Influence of Delipidification on Protein Concentrate Extracted from Fish By-Products: Technical Process, Nutritional Value, and Amino Acid Bioaccessibility

ABSTRACT Fish by-products were used to manufacture delipided fish protein concentrate (FPC), using Soxhlet method or single stage immersion, with n-hexane or anhydrous ethanol solvents. n-Hexane produced the best results. Soxhlet led to the highest protein content (86%), lipid extraction efficiency (92%), and protein digestibility (75%). The single stage immersion method resulted in higher amino acid bioaccessibility values (90%). Soxhlet associated with n-hexane led to an acceptance index above 70% for sensory attributes. FPC is a source of protein of high biological value and can be used by the food industry as an alternative for the preparation and/or fortification of foods.


Introduction
The fish processing industry results in intense generation of residues with nutritional quality similar to in natura, which could be used for co-products for human consumption.According to the literature, about 70% of processed fish becomes waste (FAO 2020), because production processes lack optimization, and edible parts are not fully utilized.However, these materials are a source of high-quality proteins, bioactive peptides, essential amino acids, vitamins, and minerals (FAO 2016;Godoy et al. 2010;Soares and Gonçalves 2012;Tacon and Metian 2013;Tilami and Sampels 2017) and can be used for product development, such as fish protein concentrate (FPC).FPC, also known as fish powder (Ahmad et al. 2021;Tanna et al. 2020) and fish flour (Senapati et al. 2016;Widodo and Sirajuddin 2018) can be manufactured from spines (Petenuci et al. 2010), filet carcasses, fins, and trimmings, minced fish (Rebouças et al. 2012) and head, bones, skin, and viscera (Bastos et al. 2014).
potassium (Abdel-Moemin 2015; Goes et al. 2016).Some products can be produced with FPC such as cookies (Abul-Fadel et al. 2018) and breads (Bastos et al. 2014), that contain increases of up to 35% and 41% of protein in the respective formulations.
The physical-chemical and sensorial characteristics of the FPC depend on its elaboration process, which consists of moisture reduction to concentrate macronutrients, especially proteins.The lipid content of the raw material should be considered, since it can have a direct influence on the sensory and rheological characteristics of FPC and food products formulated with it (Rebouças et al. 2012;Vidal et al. 2011).There are no Brazilian laws that regulate the minimum quality parameters for the production of FPC.However, there are references in the literature (Archer et al. 2001;Oetterer et al. 2006) that suggest FPC classification depending on lipid content and, consequently, nutritional and sensory characteristics.According to these authors, FPC type A has a maximum of 3% of lipids, which is considered of higher sensory quality.
Therefore, it is interesting to include the delipidification process in the FPC production flowchart, depending on the lipid content of the raw material.In addition, depending on its use, the delipidification process does not need to be considered.
Thus, several methods could be used for lipid extraction from fish and/or FPC, such as supercritical fluid (Rubio-Rodríguez et al. 2012), cooking processes (Campos et al. 2020;El-Rahman et al. 2018;Ween et al. 2017), washing cycles (Rebouças et al. 2012), and the use of solvents (Song et al. 2018).The delipidification step may also provide fish oil with biological value for the food industry due to the content of polyunsaturated fatty acids and fat-soluble vitamins (Jamshidi et al. 2020;Song et al. 2018).
With all delipidification methods, solvents are mainly used in food manufacturing processes as they have simple methodologies, can determine the amount of total lipids in food, do not require pretreatment before the extraction process, and do not leave prejudicial residues on the final product that may affect human health (Adeoti and Hawboldt 2014).n-Hexane (C 6 H 14 ) and ethanol (C 2 H 5 OH) are the most common solvents used for lipid extraction from soybean (Potrich et al. 2020;Toda et al. 2016), sunflower seed (Iglesias et al. 2012), corn germ (Espinosa-Pardo et al. 2020) green argan seeds (González-Fernández et al. 2020), chestnuts (Sampaio Neto et al. 2018), Silybum marianum seeds (Li et al. 2012), and even fish (Abdullah et al. 2020).Although solvents are mainly used for this process, they can affect the quality of the product, as high temperatures degrade heat-sensitive (as protein denaturation) and labile natural compounds (Ivanovs and Blumberga 2017).
Heating contributes to deformation of the natural features of protein molecules and could result in protein digestibility reduction caused by complex chemical (cross-linking) reactions (Abraha, Admassu, et al. 2018).In this way, bioaccessibility analysis determines the portion of nutrients that can be released from a food matrix and become potentially available for absorption in the gastrointestinal tract (Rinaldi et al. 2015).This analysis can infer the quality of products by determining protein digestibility and its components bioaccessibility (Mercadante and Mariutti 2018).The effect of the delipidification process on FPC protein bioaccessibility needs to be studied.Accordingly, the aim of this study was to evaluate the impact of the delipidification process on FPC protein recovery and evaluate its quality and amino acid bioaccessibility to improve the technical process and functional uses of FPC.
By-products of tambatinga industrial processes (Colossoma macropomum × Piaractus brachypomus) were purchased from an industry located in Várzea Grande, Mato Grosso State, Brazil.

Samples collecting
For the elaboration of the FPCs, by-products from the industrial tambatinga filleting process were used: fillet shavings with a high concentration of Y-bones and carcass after filleting (muscular portion with bones).The Y-bones, also known as intermuscular bones, are attached to the vertebrae and have y-shaped configuration.The by-products were transported from the Federal University of Mato Grosso to the Meat, Fish, and Derivatives Technology Laboratory of Nutrition College, in isothermal bags (5°C for 30 min) and kept frozen (−25°C) until analysis.

Protein recovery from fish by-products
Fish by-products were ground and homogenized for 5 min in an industrial meat grinder (Poli, Pcp22L, Brusque, Brazil) in the following proportions: 40% of carcass and 60% of Y-bones.The ground samples were placed in trays and coated with aluminum for drying in a forced air circulation oven (Solab Científica,SL 102,Piracicaba,Brazil) for 15 h at 65°C.Thereafter, the dry material was ground again for 1 min at full power (Arno, Optimix plus LN27, São Paulo, Brazil) and designated as whole fish protein (WFP) (Figure 1).

Soxhlet delipidification
Two solvents were used to remove lipids from the dry material: ethanol (SE) and n-hexane (SH).Briefly, 14 g of dry material (n = 7) was weighed in cartridges of qualitative filter paper (Unifil, 12.5 cm), placed in the Soxhlet extractor (Marconi,Piracicaba,Brazil) and washed with solvents for 4 h (3 h of percolation followed by 1 h of dipping) in a ratio of 10:1 (v/w; solvent/ sample) at 70°C.The cartridges were removed, and the sample was dried in an oven at 65°C until a constant weight was attained for solvent residue volatilization.

Single stage immersion delipidification
Single stage immersion delipidification was performed according to the protocol described by Sampaio Neto et al. (2018), with modifications.The same solvents used in the Soxhlet method were employed: ethanol (EI) and n-hexane (HI).Briefly, 20 g of WFP (n = 10) was weighed in Erlenmeyer flasks (125 mL) and placed in a Dubnoff metabolic thermostatic bath (Biovera, NT 232, Piracicaba, Brazil), with orbital rotation (225 rpm), at a ratio of 3:1 (w/w; solvent/sample) for 12 h at 70°C for ethanol solvent and 12 h at 60°C for n-hexane solvent.After extraction, the solvent was separated using a fine mesh stainless steel sieve.The solid material was dried in an oven at 65°C until a constant weight was attained for solvent residue volatilization.

Lipid extraction efficiency
The lipid extraction efficiency was determined from the weight differences and calculated as follows: lipid extraction efficiency (%) = 100 − (total lipid extracted from delipided FPC (g)/total lipid extracted from WFP (g) × 100).

Water activity (Aw)
Water activity was measured with a specific analyzer (AquaLab, Series 3 TE, Washington, USA), with three repetitions per sample, at 25°C.Results are expressed as 0-1 point.

Color assessment
Color was measured using a chromameter (Konica Minolta, CR-400/410, Osaka, Japan).The CIELab color system was used with an observation of 10º, illuminant D65, and colorimeter calibration with a white standard plate.Results are expressed as Hunter values of L* (lightness), a* (redness and greenness), and b* (yellowness and blueness).Three measurements were performed with three replicates per treatment.Delta color (ΔE) was used to compare the total color change delipided FPC with the control and calculated as follows:

Proximal composition
Proximal composition was determined according to the methodology recommended by Brazilian Standard methods (Brasil 2011), which were prepared in accordance with the Official Methods of Analysis of AOAC International (AOAC 2010c), with modifications.Moisture was measured after overnight drying to constant weight in an oven at 105°C; crude protein was measured using the Kjeldahl method with 6.25 as the conversion factor, total lipid was measured by Soxhlet extraction with n-hexane, and ash was measured by total incineration in a muffle oven at 550°C.Calories were calculated using 4 kcal•g −1 of protein and 9 kcal•g −1 of lipids (Brasil 2003).The analysis was carried out with seven repetitions/treatment, and the results are expressed based on the original moisture.

Mineral content
The microwave-assisted acid extraction procedure was used for the pre-digestion of the samples (AOAC 2010a).An inductively coupled plasma optical emission spectrometer (ICP-OES; Optima 2100 DV, Perkin Elmer, USA) was used (AOAC 2010b).For the calibration curves, certified reference materials specific for analysis by ICP-OES were used, which were produced by companies accredited in ISO 17,034 or A2LA.

Simulated digestion in vitro
In vitro digestion was performed using the harmonized protocol described by Minekus et al. (2014).The samples were digested in four digestive juices formulated in the laboratory: salivary, gastric (pepsin), duodenal (pancreatin), and biliary (bile).Falcon tubes were weighed, horny and 1.5 g of dry or freeze-dried sample was added.Digestion was performed in triplicate with the use of a control tube (white) containing only digestive juices.The first step was digestion by salivary juice, where 4.0 mL of juice +25 μL of CaCl 2 0.3 M were added, with subsequent vortex agitation for 30 seconds and incubation at 37°C for 5 minutes in a water bath with agitation.After digestion in the mouth, 8.5 mL of gastric juice +5 μL of CaCl 2 0.3 M were added to the tubes, agitated in vortex for 30 seconds, and the pH adjusted at 2.0 + 0.2 with HCl 1 M solution.A new incubation was performed at 37°C for 2 hours with constant agitation, and after 30 minutes, the temperature of the tubes was verified.After stomach digestion, 8.5 mL of duodenal juice + bile juice +40 μL of CaCl 2 0.3 M were added to each tube.The samples were stirred in vortex for 30 seconds, and the pH adjusted to 7.0 + 0.2 with NaOH 1 M solution.After each digestive step, the tubes were cooled to ice to cease enzymatic activity.After intestinal digestion, the tubes were weighed and centrifuged at 2750 × g for 10 min at 4°C.The bioaccessible or liquid portion was transferred to Falcon tubes and frozen for analysis (microelements, amino acids, and protein).To analyze the digestion percentage of the samples, the nitrogen content of the original sample and bioaccessible portion was analyzed.The protein percentage of the bioaccessible portion must be above 70% for a satisfactory digestion process (Fogaca et al. 2018).The in vitro digestion protocol used as well as each digestion fluid component is schematized in Appendix A (Supplementary Tables S1 and S2).

Separation and quantification of amino acids by acid hydrolysis
For the separation and quantification of the main amino acids contained in FPC, the method described by Pacheco (2014) was used based on the determination of amino acids by acid hydrolysis and subsequent derivatization.Thereafter, the samples were subjected to chromatographic analysis using a Waters® Alliance model 2690/5 high-performance liquid chromatographic system with a Waters® fluorescence detector.Quantification was performed by standardizing an analytical curve made with a standard solution, which later underwent the derivatization step with 6-aminoquinolyl-succinimidyl-carbamate.The bioaccessible samples were filtered through a syringe filter (0.45 µm), and a 20 µL aliquot was transferred to a vial and vortexed for 15 s.The remaining steps were performed as described above.Bioaccessible amino acids were calculated as follows: % bioaccessible amino acids = (amino acid content in bioaccessible portion/amino acid content in sample) × 100.

Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) for protein degradation
Dried samples (FPCs) were blended (1 g/30 mL) for 2 min in 0.02 M phosphate buffer at pH 7.5 and filtered through a cloth for protein extraction.The eluted fractions were used for the analysis.Digested samples (bioaccessible fractions) did not require any protein extraction steps.All samples (100 µL) were added to the sample buffer (200 µL), followed by pH adjustment to approximately 7.0 with 1 M NaOH.Proteins were analyzed by SDS-PAGE in a Bio-Rad vertical PROTEAN II xi cell (Hercules, CA, USA).Stacking and running gels were prepared using 4% and 8-12% (w/v) acrylamide solutions, respectively.A constant voltage of 100 V was applied throughout the running time (8 h).Gels were stained with Coomassie brilliant blue.High (48-204 kDa) and low (20-103 kDa) molecular weight standards from Bio-Rad were used to calculate the protein molecular masses (Laemmli 1970).

Sensory analysis
Samples with better proximal composition (in terms of protein) and higher protein digestibility were subjected to sensory analysis.The affective acceptance test (Lawless and Heymann 2010) comprises 108 untrained panelists.This study was approved by the Ethics Committee in Human Research (CAAE: 66128317.5.0000.5541).The panelists received the FPCs and evaluated them for color, flavor, texture, and global appearance using a five-point hedonic scale, which ranged from 1 (very disgusted) to 5 (very good).The results were used to calculate the acceptance index (Maia et al. 2008).The untrained panelists were asked to share their favorite attributes.They were also asked to comment on the presence and intensity of rancid odor (oxidation), fish, alcohol, and gasoline odor (petroleum), based on a scale with the following options: absent (grade 1), weak (grade 2), moderate (grade 3), strong (grade 4), or very strong (grade 5).The panelists' personal information was obtained, such as gender, age, education, frequency of fish consumption, and whether they knew the product.

Statistical analysis
SPSS version 22 was used as the statistical program, and the Shapiro -Wilk test was used to assess the data normality after data rejection by Smirnov Grubbs.For the parametric data with independent samples using K samples, an average comparison was performed with analysis of variance (ANOVA), and an average multiple comparison was carried out with post hoc Tukey's test, with a 95% confidence interval.For sensory analysis, only descriptive analysis was performed.

Lipid extraction efficiency, water activity, and color assessment
Lipid extraction was more efficient with the use of n-hexane than with anhydrous ethanol and the Soxhlet method compared to single stage immersion (Table 1).Overall, FPC delipided in Soxhlet method using n-hexane as solvent (SH) achieved an extraction efficiency of 92%, which is approximately one-third greater than that of single stage immersion method using n-hexane (HI) and more than double that of FPC delipided in Soxhlet method using ethanol (SE).The result obtained based on the method was expected, as the Soxhlet method is known to promote periodic contact between the raw material and the pure solvent.Such contact enables the largest mass transfer gradient and, consequently, the largest extraction capacity (Brum and Arruda 2009).In contrast, the immersion method tends to create an equilibrium situation, minimizing the mass transfer gradient.Of note, the difference in extraction efficiency between the solvents is related to their polarity.All treatments resulted in Aw values between 0.20 and 0.28 (Table 1), with a reduction of 60% compared to WFP (0.83).The Aw was analyzed to evaluate the delipidification process.Aw has potential effects on the chemical reaction rate and the microbiological growth rate.Therefore, it is a good parameter for food stability (Agustini et al. 2009).Microbial activities that comprise raw material quality are found in foods and/or ingredients with Aw values of up to 0.61.Thus, the use of both ethanol and n-hexane followed by drying was found to be efficient at reducing Aw, corroborating previous studies carried out by Rebouças et al. (2012) and Vidal et al. (2011).
The delipidification method and solvent used directly influence lipid extraction efficiency.Thus, the maximum efficiency can be achieved as the process is improved.If the raw material used for the preparation of FPCs has high lipid content, the delipidification process becomes extremely important to produce low lipid FPCs, as rancidity is a major problem in the storage of fish products (Shaviklo et al. 2010).FPC with low Aw and low lipid content could have longer shelf life, can be stored at room temperature, and has no fish smell (Abraha et al. 2018;Jeyasanta et al. 2013;Senapati et al. 2016).
Color is an important quality attribute of fish protein ingredients.In general, the different delipidification processes influenced the final color of the products obtained (Table 1).Therefore, the lighting of the final samples and a yellow with red undertone dominated the colorimetric behavior (higher values of b* compared to a*).This colorimetric behavior was also reported in other studies on FPC (Costa et al. 2016;Ikasari and Donny 2018).The FPC color varies from light gray to yellow or red, depending on the type of fish used and the method of extraction (Shaviklo 2015).When the three colorimetric parameters were analyzed in sets (L*, a*, and b*), the SH samples were found to be the lightest, while the EI samples were the darkest.This finding corroborates the ΔE values (Table 1) and the actual colors of the final products (Figure 2).
Product color is a determining factor of consumer acceptance and consumer purchase intention (Abraha, Mahmud, et al. 2018).Therefore, the FPC's standardization production process is important to ensure its color.There was no specific pattern for the FPC standard color; however, samples with low yellowness (Table 1) had lower lipid levels (Table 2), while samples with higher yellowness had higher lipid levels.Carotenoids are yellow-orange-red liposoluble pigments that may be present in fish fat, since they have a biological function related to their lipid metabolism (Nakano and Wiegertjes 2020).Thus, it is possible to deduce that the yellowness of the FPCs will decrease as the lipids are removed.Such a finding indicates a directly proportional relationship between these variables (i.e., lipid reduction influences FPC's final color).

Proximal composition
FPCs have a higher protein content compared to in natura waste fish.The delipidification process described in this paper using organic solvents guarantees the major concentration of nutrients such as proteins, in addition to helping nature with the use of clean and environmentally-friendly solvents.Statistically, treatments using n-hexane as a solvent were found to have the best protein recovery, reaching levels above 80%.Furthermore, compared to raw material, protein increase was up to 78% for the treatments mentioned (Table 2).The fish by-products used to produce FPCs have high nutritional value (moisture 74%, crude protein 18%, lipids 4%, and ash 2%) and protein contents indicated for protein recovery (Ogawa and Maia 1999).
Oven-drying -a low-cost technology -has been reported for developing dried fish protein (Chavan et al. 2008).After the drying process, WFP showed a 75% moisture reduction in relation to the tambatinga by-products.However, such reduction is high for FPC, as the literature indicates that FPC products must contain a maximum of 10% moisture (Oetterer et al. 2006).After the delipidification process, another reduction in moisture rates was observed, as well as a significant increase in protein and mineral values, thereby demonstrating the best treatment among the options tested (Table 2).
However, it is important to mention that FPC is considered an ingredient that will be used for another product's manufacture and human consumption is not limited to a singular product.So, the quantity of the nutrients present in the products could be variable.
According to Archer et al. (2001) and Oetterer et al. (2006), there are three types of FPC, which differ by color, odor, lipid content, and protein content: A = white or light-yellow color, tasteless and odorless odor, lipid content less than 0.75%, and protein content between 60 and 90%; B = yellow or gray color, fishy odor, lipid content of 3%, and protein content less than 65%; and C = no specification of color and odor, no limits for lipids, and protein content less than 60%.Therefore, among the FPC discussed herein, only SH can be considered the closest to an FPC type A, due to its protein content (86%) and light-yellow color.

Mineral content
The mineral quality of an ingredient or raw material should be considered, as minerals are vital nutrients for the human body and contribute to the sensory and rheological properties of food products, especially flavor and texture (Belitz et al. 2009).The results obtained for the mineral content of the FPCs are shown in Table 3. Regardless of the delipidification method or solvent used, calcium, phosphorus, and potassium had the highest concentrations.Only potassium had a significant change in its content, with the highest concentrations recorded when n-hexane was used as the extracting solvent.Potassium is directly linked to the regulation of cellular osmotic pressure; calcium is involved in the structure of the skeletal and muscular system; and phosphorus, in combination with other elements (such as lipids), transports substances between the intracellular and extracellular environments (Belitz et al. 2009).

Protein digestibility and bioaccessibility of amino acids
Even if a product has considerable protein levels, this content must be bioaccessible to the body; that is, the ingested amount must be potentially available for absorption (Etcheverry et al. 2012).The protein WFP = Whole Fish Protein Concentrate; SE = Soxhlet method using ethanol; SH = Soxhlet method using n-hexane; EI = Single stage immersion method using ethanol; HI = Single stage immersion method using n-hexane.Date expressed in original matter.Equal letters in the same column indicated that they did not have statistical difference at p < .05level by Tukey's test.
digestibility of the FPCs is presented in Table 4. SH and HI showed the highest protein and digestibility values.In addition, some treatments caused the highest lipid extraction efficiency, indicating an inversely proportional relationship between lipid levels and protein digestibility of the samples.Such findings might be due to the interference of lipids in the action of digestive enzymes during protein hydrolysis (Menezes et al. 2017), resulting in lower digestibility in SE and EI.In addition, an explanation for the low protein digestibility using solvent ethanol concerns the polarity of protein present.Since the hydrophilic part of the protein is usually inside, the hydrophobic part would be exposed to the solvent in a helical structure, thus causing the aggregation of proteins through hydrophobic interaction (Yoshida et al. 2010), decreasing the solubility and consequently the digestibility of proteins.The protein profiles presented in Figure 3 were the same for both SH and HI, indicating that the method and/or temperature used in the delipidification process did not interfere with the protein pattern.The same finding was obtained for all replicates (R1 and R2) after the digestion simulation of both samples (SH-and HI-digested), in which the nine main protein bands (ranging from 62.69-12.18kDa) seemed to be highly digestible.The presence of bands from 55.95-31.27kDa for the digested samples is related to the enzymes used during digestion, as they are also observed for the control sample comprising digestive fluids.Bands of lower molecular weight, such as those at 11.58 and 10.69 kDa, were also present for the digested samples and indicate that peptides are formed during the hydrolysis process of digestion.Altogether, these data suggest that peptides and/or free amino acids can be generated from the main proteins found in FPCs and can be of benefit to the nutritional and/or biological functions of the body.
Due to their high protein digestibility, the FPC of the SH and HI treatments were also analyzed for their amino acid profile and bioaccessibility.The results are presented in Table 4.
The samples from treatments SH and HI showed similar amino acid profiles.According to the World Health Organization's (FAO 2002) recommendations for amino acid consumption by adults, the SH and HI values indicate amino acid quality.Among the evaluated amino acids, it is possible to highlight glutamine, glycine, arginine, and leucine, which had the highest composition values for the two analyzed treatments.Based on the essential amino acids alone, higher levels of threonine, lysine, leucine, and phenylalanine were found in the SH treatment.Further, the bioaccessible portion of each amino acid (Table 4) and their respective bioaccessibility rates (Figure 4) were found, with emphasis on the HI treatment, which obtained the best results for hydrophobic amino acids like alanine (90%), tyrosine (88%), and phenylalanine (72%) and hydrophilic amino acid lysine (73%).These advantages demonstrate FPC's innovative potential as an ingredient for use in food enrichment, mainly for children and vulnerable groups (Shaviklo 2015).
Although both methods used the n-hexane solvent, the temperature of the lipid extraction process applied in SH may have influenced the protein denaturation of the samples, with a consequent increase in protein digestibility and a decrease in amino acid bioaccessibility.Similar results were reported by Menezes et al. (2017).However, it is important to highlight that at high concentrations, all organic solvents are protein denaturants, but at low concentrations they can provide molecule stability (Ustunol 2015).This justifies the highest levels of bioaccessibility of amino acids for single stage immersion, which used a concentration of 3:1 (solvent:sample) compared to the use of Soxhlet that used 10:1.
To the best of our knowledge, the identification of amino acid bioaccessibility was not performed in other studies that used FPC.Therefore, the present study is a pioneer in this analysis.

Sensory analysis
Sensory analysis was only performed with the SH sample, as this treatment presented the best results from a nutritional point of view.The results are presented in Table 5.
Of 108 panelists that participated in the analysis, 50% were males, 47% females, and 3% chose not to declare their gender.Further, 92% of the panelists were between 18 and 25 years old, and 94% declared that they were graduate students.About 44% of the panelists consumed fish 2-3 times per month, and 35% consumed fish only on special dates, such as religious festivities.The justifications for their low fish consumption included vegetarianism, preferences for other meats, the presence of fishbone, consumption habits, product prices, and lack of knowledge on how to prepare fish for consumption.These justifications are similar to those found by other authors (Maciel et al. 2015;Ribeiro et al. 2018).
In the present study, among the most preferred attributes, color received 78% of the evaluations, followed by texture (72%), and global appearance (72%), indicating that a lighter color and flour-like texture are recommended for this product type.
All the attributes evaluated (Table 5) had an acceptance index above 70%, except for the flavor attribute (52%).Further, the color attribute received the highest score from the panelists.However, for a product to be accepted in the market, its acceptance index must present at least 70% of positive responses (Maia et al. 2008).The low marks attributed to the aroma can be justified by the weak or moderate rancid odor and strong or very strong fish odor reported for the samples by 56% and 63% of the panelists, respectively.Similar results were observed in the FPC resulting from the Nile tilapia (Oreochromis niloticus) fileting process (Rebouças et al. 2012;Vidal et al. 2011).
Based on the panelists' description of the aromas for the evaluated FPCs, approximately 27% reported a weak alcohol odor, and 11% reported a weak gasoline odor.The same finding was obtained in the study by Rebouças et al. (2012), where the evaluators found an alcoholic aroma in the FPC, originating from the ethanol used in the delipidification step.These results indicate that the solvents used were not completely volatilized and were not effective for FPC deodorization.
Regarding knowledge of FPC, 90% of respondents in this study reported they had never heard of the product, unlike the study by Vidal et al. (2011), where approximately 56% of the evaluators claimed to have knowledge.The results obtained were already expected, as the FPC is still not widespread in the studied community, which makes the dissemination of this type of study of paramount importance to the community.
Notably, there is no specific legislation for FPC.Therefore, as there is no production standard, each study mentioned in this article developed FPC using their own approach, which may also account for the differences between the results.
In conclusion, the delipidification process has a considerable impact on the final nutrient content in FPC, especially lipids, proteins, and amino acids.The single stage immersion method using n-hexane obtained better results for amino acid bioaccessibility.However, the use of this solvent, associated with the Soxhlet method, made it possible to obtain the highest protein content, greater lipid extraction efficiency, and better protein digestibility with considerable amino acid content, making it an interesting method for the food industry.Thus, FPC obtained from tambatinga residues can be considered an alternative protein matrix to be implemented in the preparation and/or fortification of different types of food, especially those intended for child consumption, as children are one of the groups with the greatest nutritional needs.

Disclosure statement
No potential conflict of interest was reported by the author(s).

Figure 1 .
Figure 1.Process of recovering fish protein from tambatinga's by-products.Whole fish protein (WFP); fish protein concentrate (FPC) delipided in Soxhlet with ethanol (SE) and n-hexane (SH), and single stage immersion with ethanol (EI) and n-hexane (HI).

Figure 2 .
Figure 2. Samples of the different FPC elaborated: whole fish protein concentrate (WFP); fish protein concentrate delipided in Soxhlet with ethanol (SE); fish protein concentrate delipided in Soxhlet with n-hexane (SH); fish protein concentrate delipided in single stage immersion with ethanol (EI); and fish protein concentrate delipided in single stage immersion with n-hexane (HI).

Figure 3 .
Figure 3. SDS-PAGE electrophoresis and bioaccessible fractions of fish protein concentrates (FPC) delipided in Soxhlet with n-hexane and single stage immersion with n-hexane (HI).R1 and R2 = replicates.LMWS = low molecular protein standard solution.Arrow indicates the migration direction.

Table 1 .
Results (standard deviation) for lipids extraction efficiency, water activity and color of whole fish protein of tambatinga byproducts and FPC delipidification in Soxhlet and single stage immersion with ethanol and n-hexane.
LEE = lipids extraction efficiency; Aw = activity water; WFP = Whole Fish Protein Concentrate; SE = Soxhlet method using ethanol; SH = Soxhlet method using n-hexane; EI = Single stage immersion method using ethanol; HI = Single stage immersion method using n-hexane.Equal letters in the same column indicated that they did not have statistical difference at p < .05level by Tukey's test.Results were expressed as hunter values of L* (lightness).a* (redness and greenness) and b* (yellowness and blueness).

Table 2 .
Nutritional value (standard deviation) of tambatinga by-products in natura, FPC with whole fish protein concentrate and delipidated with ethanol and n-hexane by Soxhlet and single stage immersion processes.

Table 3 .
Minerals content of whole fish protein concentrate of tambatinga by-products and delipidated with ethanol and n-hexane by Soxhlet and single stage immersion processes.Single stage immersion method using ethanol; HI = Single stage immersion method using n-hexane; ND not detectable.Equal letters in the same line indicated that they did not have statistical difference at p < .05level by Tukey's test.

Table 4 .
Protein digestibility, amino acid profile and bioaccessibility in fish protein concentrate made with tambatinga filleting byproducts.Whole Fish Protein Concentrate; SE = Soxhlet method using ethanol; SH = Soxhlet method using n-hexane; EI = Single stage immersion method using ethanol; HI = Single stage immersion method using n-hexane.*Joint FAO/WHO/UNU (2002).

Table 5 .
Results of the attributes (%) evaluated in the sensory analysis and identification of odors (%) present in fish protein concentrate prepared from delipidated tambatinga by-products in Soxhlet using n-hexane solvent (SH).