extraction system, with a 7% increase in total phenol content. Besides, it was influenced by extractant composition and sample/extractant ratio. Data is attributed to high heating and degradation of high molecular mass polyphenols, such as lignans and tannins, producing many monomers of these compounds, so a high content of ethanol in extractant is required to recover such compounds (Ding et al., 2020).

Ultrasound-Assisted Extraction and Microwave-Assisted Extraction

UAE and MAE have short kinetics due to ultrasounds, and microwaves promote some physical phenomena such as cavitation, membrane disruption, and elevated temperature, among others, which rapidly enhance extraction. Nevertheless, due to degradation processes, special care must be taken to avoid losing polyphenols.

In the case of the UAE method, low extraction time enhanced around 50% of both antioxidant indices in obtained extracts, above all, when it interacted with down sonication time. Even though a long time of exposure of the extraction system to ultrasounds improves presence of cavitation phenomena, causing the implosion of bubbles, disrupting membrane cells, and increasing extractant permeability, this phenomenon elevates temperature, causing polyphenol degradation, and, additionally, the high number of bubbles in system makes physically impossible extractant permeability, thus, relaxation (non-irradiation) and low extraction periods are good features for polyphenol leaching (Chemat et al., 2017). Therefore, extraction time and sonication period were optimized. The relaxation period was not statistically significant, selecting the lowest time.

Concerning the MAE method, it had similar antioxidant indices enhanced at low extraction time and power. In this case, microwave power is related to wave amplitude, where its increase allows an easy cell disruption, but, at the same time, a drastic temperature elevation, which promotes the formation of radicals in extractant and, subsequently, oxidation of polyphenols (Hu et al., 2021). So, low extraction time at low power is preferred.

Determination of Optimal Extraction Conditions for the Proposed Methods

With the results obtained and using a desirability function integrated total polyphenol index, antioxidant capacity index, and extraction yield, a surface response methodology was used to determine optimal conditions for detected significant factors through Box–Behnken design. Contour graphs to see interactions in the system and to understand the behavior of variables on the extraction of polyphenols were plotted, and the obtained information was like that obtained during exploratory analysis by Plackett–Burman designs. So, experimental conditions were selected according to the experiment that got the matrix’s highest desirability. This analysis showed good performance, achieving desirability values around 0.98.

Optimal conditions for each method are summarized in Table 4.

Precision of the Proposed Extraction Methods

The intra-day and inter-day precision were estimated for the proposed methods as described in “Precision and Recovery Studies of Proposed Extraction Methods” section in Materials and Methods. The RSD values were from 10.81 to 16.40% for SOX, from 8.12 to 10.79% for SE, 7.22 to 10.19% for UAE, and 9.12 to 15.68% for MAE. In the case of recoveries, the percentages were in the range of 115.55–117.10% for SOX, 96.29–107.58% for SE, 86.41–119.85% for UAE, and 80.48–91.71% for MAE, UAE being the method with the best intra-day (6.79%) and inter-day (7.22%) precision for FC index, and the SE was the best for antioxidant capacity index with 8.12% and 8.22% for intra-day and inter-day repeatability. All results are shown in Table S10 of Supplementary information.

All results concord with the AOAC criterion (2019), expecting values under 20% for precision and 80–120% for recovery.

Influence of the Extraction Techniques

Table 5 shows the results of the indexes of data. The significance criteria were by a one-way ANOVA test (95% confidence at two tails) followed by the Fisher LSD test of data.

Table 5 Estimation of total polyphenol index, antioxidant capacity index, and extraction yield of extracts obtained from tomato by-products using optimal extraction conditions for each proposed method (n = 6)
Full size table

The ascending order for the total polyphenols index of extracts was SOX < MAE < UAE < SE (p-value: 0.002), and for antioxidant capacity index was MAE < SOX = UAE < SE (p-value: 0.0004). While SE extracts showed the highest values for both indices, SOX obtained the lowest for the total polyphenol index but an excellent antioxidant capacity index. There is a difference close to 6% in the antioxidant capacity indices between SOX and SE extracts, but it is not attributed to polyphenols because polyphenol indices showed a difference of approximately 70%. Zalewska et al. (2022) demonstrated that the extreme heating conditions in the SOX process could promote melanoidin formation from Maillard reactions, and those compounds cause positive interferences in the DPPH assay, obtaining a higher value than the authentic value.

Concentration of Polyphenols by LC-UV and LC–MS in the Extracts

Supplementary Information Fig. S1 shows an example of an LC-UV chromatogram at 260 nm for GA, QUE, and KAE and 324 nm for CAFA and CHLA, corresponding to an extract obtained by the SE method, and Table 6 lists the concentration values of the compounds for all extraction techniques.

Table 6 Polyphenol concentration by LC-UV in extracts from tomato by-products obtained by the proposed solid–liquid extraction methods (n = 6)
Full size table

ANOVA test (95% confidence at two tails) followed by the Fisher LSD test was performed to analyze data.

The ascending order for the concentration of polyphenols using the extraction methods was SOX < MAE < UAE = SE (p-value = 0.0031), while the increase of the individual concentration of polyphenols was KAE < QUE < CHL = CAF < GAL (p-value = 0.0014).

The UAE method obtained the same concentration of individual polyphenols as the SE method but 92% faster (5 min vs 60 min) due to the enhancement of analytes released by cavitation phenomena. On the other hand, MAE was 66% faster than SE and four times slower than UAE, with concentrations 30% lower than both methods. It could be attributed to the oxidation of the polyphenols due to the high power of microwaves and high rates of temperatures that can cause polyphenol degradation (Hu et al., 2021).

Principal Component Analysis from the Results by LC-UV, F-C, Antioxidant Capacity Indexes, and Extraction Methods

The extraction method is fundamental to obtaining good results in the analysis of compounds from solid matrixes, so it is necessary to know the quality and the difference or similitude among them to select the best method depending on the objective to follow. The effect of solid–liquid extraction methods on polyphenols of tomato by-product samples was done using a PCA from the results in Table 5 and 6, except for extraction yield. The protocols were plotted in Fig. 2, finding discriminations among methods at 95.80% of the total variability with three components in the 3D biplot.

Fig. 2
figure 2

Principal component analysis score for comparison of the effect of solid–liquid extraction methods on the determination of polyphenols from tomato by-products

Full size image

The SOX and MAE extracts were separated along component 1, whereas the extracts from SE and UAE exhibited complete overlapping. The SE and UAE methods’ scores fall near antioxidant capacity indices and all analyzed polyphenols, meaning that significant changes occur when those methods are used.

Regarding the loadings, the total polyphenol index was the same way for all analyzed polyphenols, along with component 1, indicating a good correlation between this index and the concentration of the polyphenols in the four extraction methods. Also, it is possible to conclude that SE and UAE methods promote the enhancement of antioxidant capacity indices as polyphenol concentration. Nevertheless, the DPPH antioxidant capacity index is distributed along component 2, is opposite to the majority of the analyzed polyphenols’ concentration, and shows a possible lack of correlation among data.

Coefficient of Pearson to Determine the Relationship of the Concentration from each Polyphenol with the Spectrophotometric Assays

The individual concentrations of polyphenols determined by LC-UV and the indices obtained by the spectrophotometric FC and DPPH assays in the extracts obtained by the four extraction methods from tomato by-products by oven dehydration were correlated using the coefficient of Pearson. It was also used to determine the significant correlation between antioxidant capacity indices and the concentration of polyphenols, expecting an r value over 70% (Sánchez de Medina et al., 2015).

Regarding the analysis of SOX extracts, the concentration of GAL, CHL, and QUE well fitted with the total polyphenol index (> 70.81%), whereas for CAF and KAE, the correlation was lower than 62.32%; meanwhile, all polyphenols’ concentrations and the antioxidant capacity index showed a lack of correlation (< 39.65%). This can be attributed to an increase in the antioxidant value due to the formation of products from the Maillard reaction by high heating temperatures in the SOX method, causing false positives.

For extracts obtained by SE, all polyphenol concentrations correlated well with the total polyphenol index above 78.21%, and CAF showed the highest value (98.99%).

For the extracts obtained using UAE, the correlation of the concentration of all compounds and total polyphenol index well fitted with values between 77.12 and 87.55%. Nevertheless, for both SE and UAE, the CAF, CHL, QUE, and KAE concentrations and the antioxidant capacity index were not fitted, less than 38.46%, and only GAL showed values higher than 77%.

The polyphenol concentrations in MAE fitted with the total polyphenol index (> 86.37%), and only the antioxidant capacity index was adjusted with GAL (88%), being higher than that of the correlation for SE and UAE. It notes that the individual polyphenol concentrations and the index values are much lower, 30%, using MAE than SE and UAE. Therefore, these inconsistencies indicate the presence of other phenolic components, non-polyphenolic antioxidants, and compound reductants in the extracts that interfere with the assays.

For instance, Moco and co-workers (2006) made a database for the chemical composition of whole tomatoes where some amino acids, sugars, organic acids, carotenoids, and other polyphenolic compounds could be related among them and their antioxidant capacity. All these compounds have a demonstrated antioxidant potential, whereas sugars and organic acids are identified as interferents for DPPH assay (Huang et al., 2005).

Concentration of Polyphenols in Tomato Peels and Tomato Seeds

Although the tomato by-products include seeds and peels, the amount of polyphenols for each is different. Thus, peel and seed extracts were analyzed separately using the SE method, which had the highest values for almost all parameters, and calculating the concentration of the polyphenols studied by LC-UV.

For peels, the content of GAL, CAF, CHL, QUE, and KAE were 3.37 ± 0.01 mg g−1, 0.57 ± 0.009 mg g−1, 0.56 ± 0.001 mg g−1, 0.54 ± 0.12 mg g−1, and 0.25 ± 0.03 mg g−1, respectively, while for seeds, the concentrations were 2.54 ± 0.004 mg g−1, 0.42 ± 0.001 mg g−1, 0.55 ± 0.001 mg g−1, 0.52 ± 0.01 mg g−1, and 0.25 ± 0.007 mg g−1, respectively. In this case, it was noticed that peel extracts showed a higher concentration of GAL and CAF than seed extracts (tc = 0.89 and t0 = 4.30) but similar concentrations for CHL, QUE, and KAE.

Characteristic and Concentration of Polyphenols from Methods Reported Previously

A bibliographic review of methods previously reported in the literature to compare their results with the obtained data from the proposed methods in this work is very relevant to establishing suitably the criteria for selecting the solid–liquid extraction method in food matrices.

Table 7 shows the results of the polyphenol extraction from tomato by-products previously reported. For extracts obtained by SE, total polyphenols and the antioxidant capacity indices obtained by the proposed method were a bit equal to those obtained by Azabou et al. (2020), but the proposed showed short extraction time (see Table 4). The main difference between both methods was the extractant; while Azabou used 100% ethanol, an ethanol/water mixture was employed for the proposed method, enhancing the extraction of more polar polyphenols in a shorter time. Also, the other SE methods showed lower values for both indices because low sample–extractant ratios were used, which decreased surface area and avoided the permeability of the extractant (Castro-López et al., 2017).

Table 7 Total polyphenol index (mg gallic acid/g sample), antioxidant capacity index (µmol gallic acid/g sample), and polyphenol concentration by LC in extracts from tomato by-products reported by different authors
Full size table

On the other hand, di Donato and co-workers (2018) showed high values for QUE and KAE, but the used model sample was composed of not only peels and seeds but also whole unripe and rotten tomatoes, providing a high number of endomembrane, tomato organ where flavonoids biosynthesized, which increased extraction of these compounds (Falcone Ferreyra et al., 2012).

Regarding the effect of ultrasounds, the proposed method showed the highest values for total and individual concentration of polyphenols compared with the reported ones. The ultrasounds promote the external organ destruction—like membranes—by cavitation phenomena, which enable the extractant permeation into cells, and in addition, the internal organs suffer deformations which allow the releasing of polyphenols from solid matrices (Chemat et al., 2017).

However, the UAE method, in focused mode, of Kalogeropoulos and co-workers (2012) obtained a threefold value for the antioxidant capacity index in comparison with the proposed UAE, attributing it to two-sequential extraction processes which could release several mid-polar and some non-polar antioxidants, such as lycopene and β-carotene. DPPH assay is based on hydrogen atoms’ transference from antioxidants to radicals, so carotenoids transfer one hydrogen for radicals’ stabilization, increasing such index value (Huang et al., 2005). For Szabo and co-workers’ UAE method in conventional mode (2019), there are two differences: the volume of extractant was not indicated, hypothesizing low sample–extraction ratios decreased extraction efficiency, and the use of the methanolic aqueous solution as extractant that seems to reduce the extraction of semi-polar components such as flavonoids. Fernández and co-workers (2018) proposed a UAE in conventional mode using a bath and a natural deep eutectic solvent (lactic acid and glucose, 5:1 (w/w)), obtaining the poorest values, especially for CAF and QUE.

By its way, the proposed MAE method obtained higher values for total polyphenols and CHL and KAE concentration than Tranfić Bakić and co-workers’ (2019) method, which uses high-pressure MAE in conventional mode. Nevertheless, several different chemical factors were observed. The technique used by Tranfić Bakić was carried out using 1 g of peels extracted with 50 mL of methanol/water mixture as an extractant for 10 min, reducing about two times the total polyphenol content and CHL and KAE concentrations. The focused energies show phenomena like powerful shear, high-frequency vibration, high-velocity impaction, and cavitation, which elevate the temperature to burst the cell wall and release the polyphenols into extractant (Castro-López et al., 2017).

During this review process, there needed to be more optimization studies in some publications, which could help to profoundly understand the effect of the chemical factors on the extraction efficiency and to determine, as better as possible, the suited experimental conditions. In addition, for a better discussion of the results, it is highly recommended to carry out at least precision and recovery studies.

Importance of the Proposed Extraction Methods on the Obtaining of High-Value Extracts

All analytical methods have advantages and disadvantages, and it is always necessary to evaluate which procedure is the more suitable depending on the purpose.The extraction methods are no exception, so data of the proposed methods is discussed to take it into account in making decisions for the extraction of polyphenols, in this case, from tomato by-products.

SE showed the highest values for total polyphenols and antioxidant capacity indices, two essential parameters for food, pharmaceutical, and cosmetical applications. Nevertheless, HPLC analysis showed that SE as UAE extracts obtained significantly similar concentrations for the evaluated polyphenols, making UAE an alternative 92% faster than SE. Another relevant parameter is the correlation between the estimated antioxidant capacity and individual polyphenol concentration; this is crucial to describe the potential of an extract for the prevention and treatment of human diseases. In both cases, the concentration of each polyphenol was well correlated with F–C assay data, indicating that these methods are suitable for high-value polyphenols, where CAF principally contributed to the total polyphenol index and GAL does for antioxidant capacity.

MAE also was a method faster than the SE method, but the polyphenol concentration and antioxidant indices were around 30% lower; however, with this method, GAL had the highest antioxidant capacity/concentration ratio, making it a suitable approach to obtain extracts in rapid extraction time with GAL showing good antioxidant potential.

For SOX, even though it is mentioned as a gold standard for exhaustive extraction process when solid matrices are analyzed, total and individual concentrations of polyphenols were much lower under the experimental conditions used, hypothesizing that high temperatures degrade such compounds. However, SOX extract showed a good antioxidant capacity but no relationship with polyphenols, ponting out that a further deep study on its chemical composition to suitably correlate composition/activity.

Conclusions

The results of this investigation demonstrated that the different extraction methods give variated data on the composition of extracts from tomato by-products. For extraction methods using oven-dried samples, SE showed the highest values for total polyphenol content and antioxidant capacity, making it suitable for obtaining extracts with elevated antioxidant potential using minimum laboratory material.

Nevertheless, PCA showed that the individual concentration of polyphenols by SE was like those by UAE, being 90% faster. All detected polyphenols in the obtained extracts by these two methods also showed good antioxidant potential given by the F–C assay, especially CAF, and with the DPPH assay, GAL showed a good relation.

On the other hand, MAE reduced 66% the extraction time compared to SE, but it was not firmly considerable because the concentration values for all polyphenols and antioxidant indices decreased by 30%; however, the ratio GAL/antioxidant capacity index is higher than that for SE. Microwaves reduce the extraction time and decrease some polyphenols because this type of energy promotes the oxidation of polyphenols, forming radicals that do not show antioxidant capacity.

In the case of extracts obtained by SOX, it was not suitable for polyphenol extraction, but it was noticed that antioxidant capacity was increased. Herein, nor other non-polyphenol antioxidants or interference compounds, the extract solution was color intense, which could be causing false positives in this spectrophotometric determination.

Considering all the studied methods, it can be said that, under tested conditions, SE, UAE, and MAE could be used for the obtaining of extracts enriched with antioxidant polyphenols as exploitation strategies of tomato peels and seeds, but the chosen method depends on the goal of each investigation.