**Development and Optimization of Djulis Sourdough Bread Using Taguchi Grey Relational Analysis**

#### **Pei-Ling Chung, Ean-Tun Liaw, Mohsen Gavahian and Ho-Hsien Chen \***

Department of Food Science, National Pingtung University of Science and Technology, Neipu 91201, Pingtung, Taiwan; plchung@tajen.edu.tw (P.-L.C.); alexliaw@mail.npust.edu.tw (E.-T.L.); mohsengavahian@yahoo.com (M.G.)

**\*** Correspondence: hhchen@mail.npust.edu.tw; Tel.: +886-8-7740402

Received: 5 July 2020; Accepted: 12 August 2020; Published: 20 August 2020

**Abstract:** Bakery products made from naturally fermented sourdough show a diversified flavor and nutritional profile. Djulis (*Chenopodium formosanum*), known as red quinoa or Taiwan djulis, originally cultivated by Taiwanese indigenous people in mountain areas in eastern and southern Taiwan, has a high nutritional value and characteristic properties. In the present study, a new bakery product (djulis sourdough bread) was developed and a combination of the Taguchi method coupled with grey theory was utilized to optimize the baking parameters (product formulation). Five main factors, i.e., djulis sourdough (A), hulled djulis (B), oil type (C), a mixture of bread flour (wet gluten content of 29.0%) and a high-gluten flour (wet gluten content of 35.5%) (D), and honey (E), (each at four levels) were chosen for the Taguchi experiment design (L16(4)5). Dependent parameters were the data from texture profile analysis (brittleness, springiness, cohesiveness, gumminess, and chewiness), color analysis *(L\*, a\*,* and *b\*),* and sensory evaluation (appearance, aroma, bitterness, sourness, chewiness, and overall acceptance) of the final product. Taguchi grey relational analysis successfully determined the optimal conditions based on combined parameters (5 factors), which highlighted the advantages of this innovative optimization technique. The result shows that the optimal formula for producing a djulis sourdough bread with the best texture, color, and sensory qualities was A3B1C1D2E2, i.e., 20% djulis sourdough, 0% addition of hulled djulis, 8% unsalted butter, 80% wheat flour + 20% high-gluten flour, and 10% honey, respectively. Such a novel application could be a reference for improving the quality of bakery products in the industry. Moreover, it seems that the new bakery product developed in this study has good potential to be commercially produced after further nutritional and economic analysis.

**Keywords:** bakery products; bread; djulis; food quality; optimization; product development; Taguchi grey relational analysis; texture profile analysis; sensory attributes; sourdough

#### **1. Introduction**

Djulis (*Chenopodium formosanum* Koidz.) belongs to the Amaranthaceae family and *Chenopodium* genus. According to Encyclopedia Britannica, the Amaranthaceae family includes about 175 genera and more than 2500 species. Many species, including beets and quinoa, are considered staple food crops, and some are cultivated as garden ornamental plants [1]. Among them, amaranth species are located mainly in tropical and subtropical areas. Grain amaranth yields tiny seeds that can be used as a grain to make flour, porridge, and other foods. For instance, amaranth grain can be processed to be added into several products including baby food, cakes, and cookies. In addition, amaranth grain has a high concentration of lysine, that is, an essential amino acid for the biosynthesis of proteins, which is vital for human tissue development and healing. Furthermore, this grain is rich in calcium, phosphorus, iron, potassium, zinc, vitamin E, and vitamin B-complex [2]. Previous studies have shown bioactive

effects for species of the Amaranthaceae family [3]. For example, Sánchez-Urdaneta et al. fed rats with breads made with amaranth (*Amaranthus dubius* Mart. ex Thell) flour and observed that consumption of amaranth-enriched bread enhanced lipid profiles of rats and prevented metabolic and cardiovascular diseases due to its hypoglycemic and hypolipidemic effects [4]. Moreover, an in vivo study revealed that phenylpropanoid extract from *Halosarcia indica* (Willd.) has analgesic and anti-inflammatory effects on Wistar albino rats [5].

In Taiwan, djulis is also called Taiwan djulis or Taiwan red quinoa and has different strains with diverse colors. This plant has been cultivated mainly in eastern and southern areas of Taiwan (Taitung and Pingtung) as a cereal crop and was previously used for worship purposes and decorations in seasonal festivals by Taiwanese indigenous people (Taiwan aborigines). In recent years, physicochemical and bioactive characteristics as well as preventive healthcare applications of this crop get attention [6]. For instance, Hong et al. demonstrated that djulis extract could protect skin from UV-induced damage [7]. Additionally, Lee et al. found that the early stages of chemically induced colon carcinogenesis were suppressed in mice after feeding them djulis for 10 weeks [8]. However, there is limited information about the possibility of using this crop for developing a bakery product with optimal characteristics.

The Taguchi method is a systematic approach for experimental design and analysis. Recently, this approach has gained popularity to be used in various sectors of the industry for new product development and quality improvement in an economical way. Previous work demonstrated that the Taguchi method was used for the optimization of the conditions for submerged culture at a laboratory-scale study and resulted in the development of an upscaled fermentation process that could yield a high concentration of monacolin K [9]. Although the application of other optimization approaches (e.g., response surface methodology) has been widely explored for food processing [10,11], there are only limited studies in the literature that explored the applicability of the Taguchi method for developing new bakery products. Moreover, it seems that the limitations of other optimization approaches (e.g., response surface methodology; RSM) can be addressed by the application of the Taguchi method. For example, A study conducted by Chen et al. showed the applicability of the Taguchi technique for the quality improvement of egg–shortening cakes [12]. Basically, when applying the Taguchi method in process optimization, the optimum combination is determined based on one quality characteristic at a time. However, in practice, in production lines usually the process involves more than one characteristic (multiple objectives). Consequently, the effects of non-linear interactions between control factors exist and cannot be ignored. Therefore, the grey system theory is an approach that can be employed to optimize multi-characteristic processes.

The theory of the grey system was proposed by Professor Deng Julong in the 1980s [13]. Grey relational analysis (GRA), procured from grey system theory, is a measurement technique to determine the relationship between sequences through the analysis of a limited number of data [14,15]. The relational grade is defined as measuring the relevance between or two sequences or two systems and can be used to describe the trend relationship between a reference sequence (objective or ideal sequence) and a comparative sequence in a specific system. In such studies, a relational grade approaching 1 suggests that the reference sequence and the comparative sequence tend toward concordance. On the other hand, a relational grade approaching 0 indicates that the reference sequence and the comparative sequence do not tend toward concordance completely. This technique requires small quantities of data, and the data are not restricted to specific statistical distributions, which makes GRA superior to classical statistical methods. In previous studies, Chen et al. used GRA successfully to investigate the adulterated cases of commercial soybean sauces [16]; Chen et al. employed GRA effectively to identify and classify undried roselle samples frozen at −20 ◦C, and roselle samples dried correctly at 20, 50, 75, and 85 ◦C, respectively [17]. Moreover, associating the Taguchi method with GRA has also shown a powerful tool to optimize the multiple performance characteristics in the food manufacturing process. Chen et al. applied the Taguchi grey relational analysis method to optimize the fish drying process based on performance characteristics such as color measurement value (*L\*, a\*, b\**), thiobarbituric acid

value (TBA), and shear stress value [18]. Chung et al. used a grey-based Taguchi approach to improve beneficial monacolin K, *Monascus* pigment synthesis, and to decrease the adverse metabolite, citrinin, in the fermentation of *Monascus purpureus* [19].

Despite the progress in the application of grey system theory in several fields, this application of this innovative approach is an ongoing topic in the food industry. For example, the grey system theory has not been well-explored for optimization of products such as sourdough breads, which are believed to have improved shelf life and sensory properties such as flavor, aroma, and texture (mainly due to fermentation by yeasts, *Aspergillus*, and lactic acid bacteria) [20,21]. Specifically, developing a new djulis sourdough-based bread need a tremendous optimization that has not been explored in the literature. Therefore, this study aims to develop a new naturally fermented bread product (djulis sourdough bread) with acceptable palatability by utilizing the Taguchi–GRA method as an innovative optimization approach that can serve as a potential practice for industrial baking. In this regard, such a novel approach was utilized through an overall evaluation process for the optimization of djulis sourdough bread manufacturing based on 14 characteristics (objectives) including attributes obtained by texture profile analysis (TPA), measurement of *L\**, *a\**, and *b\** values by a colorimeter, and sensory evaluation.

#### **2. Materials and Methods**

#### *2.1. Questionnaire and Experimental Strategy*

Thirty experienced bakers were invited to reexamine the standard formula used for making general commercial round-top white bread set up by the China Grain Products Research and Development Institute (New Taipei City, Taiwan) [22]. This standard formula is as follows (ratio of materials): 100% high-gluten flour (HGF), 10% fine granulated sugar, 8% butter, 4% fresh yeast, 54% water, 12% egg, 2% salt, and 4% milk powder. Based on the questionnaire responses, feedback, and discussions collected from these 30 experienced bakers, five potential influential factors that would affect bakery product quality were selected and identified for the experimental design and were then used to develop the djulis sourdough bread. These five factors were the addition/non-addition of djulis sourdough, honey, wheat flour (WF) or HGF, addition/non-addition of hulled djulis, and butter/oil. The abovementioned evaluations were performed to design an experiment with a specified number of tests as well as specified ranges for each test.

#### *2.2. Experimental Preparation*

Before running the experiment, a bunch of organic djulis was added and mixed with leftover baguette dough for four hours to form a djulis sourdough. Then the djulis sourdough was cultured in a refrigerator at 4–7 ◦C. Re-culturing was performed every seven days. In this regard, an appropriate amount of djulis sourdough was added into flour and water, kept at room temperature for three hours, and then refrigerated for the continuation of the low-temperature fermentation process.

The preparation of djulis sourdough bread was based on formulas derived from the factor/level assignments. Three fermentation processes were conducted for djulis sourdough preparation, i.e., the first fermentation was run at 28 ◦C, 75% of humidity for 60 min; after cutting and rounding, the second fermentation was run again at 28 ◦C, 75% of humidity for 15 min. Then after the appearance shaping, 38 ◦C, 75% of humidity for 50 min were applied in the third fermentation. After fermentation, the sourdough was baked for 35 min at 180–200 ◦C (top and bottom heat). The flow chart for making the djulis sourdough bread is shown in Figure 1.

**Figure 1.** A flow chart representing djulis sourdough bread preparation in the present study.

#### *2.3. Materials*

Djulis was collected from Machia Township (Pingtung County, Taiwan). A WF sample with a wet gluten content of 29.0%, protein content of 10.0%, and ash content of 0.6% and an HGF with a wet gluten content of 35.5%, protein content of 12.5%, and ash content of 0.4% were obtained from Yuan Shan Food Co., Ltd. (Pingtung, Taiwan); Anchor unsalted butter, originally from New Zealand, obtained from Tehmag Foods Co. (New Taipei City, Taiwan); camellia oil (Yuan Shan Food Co., Ltd., Pingtung, Taiwan); Italian olive oil obtained from Tehmag Foods Co. (New Taipei City, Taiwan); lard (I-MEI Foods Co., Ltd., Pingtung, Taiwan); honey (Longan honey, The Chen's Honey, Pingtung, Taiwan); eggs (PX Mart, Pingtung, Taiwan); Anchor full cream milk powder, originally from New Zealand, obtained from Yu Hsuan Inc. (Pingtung, Taiwan); salt (Yu Hsuan Inc., Pingtung, Taiwan); charcoal-filtered water (PX Mart, Pingtung, Taiwan); and yeast (Yu Hsuan Inc., Pingtung, Taiwan).

#### *2.4. Sensory Evaluation and Instrumental Measurement*

#### 2.4.1. Sensory Evaluation Analysis

The seven-point hedonic scale was adopted to assess the overall round-top djulis sourdough bread acceptability. Such a test was performed by 60 participants (between 20 and 35 years old) who were chosen randomly from students of the National Pingtung University of Science and Technology (NPUST). Appropriate guidance and training for all panelists were provided before conducting the sensory evaluation. Four categories of sensory attributes were stated on the score sheet as follows: appearance, smell/taste (aroma, bitterness, and sourness), texture (chewiness), and overall acceptability. Each item was scored between 1 and 7 (1: dislike extremely, 2: dislike moderately, 3: dislike slightly, 4: neither like nor dislike, 5: like slightly, 6: like moderately, 7: like extremely). Items could not be scored more than once. There were a total number of 16 slices of bread that needed to be practiced by a panelist per day. After the evaluation had been completed for one slice of bread, a 15 s interval was required before the next practice. During each evaluation, the external appearance of the sliced bread was first observed and scored. Afterward, the scoring was performed one by one for aroma, bitterness, sourness, chewiness, and overall acceptability. The sensory evaluation tests were performed in triplicate (*n* = 3) on three days, meaning that each member of the sensory evaluation panel (60 members) need to repeat the tests three times on different days. The final score of each item was calculated and obtained by averaging the three-replicate data.

#### 2.4.2. Texture Profile Analysis

Oven-fresh round-top djulis loaves were subjected to TPA using a texture analyzer (TA-XT-Plus, Stable Micro Systems, Ltd., Godalming, UK) based on a standard method according to approved methods of the American Association of Cereal Chemists (AACC), as described by Amigo et al. [23]. First, each sample was sliced into 1.25-cm-thick slices, the slices at both ends were discarded, and the slices from the middle portion of each sample were used for analysis. During the testing process, two bread slices were stacked and analyzed using a 36-mm-diameter probe, with a compression ratio of 25% and a probe compression speed of 5 m/s. Each type of sample was compressed four times using eight slices from the middle portion, and data related to brittleness, springiness, cohesiveness,

gumminess, and chewiness were recorded and calculated using the developed software (TA-XT-Plus, Stable Micro Systems, Ltd., Godalming, UK) [24,25].

#### 2.4.3. Colorimeter Analysis

After cooling for one hour, round-top djulis loaves were sliced into 1.25-cm-thick slices, and slices from the middle portion of each sample were subjected to color and lightness measurements using a colorimeter (Minolta CR 310, Konica Minolta Sensing Singapore Pte. Ltd., Jurong East, Singapore). Color and lightness values were expressed as *L\**, *a\**, and *b\**, with *L\** representing lightness (*L\** for brightest white = 100) or darkness (*L\** for darkest black = 0); *a\** representing the red/green component (+*a\**: red, −*a\**: green); and *b\** representing the yellow/blue component (+*b\**: yellow, −*b\**: blue). The average value of six measurements was used for each parameter [26].

#### *2.5. Data Analysis Models*

Analyses of multiple quality characteristics of the baked djulis sourdough bread samples were performed using the novel combination approach, the Taguchi–GRA method, as shown in the following.

#### 2.5.1. Taguchi Method

With the Taguchi method [27], an orthogonal array is first constructed by assigning known or assumed control factors and noise factors. Accordingly, the optimal parameter levels are determined with the minimum number of experiments. The orthogonal array is denoted by Ln(Xm), where n is the number of columns of the array (i.e., the number of parameter and level combinations in the experiment), X is the number of levels, and m is the number of rows of the array (i.e., the number of factors). The orthogonal array used in the present study is denoted by L16(45), meaning that five control factors with four levels were used in 16 bakery product experiments. The five control factors used in this study were djulis sourdough (A), hulled djulis (B), butter/oil (C), Taiwan flour (D), and honey (E). Four levels (the ratio of formula) were set for each control factor (Table 1) in which the common ingredients were 3.5% fresh yeast, 54% distilled water, 12% egg, 2% salt, and 4% milk powder. The selected orthogonal array L16(45) and factor/level assignments are shown in Table 2 as the mean of three replicates.


**Table 1.** Parameter design factors and levels \*.

\* The common ingredients were 3.5% fresh yeast, 54% distilled water, 12% egg, 2% salt, and 4% milk powder. \*\* WF: wheat flour; HGF: high-gluten flour.

#### 2.5.2. Calculation of S/N (Signal-to-Noise Ratio) Values

Experimental data of those multiple quality characteristics in the orthogonal table were used to calculate the signal-to-noise ratio (S/N ratio, η). The S/N ratio did create a transformation function of the repetition data to another value and was used as a measure of the variation present in the experiment. S/N is a function indicator that measures performance, with higher S/N values indicating smaller quality losses. There are three types of quality characteristics for S/N values: nominal-the-best, smaller-the-best, and larger-the-best. In this study, we aimed to find the optimal operational parameters for the manufacturing of djulis sourdough bread retaining taste, nutrients, and supple flavors. Accordingly, the larger-the-best loss function was, therefore, used to calculate the S/N ratio as described in Equation (1).

$$\eta = -10 \log \left( \frac{1}{n} \sum\_{i=0}^{n} 1/y\_i^2 \right) \tag{1}$$

where *yi* is the *i*th value of the quality attribute, and *n* is the number of trials.


**Table 2.** L16 (45) orthogonal array and factor/level assignments \*.

\* A: djulis sourdough; B: hulled djulis; C: butter/oil (8%); D: wheat flour + high-gluten flour; E: honey.

#### 2.5.3. Algorithm of GRA

GRA was used to develop multiple quality characteristics of the djulis sourdough bread, to assess the optimal combination of parameters that best satisfies all the quality characteristics and to proceed with overall evaluation. These quality characteristics (dependent parameters) include color values, sensory attributes, and textural property. Data pretreatment was performed before employing GRA for data normalization, i.e., normalizing the raw data or their S/N ratios in the range of 0–1. All these characteristics and their S/N ratios were in the nature of the larger-the-better characteristics.

According to the literature [12,13,16,17], normalized functions could firstly be represented as Equation (2).

$$X\_i^\*(k) = \frac{X\_i(k) - \min[X\_i(k)]}{\max[X\_i(k)] - \min[X\_i(k)]} \tag{2}$$

where *X*∗ *i* (*k*) is normalized raw data, *X*∗ *i* (*k*) is a comparative sequence with *k*th entities, *i* = 1, ... , *m*; *k* = 1, ... , *n*; and max[*Xi*(*k*)] and max[*Xi*(*k*)] are the maximum and minimum ones in the comparative sequence.

The grey relational grade (GRG) could depict the degree of relationship between a reference sequences (objective sequence or ideal sequence) and a comparative sequence in which it is comprised of 14 characteristics including attributes obtained by TPA; measurement of *L\**, *a\**, and *b\** values; and sensory evaluation. Equations (3)–(5) were used for the calculations related to GRA.

Let *X*<sup>0</sup> *(k)* be the reference sequence with *k*th entities, that is,

$$X\_0 \ (k) = \langle \mathbf{x}\_0 \ (1), \mathbf{x}\_0 \ (2), \dots, \mathbf{x}\_0 \ (n) \rangle,\tag{3}$$

where *k* = 1, 2, 3, ... , *n*.

Let *X\*j (k)* be the compared sequence; each *X\*j* possess the same number of entities as *X*0, that is,

$$X^\*\_{\\_}(k) = \langle \mathbf{x}^\*\_{\\_}(1), \mathbf{x}^\*\_{\\_}(2), \dots, \mathbf{x}^\*\_{\\_}(n) \rangle,\tag{4}$$

where *k* = 1, 2, 3, ... , *n*.

The grey relational coefficient between the reference sequence of *X*<sup>0</sup> and the compared sequence X\**<sup>j</sup>* and at the *k*th entity are described in Equation (5):

$$\gamma \left( X\_0(k), X^\*\_{\ j}(k) \right) = \frac{\Lambda \text{min} + \xi \Lambda \text{max}}{\Delta\_{0j}(k) + \xi \Lambda \text{max}} \tag{5}$$

where


The optimal settings of process parameters combine multiple quality characteristics into one integrated numerical value, that is, GRG. This parameter for the sequence of *X\*j* is represented in Equation (6).

$$\Gamma\_{0j} = \Gamma\{X\_0, X\_j\} = \sum\_{k=1}^n w\_{k\uparrow} \left(X\_0(k), X\_j(k)\right) \tag{6}$$

where *Wk* is the *k*th weighting of γ0*j*.

The value of the GRG (Γ0*<sup>j</sup>* in Equation (6)) represents the level of similarity between the comparative sequence *X*\**<sup>j</sup>* (the *j*th of the experimental trials) and the referential sequence *X*0. The GRG of each experimental trial can be treated as a response (Γ*j*) for each row of the orthogonal array of Table 2. The response graph can be set up by grouping the response values of the corresponding same factor levels of the column in the array, taking the sum, and dividing by the number of responses, as follows:

$$L\_j = \frac{\sum\_{j=1}^{n} \Gamma\_j}{n} \tag{7}$$

where Γ*<sup>j</sup>* is the response value of corresponding same factor levels of the column in the array, *Lj* is the mean response of the corresponding factor level.

#### *2.6. Statistical Analysis*

All the experiments were analyzed in triplicate. The analysis of variance (one-way ANOVA) was applied to the date to determine the significance of influences of control factors used for making djulis sourdough bread and was performed using an SPSS Statistics V.22.0 for Windows Statistical package (IBM Corporation, Armonk, NY, USA). The differences were significant statistically when *p* < 0.05 using the Duncan multiple range tests.

#### **3. Results and Discussion**

#### *3.1. Appearance and Bread Volume*

Sixteen round-top djulis sliced bread samples were prepared using different formulas based on the L16(45) orthogonal array and factor/level assignments (Tables 1 and 2). These samples were numbered as samples No. 1–16 as can be seen in Figure 2. The first four types of bread in Table 2 (samples No. 1–4 in Figure 2) were prepared without the addition of djulis sourdough. The average length, width, and height of these bread were 30.3, 10.3, and 13.1 cm, respectively. The other 12 types of samples that were prepared with the addition of djulis sourdough (samples No. 5–16 in Figure 2) had smaller sizes, that is, the average length, width, and height of these sourdough breads were 29.9, 10.0, and 11.88 cm, respectively. These observations suggest that the incorporation of djulis in the sourdough bread may reduce the loaf rising. Similarly, previous studies showed that change in the formulation may affect the loaf volume [28,29]. Such changes could be related to several parameters including the effect of the formulation on the gluten network as well as on the fermentation process.

**Figure 2.** Sixteen numbered round-top djulis bread sliced samples (No. 1 to No. 16) made of different formulas as described in Table 2. No.: number.

#### *3.2. Sensory Evaluation, Texture, and Color Analysis*

The raw sensory data from questionnaires (Supplementary Materials; Figure S1) were used to calculate the results of the sensory evaluation of the 16 sliced bread samples (Table 3). According to the results, the trial No. 9 (20% djulis sourdough, 0% addition of hulled djulis, 8% lard, 40% WF + 60% HGF, and 10% honey) consistently gained higher scores on sensory attributes, including appearance, aroma, bitterness, sourness chewiness, and overall acceptance (Table 3). As djulis would release a bitter taste, the sensory evaluation of trial No. 9 showed the highest score on bitterness (less bitter, in the level of like slightly). In terms of sourness, sample No. 9 got the highest score, which showed that incorporation of djulis in the formulation can affect the sourness of the bread. This could be related to both the direct effect of djulis on the final taste of the product as well as its effect on the fermentation process and fermentation products. Similarly, for the appearance the bread, the sensory evaluation team preferred the appearance of sample No. 9. As can be seen in Figure 2, this sample has a unique distribution of air bubbles and color, which is related to the different formulation compared to other samples. Additionally, the panelists found sample No. 4 more chewy, which is in line with the observation about the air bubble distribution (Figure 2) as well as the reduced volume of the sample. This observation depicted that the sensory panelists in this study did not like bitterness in djulis sourdough sliced bread. Nevertheless, it could not be concluded that trial No. 9 was the best product among all bread samples only based on these sensory observations. Objective criterion data from instrumental analyses such as TPA and colorimeter analysis are also equally crucial parameters that should be included in the calculation (Table 3). In other words, an overall evaluation, based on both sensory and instrumental data, is necessary to determine the best set among others. In such cases, different results might be generated depending on the objectives of interest. According to the instrumental data, the addition of djulis affected the color values and textural properties of the final product. It was observed that the addition of djulis can alter the bread color values. Moreover, the instrumental texture analysis data were in line with those of the sensory evaluation. For example, similar to the panelists, TPA also confirmed that sample No. 9 is among the chewiest samples. The results of the present study were in line with those reported in the literature. For example, researchers observed a notable impact of formulation on the textural attributes of bread [28,29].

#### *3.3. Calculation of S*/*N Values of Quality Characteristics*

Table 3 shows data on the 14 quality characteristics obtained from sensory evaluation, texture analysis, and colorimetric analysis. Using the Taguchi method, the values of quality characteristics were transformed into S/N values (Table 4), which were then used to determine a formulation with the best quality and lowest variance. When the Taguchi method is used for process optimization, in some cases, a single quality characteristic is set as the target. As a result, experimental results can be shown as a simple linear relationship through the calculation of S/N values, and the best experimental combination can be directly determined from the response graph of the Taguchi orthogonal array. However, in practical production lines of the bakery industry, the investigation of a single quality objective is extremely rare. It means that, for food processing and product development, a number of parameters affect the overall quality of the product. Therefore, process optimization based on a single quality parameter usually cannot provide practical information for the industry. In the present study, 16 experimental trials with multiple quality objectives had to be investigated at a time. As the different quality characteristics had different units and attributes, data incomparability existed in the sequences. Therefore, data preprocessing was required to convert the data of the sequence with distinct scale and dimension into ones having a consistent unified scale and no dimension. The optimization practice in this study has successfully taken into account various processing and quality parameters. Therefore, the grey relational analysis could be employed with available comparable sequences [17].



162

followed by standard deviation (SD), i.e., mean ± SD. For sensory evaluation data, *n* = 3 means that 60 panelists repeated the sensory tests on 3 different days.


 table.


#### *Foods* **2020** , *9*, 1149

#### *3.4. Grey Relational Analysis*

Before using GRA, the S/N values of the various quality characteristics were preprocessed and converted into normalized values ranging from 0 to 1 through Equation (2), as shown in Table 5. The normalized data possessed good consistency and satisfied the three basic conditions for sequence comparison mentioned in the previous section. As the larger-the-best S/N values were calculated for the multiple quality characteristics of the present study, the value of the reference sequence for GRA was set to 1. Therefore, the normalized GRG closer to 1 in the sequence indicated greater closeness to the target value.

With the calculation of GRG, the distinguishing coefficient ζ is set within the range of zero to one (0 < ζ ≤ 1). This ensures that the maximum sequence difference Δmax does not become excessively large and causes a loss of the influencing power of the minimum sequence difference Δmin. Although excessively high or low values of ζ will lead to linear biases in data [13], the main function of ζ is to adjust the degree of contrast between the background value and the object being tested. Therefore, the value can be adjusted based on actual needs, as changes in the value of ζ only lead to changes in relative values without affecting the order of the GRG [13]. In the present investigation, a value of 0.5 was used for ζ.

Weightings were assigned to the quality characteristics. Fourteen quality characteristics, including appearance, smell/taste (aroma, bitterness, and sourness), texture (chewiness), overall acceptance, TPA attributes (gumminess, chewiness, brittleness, springiness, and cohesiveness), and colorimetric values (*L*, *a\*,* and *b\**), were classified into four categories based on the characteristics preferred by consumers who purchase bakery products, and a weighting of 1/4 was assigned to each category (Table 6). The reference sequence (ideal values) was chosen as *X*<sup>0</sup> *(k)* = (1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1), in which all 14 characteristics employed the concept of the larger-the-better for this work. All 16 sequences were treated as comparative sequences, and each sequence was composed of 14 characteristics (entities). The GRGs of 16 comparative sequence, which were calculated according to Equations (2) and (6), are presented in the last column of Table 6. Then, the grey relational analysis user interface was developed accordingly, to perform computer computing instead of manual calculation (Figure 3).



*Foods* **2020** , *9*, 1149

kgf:

kilogram-force;

 OA: overall acceptance; *L*\*: lightness; *a*\*: green–red coordinate; *b*\*: blue–yellow coordinate.


Grey relational analysis (GRA) calculation model (<sup>ζ</sup> = 0.5).

> **Table 6.**

\* OA: overall acceptance; *L*\*: lightness; *a*\*: green–red coordinate; *b*\*: blue–yellow coordinate.


#### (**A**)


**Figure 3.** (**A**) First three results obtained directly from the GRA user interface; (**B**) last three results obtained directly from the GRA user interface.

(**B**)

#### *3.5. Parameter Optimizaion*

#### 3.5.1. Optimal Factors and Levels

The larger the GRG, the closer the product quality to the objective value. For instance, in Table 6, experimental trial No. 9 seems to be acceptable and closer to the reference sequence (ideal sequence), in which the highest GRG, 0.8675, was obtained. Accordingly, the order of trials (No. 1–No. 16) was rearranged as No. 9 (0.8675) > No. 11 (0.7131) > No. 13 (0.7026) > No. 1 (0.6163) > No. 16 (0.6016) > No. 10 (0.5861) > No. 14 (0.5855) > No. 6 (0.5710) > No. 2 (0.5439) > No. 3 (0.5354) > No. 11 (0.5290) > No. 15 (0.4930) > No. 12 (0.4881) > No. 8 (0.4839) > No. 4 (0.4690) > No. 5 (0.4597). However, non-linear interrelations were observed between the processing parameters in bakery manufacturing. A previous investigation showed that the mean GRG of each experimental trial could be regarded as the response and processed to determine optimal combinations of process parameter levels when a system with multiple performance characteristics is evaluated [18].

Based on the L16(45) orthogonal array (Table 2), the GRG for all 16 experimental trials (Table 6), and Equation (7), mean responses of the control level were calculated by taking the sum of the same levels in the column divided by the number of levels. For instance, values of A1, A2, A3, and A4 were 0.541 ((0.6163 + 0.5439 + 0.5354 + 0.4690)/4), 0.511 ((0.4597 + 0.5710 + 0.5290 + 0.4839)/4), 0.664 ((0.8675 + 0.5861 + 0.7131 + 0.4881)/4), and 0.596 ((0.7026 + 0.5855 + 0.4930 + 0.6016)/4), respectively. Likewise, mean responses were calculated for all factor levels of B, C, D, and E to generate a response graph (Figure 4). In this response graph, the highest value of the level for each factor represents the strongest effect. Therefore, optimal parameters were selected based on the highest response values in Figure 4, which were A3 (20% djulis sourdough), B1 (0% addition of hulled djulis), C1 (8% unsalted butter), D2 (80% WF + 20% HGF), and E2 (10% honey). The suggested condition (A3B1C1D2E2) is a generated optimal factor-level combination, which is not among the 16 experimental trials that are listed in Table 2. Although the present study mainly focused on the technological, physical, and sensory properties of the newly developed djulis sourdough bread, it should be noted that the new product developed in the present study can possess unique nutritional values considering previous reports about the bioactive effects of djulis [7]. Further investigation about the nutritional profile, bioactive effects, and potential healthcare applications of this product can be investigated in future studies. Similarly, studies regarding the properties of sourdough bread incorporated with other strains of djulis can be suggested as researchers showed that bread ingredients (e.g., wheat flour) originated from different locations possess various bioactive compounds that can affect the nutritional characteristics of the final product, which might be the case for djulis sourdough bread [30].

#### 3.5.2. ANOVA Analysis

The results of ANOVAs indicate the degrees of influence over multiple quality characteristics (Table 7). As GRA provides a comprehensive analysis of the various characteristics, in the way of balanced consideration of the point of view of consumers, the influences of conflicting factors had already been weakened during the analysis process. Table 7 shows that all five factors (addition/non-addition of djulis sourdough, addition/non-addition of hulled djulis, butter/oil type, WF + HGF, and honey) significantly influenced the quality characteristics of the naturally leavened sourdough bread developed in the present study.

**Figure 4.** Response graph of grey relational grades of djulis sourdough bread. WF: wheat flour; HGF: high-gluten flour.


**Table 7.** The summary of ANOVA results.

\* SS: sum of squares. \*\* DF: degrees of freedom. \*\*\* *p* < 0.001.

#### **4. Conclusions**

The present study demonstrated that a combination of Taguchi and grey relational analysis, i.e., a Taguchi–GRA approach, could be employed to investigate the effects of processing parameters on the quality of djulis sourdough bread and to identify the optimal settings for manufacturing new bakery products when multiple characteristics are involved. Such multiple characteristics of bread (e.g., aroma, color, and texture) are important for consumers. At this moment, it seems that no systematic approach has been implanted in the bakery industry that can consider multiple objectives at a time for developing new products. Therefore, the Taguchi–GRA approach, which was explored in this study, could be a prospective optimization technique that can be implemented in the bakery and other sectors of the food industry. The novel Taguchi–GRA approach introduced in this study could provide a reference of the basis for the enhancement of consumer-oriented products in the bakery industry. Furthermore, it was depicted that, sometimes, sensory evaluation could not be the only decisive approach to determine the optimal bakery products. Therefore, a combination of instrumental and sensory analysis can provide realistic data to develop a product with optimal quality parameters. Further studies in the nutritional aspects of such an innovative product can be suggested for future studies.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2304-8158/9/9/1149/s1, Figure S1: The raw sensory data from questionnaires.

**Author Contributions:** Conceptualization, H.-H.C. and P.-L.C.; methodology, H.-H.C. and E.-T.L.; software, H.-H.C. and P.-L.C.; validation, H.-H.C., E.-T.L. and M.G.; formal analysis, H.-H.C.; investigation, P.-L.C.; resources, H.-H.C.; data curation, H.-H.C. and P.-L.C.; writing—original draft preparation, H.-H.C. and P.-L.C.; writing—review and editing, H.-H.C. and M.G.; visualization, H.-H.C. and M.G.; supervision, H.-H.C.; project administration, H.-H.C.; funding acquisition, H.-H.C.. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

#### *Article*

### **E**ff**ects of the Addition of Flaxseed and Amaranth on the Physicochemical and Functional Properties of Instant-Extruded Products**

**Jazmin L. Tobias-Espinoza 1,2, Carlos A. Amaya-Guerra 2, Armando Quintero-Ramos 1,\*, Esther Pérez-Carrillo 3, María A. Núñez-González 2, Fernando Martínez-Bustos 4, Carmen O. Meléndez-Pizarro 1, Juan G. Báez-González <sup>2</sup> and Juan A. Ortega-Gutiérrez <sup>5</sup>**


Received: 2 May 2019; Accepted: 27 May 2019; Published: 30 May 2019

**Abstract:** The addition of flaxseed and amaranth on the physicochemical, functional, and microstructural changes of instant-extruded products was evaluated. Six mixtures with different proportions of amaranth (18.7–33.1%), flaxseed (6.6–9.3%), maize grits (55.6–67.3%) and minor ingredients (4.7%) were extruded in a twin-screw extruder. Insoluble and soluble fiber contents in extrudates increased as the proportions of amaranth and flaxseed increased. However, the highest flaxseed proportion had the highest soluble fiber content (1.9%). Extruded products with the highest proportion of flaxseed and amaranth resulted in the highest dietary fiber content and hardness values (5.2 N), which was correlated with the microstructural analysis where the crystallinity increased, resulting in larger, and more compact laminar structure. The extruded products with the highest maize grits proportion had the highest viscosity, expansion, and water absorption indexes, and the lowest water solubility index values. The mixtures with amaranth (18.7–22.9%), flaxseed (8.6–9.3%), and maize grits (63.8–67.3%) resulted in extruded products with acceptable physicochemical and functional properties.

**Keywords:** extruded products; flaxseed; amaranth; dietary fiber; extrusion-cooking

#### **1. Introduction**

Currently, an increasing trend in the demand for processed foods that include pro-health compounds such as soluble fiber is occurring due to evidence of potential health benefits to consumers. Reduction in various types of chronic diseases such as cancer, cardiovascular disease, type II diabetes, and various gastrointestinal disorders are among them [1]. The development of products and processes that incorporate high-fiber ingredients without altering the physical, functional, and sensory properties

of the processed foods are of interest and desirable to meet current consumer trends. A technological alternative for the incorporation of ingredients in processed products is the extrusion-cooking process, a very versatile technique widely used for the development of breakfast cereals and instant foods [2]. Some extruded products marketed as breakfast cereals have significant caloric value. One of the strategies that has allowed the food industry to reduce the energetic density and produce more healthy products has been the incorporation of dietary fiber. However, the addition of dietary fiber during extrusion, especially if it is insoluble, results in products with less expansion and crispness and a higher bulk density and hardness, which are properties less preferred by consumers [3]. These characteristics can be explained by the interactions of fiber with starch that impact the mechanisms of starch gelatinization. These, in turn, are related to water absorption of the extrudates and other physicochemical transformations that occur during extrusion, such as viscoelastic properties associated with the stabilizing membranes of the bubbles formed during bubble growth in the final product [4]. Adding insoluble fiber to extruded products has been shown to decrease the proportion of starch in the food matrix, thereby reducing the water absorption capacity and, in turn, the viscosity caused by the gelatinization of the starch, which results in a reduction in the expansion of the final product [3]. A sectional reduction in the expansion of extruded products had been reported by Brennan et al. [5], due to an increase in insoluble fiber results in structures with a high number of small cells and a high cell density. Several authors reported that the bulk density is increased by adding insoluble fiber to extruded products [5,6]. Also, it has been reported that extrusion process causes significant effects on the dietary fiber content, as breakage of structural polysaccharides or complex carbohydrates formation [3]. These effects include the formation of resistant starch that may occur during extrusion and the formation of covalent interactions between macronutrients and insoluble components (such as insoluble fiber) that make the extrudates indigestible by amylase or protease activity [7].

The development of ready-to-eat products with high fiber content (6 g of fiber/100 g of product) [8], is based on the use of ingredients that meet this requirement and at the same time, provide numerous health benefits. Flaxseed and amaranth have been used specifically in extruded products due to their health benefits for consumers, but, usually, they are used individually [9–13]. Amaranth contains a good balance of amino acids, including the essential amino acid lysine, which is present in limited amounts in most cereals [14,15]. Flaxseed is low in carbohydrates (sugars and starches), high in fiber and protein, and rich in polyunsaturated fatty acids, particularly alpha-linolenic acid (ALA or ALN) and linoleic acid (AL), known as omega-3 and omega-6 essential fatty acids, respectively [16,17]. An important component of these two grains is soluble fiber, considered a functional ingredient because it generates high-viscosity products by causing the gelation of chyme. Chyme acts as a network to capture glucose and cholesterol molecules in their passage through the gut, hindering their absorption and thereby decreasing blood glucose and cholesterol levels, resulting in beneficial health effects [18].

Despite the health benefits of amaranth and flaxseed, little information has been reported on the combined effect of both ingredients on the physicochemical and functional properties of extruded products. Therefore, the aim of this study was to develop an extruded product with high-fiber ingredients and to evaluate the effects of the addition of flaxseed and amaranth on the physicochemical and functional properties of instant-extruded products.

#### **2. Materials and Methods**

#### *2.1. Materials*

Grains of amaranth (*Amaranthus hypochondriacus* L.), flaxseed (*Linum usitatissimum* L.) and minor ingredients such as sucralose, cocoa, and cinnamon were obtained from a local distribution store (Chihuahua, Chihuahua, Mexico). Also, maize grits number 4 (GPC, Muscatine, IA, USA) was used for the extrusion food matrix. The amaranth and flaxseed grains were milled in a roller mill (Zhengzhou Chengli Grain & Oil Machinery Co., Ltd., model 6F-2240, Zhengzhou, China), separately and sieved in a mesh number 35 (model RX-24, Tyler industrial products, Mentor, OH, USA) to obtain a particle size of 0.5 mm. All materials were stored in plastic bags at room temperature until their use.

#### *2.2. Chemicals*

Hydrochloric acid 37.2%, sulfuric acid 97.9%, hexane 99.8%, ethanol 99.9%, and boric acid 99.5% were all analytical grade and obtained from J.T. Baker (Mexico City, Mexico). Analytical grade sodium hydroxide (97.0%) was obtained from Sigma-Aldrich (St. Louis, MO, USA). The selenium reaction mixture was obtained from Merck (Darmstadt, Germany). The kit for soluble and insoluble dietary fiber was obtained from Sigma-Aldrich (St. Louis, MO, USA) [19].

#### *2.3. Methods*

#### 2.3.1. Mixtures Preparation

Different proportions of amaranth and flaxseed flours were mixed with maize grits and fixed minor ingredients (sucralose, cocoa, and cinnamon). The ingredients were mixed in an industrial mixer (Bathammex, Mexico City, Mexico) for five minutes obtaining six different mixtures as shown in Table 1. The proportion of each ingredient in each mixture was determined considering fat and crude fiber contents around 5.6% and 2.5%, respectively; this according to the literature [2,20] to obtain extruded products with acceptable physicochemical and sensory characteristics.


**Table 1.** Proportion of ingredients of the different mixtures \*.

\* Minor ingredients: sucralose (1.6%), cocoa (2.5%), cinnamon (0.6%).

#### 2.3.2. Extrusion Process

For the extrusion process, we used a twin-screw corotating extruder (BCTM-30, Bühler, AG, Uzwil, Switzerland) with a 600 mm length, a length to diameter ratio (L/D) of 20:1, a die opening of 4 mm, the screw configuration was selected specifically to create high levels of shear. The mixtures were fed to the extruder at a rate of 7.5 kg/h and were processed at a speed of 272 rpm at moisture content of 0.22 kg water/kg dry matter, which was adjusted within the extruder, at a temperature of 150 ◦C. It was controlled at the final stage of the extruding chamber by using a TT-137N water heater (Tool-temp, Sulgen, Switzerland). All extrudates were dried at 120 ◦C for 15 min in an air convection oven (Electrolux 10 GN/1, Stockholm, Sweden) at air cross-flow velocity of 1.5 m·s-1 until the extrudates reached a range moisture level of 0.017–0.031 kg H2O·kg ss<sup>−</sup>1. The extrudates were packed and stored at room temperature (25 ◦C) until evaluation.

#### *2.4. Analytical Methods*

#### 2.4.1. Proximate Analyses

The starting materials and extruded products were analyzed for moisture, protein, fat, crude fiber, and ash content according to methods 950.02, 960.52, 920.39, 962.09, and 923.03 of AOAC [19], respectively. Carbohydrates mass was calculated by difference. The analyses were carried out in triplicate for each treatment, and the results were expressed in g/100 g.

#### 2.4.2. Insoluble and Soluble Dietary Fiber

The insoluble and soluble fiber in the ingredients, the mixtures and the extruded products were determined with the total dietary fiber assay kit (Sigma-Aldrich, St. Louis, MO, USA) according to method 991.43 of AOAC [19]. The analyses were carried out in triplicate for each treatment, and the results were expressed in g/100 g.

#### *2.5. Functional Properties of the Extruded Products*

#### 2.5.1. Water Absorption and Water Solubility Indexes

The water absorption index (WAI) and water solubility index (WSI) were determined in triplicate following the procedures described by Anderson et al. [21]. The methods measure the quantity of water incorporated in the flour and the soluble solids that dissolve in water at 30 ◦C. Samples were weighed (2.5 g) into plastic tubes and mixed with 30 mL of distilled water. The samples were manually shaken, the slurries were centrifuged for 10 min at 3200× *g* (Thermo IEC model CL3-R, Thermo Scientific, Waltham, MA, USA), and the supernatant was decanted into pre-weighed porcelain capsules. Capsules were dried for 24 h at 105 ◦C and weighed. The gel remaining in the tubes after decanting the supernatant was weighed. The ratio between gel-forming solids and soluble solids was measured as grams of water per gram of flour. The WAI was calculated as a percentage of remaining gel weight compared to the pre-dried weight from the extruded products. The WSI was calculated as a percentage of the dried supernatant weight compared to the pre-dried weight from the extruded products.

#### 2.5.2. Bulk Density

The bulk density (BD) was determined according to Jin et al. [22] in which the ground extrudate (40/60 mesh) was poured into a cylindrical container. Excess extrudate was scraped off, and the net weight of the powder was divided by the volume of the container. Bulk density was expressed in kilograms per liter (kg L<sup>−</sup>1). The analysis was performed in triplicate, and mean values were reported.

#### 2.5.3. Expansion Index

The expansion index (EI) was reported as the ratio of extruded product diameter and the diameter of the die hole [23]. Values were reported as means of 60 measurements.

#### *2.6. Physical Properties of the Extruded Products*

#### 2.6.1. Textural Measurement: Hardness and Crispness

The evaluation of the hardness and crispness of the extrudates was performed according to the method described by Ding et al. [24], and carried out using a Texture Analyzer TA.XT (Texture Technologies Corporation, Scarsdale, New York/Stable Micro Systems, Haslemere, Surrey, UK) configured with a 2 mm punch at a crosshead speed of 5 mm/s and a travel distance of 15 mm. Twenty-four extruded unit samples were taken randomly from each treatment and analyzed. A force time curve was recorded and analyzed by the Texture Exponent 32 (Surrey, UK) program to calculate the maximum force (N) to determine the hardness and the area under the curve (N/mm) to determine the crispness.

#### 2.6.2. Pasting Properties of the Extruded Products

The amylographic viscosity profile was determined according to Sánchez-Madrigal et al. [25], with some modifications, using a Rapid Visco Analyzer (RVA SUPER 4 (Newport Scientific, Sydney, Australia). Flour sample suspensions were prepared by weighing 4 g of milled and dried (50 ± 2 ◦C, 12 h) extrudates with a 7.5 to 8.5% moisture content and a small particle size (0.25 mm) into an RVA canister and individually adjusting each sample to a total weight of 28 g using distilled water. The rotating paddles were held at 50 ◦C for 1 min to stabilize the temperature and ensure

uniform dispersion and heated to 92 ◦C at a rate of 5.6 ◦C/min, which was held constant for 5 min. The dispersion was cooled to 50 ◦C at the same rate and was held at 50 ◦C for 1 min. The maximum viscosity (MaxV) at 92 ◦C, the minimum viscosity (MinV or lowest viscosity at the end of heating constant period at 92 ◦C) and the final viscosity (FinV attained during cooling to or holding at 50 ◦C) were recorded. The total setback viscosity or viscosity of retrogradation (final viscosity minus minimum viscosity) was calculated from these parameter values. The viscosity with RVA was obtained in RVU units (1 RVU = 10 centipoises). Each treatment was performed twice.

#### *2.7. Scanning Electron Microscopy*

This analysis was performed according to the method described by Sánchez-Madrigal et al. [25]. Flours of each extruded cereal with a particle size <0.15 mm and a moisture content of 1% were stuck to stubs and coated with a gold layer in a high vacuum using a Denton vacuum evaporator (Desk II), set to a pressure of 7.031 <sup>×</sup> 10−<sup>2</sup> kg cm−2. The samples were examined using a scanning electron microscope (JSM-5800LV, JEOL, Akishima, Japan) equipped with a secondary electron detector at an acceleration rate of 10 kV.

#### *2.8. Statistical Methods*

A univariate analysis of variance was performed adjusting a model that included the main effects and their interaction (Minitab 16). When the effect of the interaction factor or the main effects was significant (α 0.05); means comparison was performed by Tukey's test [26].

#### **3. Results and Discussion**

#### *3.1. Raw Materials Characterization*

Proximate analysis showed a significant difference (*p* < 0.05) between the raw materials for each of the components (Table 2) and indicated that they were of high nutritional value. Is important to highlight that flaxseed had the highest protein, fat, and fiber content, with the lowest carbohydrate content. Whereas the amaranth had protein content twice as maize grits. These values (%) are consistent with those reported in the literature [16,27], where amaranth contains a good balance of amino acids, including lysine, an essential amino acid that is not found in most cereals [14]. Flaxseed has been reported to be low in carbohydrates (sugars and starches), high in quality protein, fiber and rich in polyunsaturated fatty acids [16]. Maize grits were the main source of carbohydrates, as shown in Table 2. Meanwhile amaranth and flaxseed showed the highest dietary fiber contents (Table 3), these results agree with Morris [16] and Cervantes [27]. The soluble fiber content in the ingredients was in the following relevance order: flaxseed (9%), amaranth (1.3%) and maize grits (0.71%).


**Table 2.** Proximate composition of the raw materials and the extruded products \*.

Means ± standard error (SE). Means by files for raw materials and extruded products, with different letters show significant difference, contrast test (*<sup>p</sup>* < 0.05). Carbohydratescalculated by difference.

 were

**Table 3.** Dietary fiber content of the raw materials and the mixtures without extruding \*.


\* Means ± standard error (SE). Means by files for raw materials and mixtures without extruding, with different letters show significant difference, contrast test (*<sup>p</sup>* < 0.05). SDF, soluble dietary fiber; IDF, insoluble dietary fiber; TDF, total dietary fiber.

#### *3.2. Extrudate Characterization*

Proximate analysis of different extruded products is shown in Table 2. Chemical characteristics were significantly affected (*p* < 0.05) for amaranth and flaxseed additions. The extruded products had high percentage of protein compared to other commercial extruded cereals, which typically have protein content between 5 and 8% [28]. This is due to the contributions of protein of amaranth (17.4%) and flaxseed (22.4%) (Table 2). Similar protein content for extruded amaranth with maize grits were reported [10]. The addition of amaranth could lead to a good balance of amino acids because it contains lysine, an essential amino acid, which is not found in most cereals [14]. Whereas flaxseed protein is rich in arginine, aspartic acid and glutamic acid and deficient in lysine [16].

The extrudates had a desirable crude fiber content (<2%; Table 2), with acceptable physicochemical and sensory characteristics for the consumer without a negative effect on the caloric and nutrient content, which could especially benefit young children [2]. Additionally, the fat content in all extruded products was below the minimum acceptable value of 5.6% [20] to reach desirable characteristic in extruded products. From the nutritional point of view (<5%) a low-fat product was obtained.

#### *3.3. Dietary, Insoluble and Soluble Fiber*

Table 3 shows the soluble and insoluble fiber content of the different mixtures before being subjected to the extrusion process. The addition of flaxseed significantly affected the soluble and total dietary fiber content. After the extrusion process of the mixtures, the results for the dietary fiber content showed a significant difference between treatments (*p* < 0.05), indicating that variation in the proportions of ingredients (amaranth and flaxseed) and their interaction, significantly affected the content of insoluble and soluble fiber in extruded products (Figure 1). Additionally, it was observed that the extrusion process caused an increase in the soluble fiber content and a decrease in insoluble fiber compared to the non-extruded mixtures (Table 3, Figure 1). The extruded products from the mixtures 3, 4, and 5 had the highest percentage of soluble fiber but they did not present a significant difference. This can be explained by the fact that extrudates 4 and 5 had the highest content of flaxseed (9.3% and 8.6% respectively), which is a significant source of soluble fiber, reaching up to 9% (Table 3). In turn, extrudate 3 had the highest percentage of amaranth and flaxseed (41.7%), which resulted in high soluble fiber content. The rest of the extrudates (mixtures 1, 2 and 6) had a low percentage of soluble fiber without showing significant differences among them (Figure 1) because they had low percentages of flaxseed. Various studies had shown that dietary fiber, especially soluble fiber in extrudates, increases when they are subjected to the extrusion process [29]. Additionally, other biomolecules such as starch undergo structure changes, leading to the formation of resistant starch, another possible mechanism causing the increase in fiber during the extrusion process [7]. On the other hand, total dietary fiber content could decrease due to the fact that during the extrusion process, shear stress caused by high screw speed, combined with high process temperatures causes chemical bond breakage from complex carbohydrates, releasing molecules as xylose, glucose, arabinose, oligosaccharides, and, preferentially, slightly branched arabinoxylans which are solubilized [29,30]. Similarly, this decrease in total dietary fiber content could be observed in some of our treatments.

A food product is considered high in dietary fiber when it contains >6% [8]. Therefore, the extruded products presented high fiber content (Figure 1), due to the addition of amaranth and flaxseed, which contain a high percentage of dietary fiber; 13.1 and 59.6% respectively (Table 3).

**Figure 1.** Insoluble and soluble dietary fiber content of extruded products. Means ± standard error (SE). SE insoluble fiber, 0.2; SE soluble fiber, 0.05. Means by columns and colors with different letters show significant differences based on contrast tests (*p* < 0.05).

#### *3.4. Water Absorption and Water Solubility Indexes*

The WAI in the extruded products was significantly affected (*p* < 0.05) by the addition of flaxseed and amaranth and their interaction, in the mixtures (Table 4). The extruded cereal from mixture 4 (9.3% flaxseed, 18.7% amaranth, and 67.3% maize grits) resulting in high WAI values due to its high maize grits (cornstarch) content which, during the extrusion process, undergoes pronounced changes in gelatinization properties, favoring a higher water absorption. This is consistent with the amylographic viscosity profile, where the extrudates with the highest values of viscosity have the highest values of WAI (Figure 2). On the other hand, low WAI values resulted for extrudate from mixture 3 (6.6% flaxseed, 33.1% amaranth, and 55.6% maize grits), which had the lowest maize grits (cornstarch) content but the highest proportion of high-fiber ingredients (39.7% amaranth and flaxseed). Similar results were obtained in extrudates from mixtures 2 and 6 (Table 4). Similar effects of fiber addition on extruded products were reported by Altan et al. [31] for the extrusion of barley mixtures with tomato pomace.

**Figure 2.** Amylographic viscosity profile of extruded products obtained with Rapid Visco Analyzer (RVA).

WSI is an indicator of the degradation of molecular components: an example is the amount of soluble polysaccharide of starch released after extrusion, which is a measurement of the degree of conversion of starch during extrusion [24]. Table 4 shows significant effects on the WSI of the extruded products by the addition of amaranth and flaxseed, in the mixtures. Extrudates from mixtures 2, 5, and 6, with a high fiber content (Table 4), had the highest WSI values (0.5). Whereas the extrudate from mixture 1, with a low fiber content, and the extrudate from mixture 4, with a high starch content, had the lowest WSI values (0.46 and 0.45, respectively). The increase in the fiber content caused an increase in the WSI values, which can be attributed to the rupture of the structural polysaccharides by the extrusion process [29,30]. A similar finding was reported by Ganorkar et al. [13] and Altan et al. [31].

#### *3.5. Bulk Density*

The bulk density in extruded cereals shows significant changes (*p* < 0.05) due to the addition of the ingredients in the mixtures (Table 4). The extrudate from mixture 6, with a high fat content, presented the lowest density value: this can be attributed to the fat's low-density and the oil contained in cereals, which are emulsified during extrusion due to the high pressure reached during the process. The fine drops of fat are coated by starches and proteins, leaving the fat encapsulated and causing a decrease in the density [32]. The extruded products from the mixtures 2 and 4 were high in protein and presented the highest density values. The rigid tertiary structure, high cohesiveness, high molecular weights, and structural functions in cereal proteins such as corn, can increase the density of food products [33]. Ryu et al. [23], reported that the density of an extruded product is strongly affected by water, fiber, fat, and starch content. The extruded products from the mixtures 6 and 3, with similar amaranth contents (26.4 and 33.1 g/100 g respectively), resulted in the lowest density values (4.6 and 4.7 kg L<sup>−</sup>1, respectively; Table 4). These results were consistent with those found by Ilo et al. [9], who evaluated the effect of extrusion-cooking process on the properties of extruded rice flour and amaranth blends. They observed that amaranth had an important influence on the product density, resulting in a minimum density value at amaranth content of 30 g/100 g. Another report had shown increases in the bulk density values during the extrusion of rice flour and corn fortified with flaxseed [12].


**Table 4.** Functional and physical properties of the extruded products \*.

\* Means ± standard error (SE). SE expansion index, 0.007; SE bulk density, 0.006; SE WSI, 0.007; SE WAI, 0.05; SE Hardness, 0.16; SE Crispness, 1.3; SE MaxV, 2.4; SE MinV, 2.3; SE FinV, 2.8; SE setback viscosity, 2.9. Means by columns with different letters show significant difference, contrast test (*p* < 0.05). BD, bulk density; EI, expansion index; WSI, water solubility index; WAI, water absorption index; MaxV, maximum viscosity; MinV, minimum viscosity; FinV, final viscosity.

#### *3.6. Expansion Index*

The addition of the different ingredients in their different proportions, and their interaction, significantly affected the expansion of extruded products (*p* < 0.05) (Table 4). The EI of instant-extruded products is very important since it is directly related to consumer acceptability; related typically to an inflated, lightweight and crunchy structure [34], mainly attributed to the presence of starch in the final extruded products [35]. The extrudate of the mixture 4 resulted in the highest (*p* < 0.05) EI (3.33), followed by the extruded mixture 5 (3.17) due to a higher percentage of maize grits (starch), and finally the extruded products containing less maize grits (mixtures 1, 2 and 6) did not have significant difference between them and they presented the lowest EI values (Table 4). It is important to note that the mixtures 4 and 5 (with higher expansion values), contained a higher starch amount and a lower percentage of the ingredients with high dietary fiber content: amaranth and flaxseed. Mixtures 1, 2, and 6 with higher amaranth and flaxseed content presented the lowest EI, due to their high dietary fiber content, which affects the expansion of extruded products. The effect of the fiber on the expansion of the extruded products depends mainly on its interactions with the starch and, therefore, on the type and amount of fiber. Insoluble fiber significantly reduces expansion volumes and increases the density of extruded products. Conversely, soluble fiber leads to better expansion volumes, unaffecting the bulk density of the extruded products compared to the insoluble fiber components [3]. The difference in expansion behavior between soluble and insoluble fiber can be explained by their interactions with starch, differences in water absorption and plasticizing behavior, but also by the physicochemical transformations they undergo during extrusion [3]. This is consistent with the results reported by Altan et al. [31], who made extruded barley using tomato pomace as fiber source, noting that additions of tomato pomace, provoked a decrease of the EI on the final products. Similar results were reported by other authors [35–37].

#### *3.7. Textural Measurement (Hardness and Crispness)*

An important quality parameter of ready-to-eat extrudates is texture. Table 4 shows the values of hardness and crispness of extruded products made from the different mixtures of ingredients. The crispness of the extruded products was not significantly different (*p* > 0.05) among the various mixtures. Similar findings have been reported for corn extruded with amaranth, where the amaranth content from 20 to 35% had no substantial effect on crispiness [11]; this amaranth percentage was close to the one used in this study (18.7–33.1%).

However, the hardness of the cereals was significantly affected (*p* < 0.05) by the added ingredients (flaxseed and amaranth) and their interaction. Extruded products from mixture 1and 5 had the highest hardness (*p* < 0.05) attributed to their high dietary fiber content. As was described above, the addition of dietary fiber leads to reduced expansion volumes and increases in density of the extruded products, inducing harder textures and less crispiness [3]. This result is consistent with the results reported by Ganorkar and Jain [12], who showed that an increase in added flaxseed caused an increase in the hardness of the extruded products. A similar finding was reported by Brennan et al. [5] where they showed that increases in the wheat bran content up to 15% in extruded breakfast cereals causes breaking force increases. In contrast, it has also been reported that soluble fibers, such as inulin, deliver a more favorable texture compared to insoluble fibers, such as bran fiber [6]. This was corroborated by Brennan et al. [5], who observed a slight hardness change when adding either inulin or guar gum to extruded corn flour. This can be corroborated by our results, where the extruded products from mixture 3 and 4 (Figure 1) contained the highest percentage of soluble fiber and the lowest hardness values (4.7 N; Table 4).

#### *3.8. Pasting Properties of the Extruded Products*

The addition of amaranth and flaxseed to the mixtures significantly affected (*p* < 0.05) the amylographic viscosity profile (RVA) of the extruded products (Table 4). MaxV is the peak viscosity where the highest degree of starch gelatinization occurs. The MaxV values obtained for each of the extruded products were very low, due the damage in the starch granules during the extrusion process [38]; this led to a notorious decrease of viscosity values in the extruded products, as shown in Figure 2. A similar trend on other viscosity parameters (MinV, FinV, and setback viscosity) values is shown in Table 4. The mixture 4, with the highest starch content (67.3% maize grits), had the highest MaxV value (*p* < 0.05); whereas mixture 3, with the lowest percentage of starch (55.6% maize grits), had the lowest value (*p* < 0.05). On the other hand, it is possible that the formation of complex structures during extrusion-cooking through interactions between starch-lipid complexes and/or starch-protein walls prevents adequate gelatinization of the starch [39,40]. Another factor influencing the pasting properties of extruded products is the presence of dietary fiber, which leads to a decrease in the fraction of water-swelling starch, due to its replacement by the fiber [37]. All these factors limited a complete starch gelatinization, causing a decrease in viscosity. Similar results were observed in

a study where increases in amaranth in rice-amaranth blends generally decreases the viscosity of extruded products [9]. Similar findings were reported for extruded cornstarch blends with whey protein concentrate and Agave tequilana fiber [37].

Additionally, low setback values were found for all extruded products, indicating low rates of starch retrogradation and syneresis. During cooling, reassociation of starch molecules, especially amylose, result in viscosity increase favoring the final viscosity. This phase is commonly described as the setback region during which retrogradation and reordering of starch molecules occurs [41].

#### *3.9. Scanning Electron Microscopy*

The scanning electron micrographs revealed the impact of the different ingredients (Figure 3). After extrusion, it can be observed that combination of shear force and temperature inside the barrel caused microstructural changes in the extrudates of the different treatments [42]. The microstructural analysis shows that the addition of amaranth and flaxseed, increased the fiber content in the mixtures, resulting in compact agglomerates, increased crystallinity, and larger, more compact laminar structures (Figure 3c–f). This can be attributed to high protein (12%) and fiber (9–13%) content in the extrudates. Fiber tends to rupture cell walls and promotes breakage of air cells during extrusion, which prevents matrices from expanding [43], resulting in harder textures, higher densities and more compact structures as shown in the micrographs (Figure 3c–f). Similar results were found by Zhang et al. [29] and Cueto et al. [42].

**Figure 3.** Micrographs of the extruded products from: (**a**) Mixture 1, (**b**) Mixture 2, (**c**) Mixture 3, (**d**) Mixture 4, (**e**) Mixture 5, (**f**) Mixture 6.

#### **4. Conclusions**

This research shows that different levels of amaranth and flaxseed in the development of extruded products had a significant impact on their functional and physicochemical properties. The extruded products obtained had high protein content (>12%), which is higher than in the commercial breakfast cereals. Besides these characteristics, the obtained extruded products presented a healthy fat content (<5%) and a high content of soluble and insoluble dietary fiber.

Another important ingredient was the maize grits, which was the source of starch, and it served as a basis to produce some expansion level in the extrudates. Extruded products with low levels of starch (maize grits) and high levels of fiber in the mixtures resulted in extrudates with low EI and high hardness values. These results suggest that the extrusion-cooking process of high-fiber flours such as flaxseed (8.6–9.3%) and amaranth (18.7–22.9%) in mixture with maize grits (63.8–67.3%) and minor ingredients results in an extruded product of good nutritional quality, with suitable functional and physicochemical characteristics.

**Author Contributions:** Conceptualization, J.L.T.-E. and A.Q.-R.; investigation, J.L.T.-E.; writing—original draft, J.L.T.-E. and A.Q.-R.; methodology, J.G.B.-G., C.A.A.-G. and F.M.-B.; writing—review and editing, E.P.-C., M.A.N.-G., C.O.M.-P.; performed the analysis, data curation, J.A.O.-G.

**Funding:** This research received no external funding.

**Acknowledgments:** The authors acknowledge the Universidad Autónoma de Nuevo León and the Universidad Autónoma de Chihuahua for supporting this investigation. This paper is based on the postgraduate studies of Jazmin Leticia Tobias Espinoza, who was supported by a PhD scholarship from the National Council of Science and Technology of Mexico (CONACYT).

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

#### *Article*

### *Brosimum alicastrum* **Sw. (Ramón): An Alternative to Improve the Nutritional Properties and Functional Potential of the Wheat Flour Tortilla**

**Rodrigo Subiria-Cueto 1, Alfonso Larqué-Saavedra 2, María L. Reyes-Vega 3, Laura A. de la Rosa 1, Laura E. Santana-Contreras 1, Marcela Gaytán-Martínez 3, Alma A. Vázquez-Flores 1, Joaquín Rodrigo-García 1, Alba Y. Corral-Avitia 1, José A. Núñez-Gastélum <sup>1</sup> and Nina R. Martínez-Ruiz 1,\***


Received: 23 October 2019; Accepted: 21 November 2019; Published: 24 November 2019 -

**Abstract:** The wheat flour tortilla (WFT) is a Mexican food product widely consumed in the world, despite lacking fiber and micronutrients. Ramón seed flour (RSF) is an underutilized natural resource rich in fiber, minerals and bioactive compounds that can be used to improve properties of starchy foods, such as WFT. The study evaluated the impact of partial replacement of wheat flour with RSF on the physicochemical, sensory, rheological and nutritional properties and antioxidant capacity (AC) of RSF-containing flour tortilla (RFT). Results indicated that RFT (25% RSF) had higher dietary fiber (4.5 times) and mineral (8.8%; potassium 42.8%, copper 33%) content than WFT. Two sensory attributes were significantly different between RTF and WFT, color intensity and rollability. RFT was soft and it was accepted by the consumer. Phenolic compounds (PC) and AC were higher in RFT (11.7 times, 33%–50%, respectively) than WFT. PC identification by ultra-performance liquid chromatography quadrupole time of flight mass spectrometry (UPLC-QTOF-MS) showed that phenolic acids esterified with quinic acid, such as chlorogenic and other caffeoyl and coumaroyl derivatives were the major PC identified in RSF, resveratrol was also detected. These results show that RSF can be used as an ingredient to improve nutritional and antioxidant properties of traditional foods, such as the WFT.

**Keywords:** nutritional value; antioxidant capacity; phenolic compounds; sensory properties; functional foods

#### **1. Introduction**

Tortilla is an iconic food in the Mexican and Central American diet. Mexico is the main tortilla consumer around the world (75 kg/person) [1], with 11.5 million tons consumption per year [2]. The growing demand of tortilla has caused its globalization, representing nowadays a food of importance in countries such as USA or China. Tortilla is usually made from wheat or corn [3]. On the one hand, the corn tortilla provides proteins, calcium, potassium and carbohydrates, which makes this food an important vehicle to provide nutrients to people with malnutrition. On the other hand, the

wheat flour tortilla (WFT) provides more calories (47%), proteins, lipids and carbohydrates and less dietary fiber and micronutrients than corn tortilla [4], which makes this type of tortilla a hypercaloric and a poorly nutritious food. However, WFT has a wide demand in northern Mexico and USA and it is highly consumed by children, youth and adults from ethnic and low-income populations contributing to exacerbate states of undernourishment [5] or obesity [6]. Malnutrition is one of the main public health problems affecting many countries [7] and is associated, in many cases, with the high intake of hypercaloric foods with low dietary fiber, minerals and vitamins [6]. Therefore, it is of interest to make culturally rooted foods, such as tortilla, with greater nutritional value in order to contribute to a healthier diet of the population; particularly for groups in poverty and/or malnutrition.

*Brosimum alicastrum* Sw. (ramón) is a tree of the Mexican tropics whose fruit and seed have high nutritional value. This tree was appreciated by the Mayan culture from 300 to 900 A.D. [8]. The flour obtained from the seed (RSF) is characterized by high protein, dietary fiber and micronutrient content. *Brosimum alicastrum* Sw. is considered as an underexploited natural resource in Mexico with potential nutritional and functional properties [9], which can be incorporated into foods with little nutritional value. In this context, the aim of this study was to evaluate the impact of partial replacement of wheat flour with RSF on the physicochemical, sensory, rheological and nutritional properties and antioxidant capacity of RSF-containing flour tortilla (RFT). The hypothesis proposed was that the partial replacement of WF with RSF in the tortilla improves its nutritional contribution in dietary fiber and micronutrient content, increases its antioxidant capacity and the RFT is sensory accepted by the consumer compared as well as a tortilla made with 100% WF.

#### **2. Materials and Methods**

#### *2.1. Materials*

The ramón seed flour (RSF) (*B. alicastrum* Sw.) was provided by CICY (Herbarium Roger Orellana, Centro de Investigación Científica de Yucatán A.C.). The seeds were collected from growing ramón trees at rancho Xoccheila (20◦33 N; 89◦34 W), municipality of Sacalum, Yucatán. The seeds were dried in the sun, the testa was removed and the seed was ground to a fine flour. The wheat flour (WF), dry (salt, baking powder) and moist (shortening) ingredients were purchased in the local markets of Ciudad Juárez, Chihuahua, México.

#### *2.2. Tortilla Development*

Different formulations were proposed partially replacing the content of WF for RSF (20%, 25%, 28%, 30% and 40%). The ramón flour tortilla (RFT) and the wheat flour tortilla (WFT) were developed following the previously reported method with some modifications [10]. Extra water was added to the RSF doughs until a more elastic texture was obtained. Microbiological quality in total coliforms bacteria (CC), aerobics mesophilic bacteria (AC), yeast and mold (YM) was determined in tortillas samples according to the plate count method (3M™, Petrifilm™, Minneapolis, MN, USA). Briefly, tortilla samples were diluted 1:10 in saline solution (0.9%), homogenized and one mL plated onto Petrifilm. Plates were incubated at 35 ± 1 ◦C for 24 h for CC, 48 h for AC and 25 ± 1 ◦C for YM. Official Mexican regulation was observed for the limits of CC, AC and YM in tortillas [11].

#### *2.3. Proximate Composition, pH, Activity Water (Aw) and Titratable Acidity*

All tortilla samples were homogenized using a commercial blender (Nutribullet®). The samples were analyzed in triplicate following the AOAC methods [12]: ash was determined in muffle furnace (Felisa®, Model FE-340, Jalisco, México) at 550 ◦C for 5 h; crude protein by Kjeldahl method (Labconco®, Model RapidStill II, Kansas city, MO, USA); fat by Soxhlet method (Soxtec™, Model 2043, Foss™, Hilleroed, Denmark); total carbohydrates by difference method; crude fiber by gravimetric method, dietary fiber by enzymatic-gravimetric assay, water activity in AQUA LAB® (Model Serie 3, Meter Food, Washington, D.C., USA) equipment; pH and titratable acidity by potentiometric method (Accumet®, Model AB15 Plus, Westford, MA, USA). Moisture analysis was performed by the AOAC method [12] with the following modifications: it was determined in an oven (VWR®, Model 1324, Irving, TX, USA) at 105 ◦C for 8 h.

#### *2.4. Extraction and Quantification of Minerals*

The mineral content (Cu, Zn, K, Fe and Na) was obtained from the ashes of the samples following the method by the AOAC with minor modifications. Flours and tortilla samples (5 g) were calcined in a muffle furnace (Felisa®, Model FE-340, Guadalajara, Jalisco, México) for 8 h at 550 ◦C. Subsequently, 3 mL of HNO3 (SCP®, Quebec, Canada) (0.2%, *v*/*v*) was added and taken to dryness on a hot plate (100 ◦C). The samples were calcined again for 2 h at 550 ◦C until white ashes were obtained, 5 mL of 6 M HCl (JT Baker®, Fisher Scientific, West Palm Beach, Florida, USA) were added and dried. Finally, 20 mL of HNO3 (0.2%, *v*/*v*) (Merck®, Toluca, Estado de México, Mexico) were added and samples were transferred into plastic conical tubes (Corning®, Merck, Toluca, Estado de México, México). The samples were analyzed using atomic absorption spectroscopy (Perkin Elmer®, Model AAnalyst 200, Madrid, Spain) with acetylene flame adjusting the wavelength for each mineral [12].

#### *2.5. Extraction and Quantification of Carotenoids*

The content of carotenoids was determined following the method by Moreno-Escamilla et al. [13] with some modifications. Flours and tortillas were dried at 45 ◦C in a vacuum oven (Shel Lab®, Model VWR A-143, Tualatin, OR, USA) at 20 mm Hg for 36 h. Next, they were ground using a commercial blender (Nutribullet®) and kept at <sup>−</sup><sup>18</sup> ◦C in darkness for no more than 48 h. Subsequently, 0.3 g of dry ground sample were mixed with 10 mL of acetone (JT Baker®, Fisher Scientific, West Palm Beach, FL, USA), sonicated (Branson®, Model 5800, Fisher Scientific, West Palm Beach, Florida, USA) for 20 min and centrifuged (Eppendorf®, Model 5810 R, Alto da lapa, Sâo Paulo, Brazil) at 3500 rpm for 10 min. The supernatant was recovered, and the residue was extracted two more times under the same conditions. The absorbance of the combined supernatants was read in an ultraviolet–visible (UV–Vis) microplate reader (BioRad®, Model XMark, Ciudad de México, México) at a wavelength of 474 nm. Results were reported as milligrams of β-carotenoids per 100 g of sample.

#### *2.6. Extraction and Quantification of Ascorbic Acid*

Ascorbic acid determination was realized according to the technique described by Alvarez-Parrilla et al. [14], with some modifications. Extracts were obtained by mixing 0.2 g of flour or tortilla samples (DW) with 5 mL of 5% metaphosphoric acid (Merck®, Toluca, Estado de México, México), the mixture was sonicated (Branson®, Model 5800 Fisher Scientific, West Palm Beach, FL, USA) for 20 min in dark conditions, and centrifuged (Eppendorf®, Model 5810 R, Alto da lapa, Sâo Paulo, Brazil) at 3500 rpm for 10 min. For quantification, 300 μL of each supernatant was taken and mixed with 200 μL of 6.65% tricloriacetic acid (Merck®, Toluca, Estado de México, México) and 75 μL of the DNPH (2,4-dinitrophenylhydrazine) reagent (Merck®, Toluca, Estado de México, México) in 100 mL of 5 M H2SO4 (JT Baker®, West Palm Beach, Fisher Scientific, FL, USA), then incubated for 3 h at 37 ◦C and 0.5 mL of H2SO4 (JT Baker®, Fisher Scientific, West Palm Beach, FL, USA) (65%, *v*/*v*). Absorbance was measured in the UV-Vis microplate reader (BioRad®, Model xMark, Ciudad de México, México) at 520 nm, using ascorbic acid as standard. Results were reported as milligrams of ascorbic acid per 100 g of sample.

#### *2.7. Sensory Characterization*

Tortillas (RFT and WFT) were sensory characterized with a descriptive analysis by a trained panel of 8 judges. The attributes in the olfactory phase were odor intensity and descriptors determined by focus group technique. In the oral phase, mouthfeel characteristics were evaluated: such as cohesiveness, hardness, moistness, adhesiveness and astringency; taste: such as sour, salty and bitter. Color and texture attributes, such as elasticity, firmness and rollability were also evaluated. All tests

were conducted in individual booths and the judges used a 150 mm linear scale, labeled at the end as "Not all ... " and "Extremely ... " for each attribute or descriptor. Each judge was provided with slices of tortilla (10 g), previously heated in a microwave for 30 s, and they were placed in 2 oz plastic cups, identified with three-digit random numbers. The samples were served to the judges in a balanced and randomized form, together with evaluation sheets. Judges rinsed their mouths with purified water (Alaska®, Chihuahua, Mexico) at the beginning and between samples for the oral phase and they used eye covers in all tests, except in the visual phase. Pantone® scale was used in color test. Two attributes or descriptors were evaluated per session of 60 min, standards for each attribute or descriptor were used at the beginning of the test and each test was performed by duplicate [15,16].

#### *2.8. Consumer Acceptance Test*

An acceptance test was carried out to evaluate the consumer degree of liking for RFT and WFT. The test was performed in 120 consumers using a 9-point hedonic scale, ranging from "Like extremely" to "Dislike extremely". Participants were given 5 g of RFT or WFT (freshly made) and kept warm in thermal containers (35 ± 2 ◦C) for a limited time of 15 min. Two samples were evaluated by session and each sample was presented monadically in disposable dishes (12 cm diameter), labeled with three-digit random numbers. Participants rinsed their mouths with purified water (Alaska®, Chihuahua, Mexico) at the beginning of the session and between samples. They tasted each sample and indicated on the hedonic scale the degree of liking for the sample [16].

#### *2.9. Rheological Measurements*

For the rheological characterization, texture profile analysis (TPA) and cohesiveness tests were performed in the RFT and WFT dough samples using the methods described by Reyes-Vega et al. [17] and Flores-Farías et al. [18] with modifications. Spherical dough fractions (2.0 cm diameter) were tested to obtain gumminess (N), hardness (N), adhesiveness (J), elasticity (mm) and chewiness (Nm). Cohesiveness was performed in dough samples (20 g) placed in a cylindrical container and a penetration speed of 2.0 cm·s−<sup>1</sup> was applied. In the tortilla samples (2 <sup>×</sup> 6 cm), cutting, elongation and extensibility tests were made to obtain the force (N) required to separate it, the elongation (mm), the distance that can be stretched before breaking (mm), work necessary for extension (J) and cohesiveness (N) following the methods by Kelekei et al. [19] and Suhendro et al. [20]. All measurements were carried out in a texture analyzer (Lloyd Instruments™-Model TAPlus AMETEK, Elancourt, France), adjusting different probes for each test.

#### *2.10. Content of Total Phenolic Compounds and Flavonoids*

Extraction of phenolic compounds was performed following the methodology established by Alvarez-Parrilla et al. [14] with modifications. Flour and tortilla samples were dried, ground and stored as described for the extraction and quantification of carotenoids. Next, 5 g of the ground samples were sonicated for 30 min with 10 mL of methanol (JT Baker®, Fisher Scientific, West Palm Beach, FL, USA) (80%, *v*/*v*), centrifuged at 3500 rpm for 15 min and the supernatant was collected by filtration, adjusted to a volume of 25 mL, and kept in refrigeration at 4 ◦C for less than 12 h, until analysis. Total phenolic content (TPC) was determined by the Folin–Ciocalteu method and total flavonoids (TF) were determined by the AlCl3 method, according to the methodology described by de la Rosa et al. [21]. Results were expressed as milligrams of gallic acid equivalents (GAE) and milligrams of catechin equivalents (CE) per 100 g of sample (DW), respectively.

#### *2.11. Antioxidant Capacity*

The FRAP (ferric ion reducing antioxidant power), DPPH (2,2-diphenyl-1-picryl-hydrazyl-hydrate free radical) and ABTS (2,2'-azinobis-(3-ethylbenzothiazoline-6-sulfonate) assays were used to quantify the antioxidant capacity of the methanolic extracts of flour and tortillas samples, according to the methodology described by Alvarez-Parrilla et al. [14], de la Rosa et al. [21] and Brand-Williams et al. [22].

The extracts were obtained as described for the content of total phenolic compounds and flavonoids. For the FRAP assay, the FRAP reagent was prepared by mixing 0.3 M acetate buffer (Hycel®, Zapopan, Jalisco, México) with 10 mM TPTZ (2,4,6-tripyridyl-s-triazine; Acros Organics®, Morris Plains, NJ, USA) dissolved in 40 mM HCl (Hycel®, Zapopan, Jalisco, México), and 20 mM FeCl3 (Hycel®, Zapopan, Jalisco, México); in a ratio 10:1:1, v / v / v. The FRAP reagent was heated at 37 ◦C for 30 min and the assay was performed by mixing 180 μL of the FRAP reagent with 24 μL of sample in microplate wells. Absorption was measured at 595 nm every 60 s for 30 min in the UV–Vis microplate reader. The results were reported in millimole Trolox equivalent/g dry weight sample.

The DPPH assay was performed by mixing 50 μL of sample with 200 μL of the DPPH radical (190 μM in methanol; Merck®, Toluca, Estado de México, México) in microplate wells and absorbance was read at 515 nm for 10 min in the UV–Vis microplate reader.

For the ABTS assay, ABTS radical cation was prepared by diluting ABTS salt (7 mM; Merck®, Toluca, Estado de México, México) and K2S2O8 (2.45 mM; Merck®, Toluca, Estado de México, México) in phosphate buffered saline (PBS, pH 7.4, 0.15 M KCl; Merck®, Toluca, Estado de México, México), and the solution was incubated in refrigeration for 16 h. Then 12 μL of the sample was mixed with 285 μL of the ABTS radical cation in microplate wells and the absorbance was read at 734 nm for 30 min in the UV–Vis microplate reader. Results of DPPH and ABTS assays were reported as inhibition percentage.

#### *2.12. Identification of Individual Phenolic Compounds in Ramón Seed Flour (RSF) by Ultra-Performance Liquid Chromatography Quadrupole Time of Flight Mass Spectrometry (UPLC-QTOF-MS)*

Phenolic extracts of RSF, were cleaned up by solid phase extraction (SPE) using a C18 column (SupelCo, Merck®, Darmstadt, Germany) to reduce the presence of sugars and small organic acids. Briefly, 6 mL (1 volume) of extract was washed with 2 volumes of water (fractions discarded) and phenolic compounds were eluted with 2 volumes of pure methanol. Solvent was eliminated in a rotary evaporator and the dried extract re-dissolved in methanol high-performance liquid chromatography (HPLC) grade (2 mg/mL), passed through a nylon filter (0.45 μm) and applied into the chromatography system. Separation and identification of phenolic compounds was carried out by using an Infinity II LC System equipped with a photodiode array detector with a binary solvent pump and autosampler (Agilent Technologies, California, USA). Separation of individual phenolic compounds was carried out using a rapid-resolution high-definition (RRHD) reverse-phase C18 column (2.1 × 50 mm; 1.8 particle; ZORBAX Eclipse Plus®, Agilent, California, USA) at 25 ◦C, with a pre-column cartridge. The samples (1 μL) were injected and elution of phenolic compounds was completed in 12 min with a linear gradient and constant flow rate of 0.4 mL/min, as described before by Torres-Aguirre et al. [23]. The mobile phase consisted of solvent A (formic acid, 0.1% *v*/*v*, from Tedia®, Fairfield, Ohio, USA) and solvent B (acetonitrile from Tedia®, Fairfield, Ohio, USA). The linear gradients were as follows: 0–4 min, 90% A, 4–6 min, 70% A, 6–8 min, 62% A, 8–8.5 min, 40% A, 8.5–9.5 min, 90% A. Elution of phenolic compounds was detected at 255, 275 and 320 nm.

The LC equipment was coupled to a quadrupole time of flight (Q-TOF) mass spectrometer with electrospray ionization (ESI) source. The mass spectrometer was operated in negative mode and specific conditions as follows: capillary voltage of 4500 V, gas nebulizer pression 30 psi, dry gas (nitrogen) temperature of 340 ◦C and flow at 13 L/min. Mass range was monitored from 100 to 3000 m/z. Phenolic compounds were identified by comparing the accurate mass and isotopic distribution of their molecular ions [M − H]−, and in some cases their retention times with those of commercial standards, and compounds listed in a specialized database, using a find by database algorithm in the MassHunter Workstation qualitative analysis, version B. 07.00 (Agilent Technologies, Santa Clara, CA, USA).

#### *2.13. Statistical Analysis*

Data from microbiological analysis were compared with the permissible limits established by the Mexican regulation [9]. Physicochemical parameters, minerals, vitamins and phytochemical data were analyzed by Student *t*-test. Data from sensory descriptive analysis were analyzed using a repeated measure analysis of variance (ANOVA) and Fisher's multiple comparisons (LSD) and data from acceptance test were analyzed Chi square test. All the analyses were carried out with XLSTAT program, version 2016.05 (Addinsoft®, Paris, France). The results are presented in mean values <sup>±</sup> standard deviation (SD). The criterion for statistical significance was *p* < 0.05.

#### **3. Results and Discussion**

#### *3.1. Tortilla Preparation and Food Safety*

The preliminary sensory acceptance tests were considered for the selection of the RFT formulation, establishing the addition of 25% RSF for the tortilla. The lack of gluten in RSF influenced the properties of elasticity and firmness imparted by prolamines (gliadin) and glutelins (glutenin) [24]. Therefore, slight changes were made in the steps to prepare the tortilla (RSF), in order to obtain better physical characteristics. First, the kneading time was increased to 20 min and more water was added until a flexible and firm dough was obtained. In this way, higher hydration was achieved, breaking the endosperm protein bodies and promoting covalent and non-covalent interactions between larger polypeptides [25]. Second, water incorporation increased the elasticity and adhesiveness in the dough, so it was necessary to change the use of the rolling pin for a manual tortilla press, which is a simple, traditional and low-cost procedure. Third, salt content was increased by 25% in order to increase the hydration property of the proteins in RSF. Also, sodium facilitates denaturalization of proteins exposing their hydrophilic groups and increasing their degree of aggregation. This effect is generated by the increase of ionic force producing a reduction in protein solubility [26]. Finally, the cooking time was adjusted for RFT (26 s per side) according to the other changes made in the process, in order to avoid an impact on the consistency and rheology of the tortilla.

Microbiological analysis showed that both tortillas (RFT and WFT) were within the permissible limits, according to the Mexican legislation, in colony forming units of aerobic mesophilic bacteria (AC) (8 <sup>×</sup> 101 and 1 <sup>×</sup> 101 CFU/g, respectively. Permissible limit: 10,000 CFU/g), total coliforms bacteria (CC) (<1 <sup>×</sup> 101 CFU/ g in each sample. Permissible limit: <30 CFU/g) and yeasts and molds (YM) (<10 <sup>×</sup> 101 CFU /g. Permissible limit: 300 CFU/g) [11]. These microorganisms are indicators of management conditions or efficiency of the food preparation. They timely warn of inadequate handling or contamination that increases the risk for the presence of pathogenic microorganisms in the product [27].

#### *3.2. Physicochemical and Micronutrient Characterization*

Chemical and micronutrient composition of flours and tortillas is presented in Table 1. In flours, the contents of protein, ash, crude, and dietary fiber were higher in RSF than in commercial WF. In minerals, RSF was 2.5 times higher in copper (*p* = 0.01), 8 times higher in potassium (*p* < 0.01) and 2.3 times higher in sodium (*p* < 0.01) than WF. RSF was equal to WF in iron content (*p* = 0.13) and 6 times lower in zinc content (*p* < 0.01). RSF showed higher acidity and Vitamin C content (3.8 times) than WF, but carotenoid content was equal in both flours.

In tortilla, RFT retained more moisture (2.9%) than WFT, reflecting the extra water added in the wetting process, but the water activity was the same in both samples. Protein content was equal in RFT and WFT, showing RFT protein content was not affected by the partial substitution with RSF. RFT had an increase in the mineral content, showing an increase in copper (1.5 times higher) (*p* = 0.02) and potassium (1.8 times higher) in RFT (*p* < 0.01), while the content of iron (*p* = 0.84) and zinc (*p* = 0.81) remained similar. Dietary fiber was 4.5 times higher in RFT (14% DRV/100 g product) than in WFT (3% DRV/100 g product).


**Table 1.** Physicochemical characteristics of flour and tortillas samples.

Mean ± SD. RSF—ramón seed flour, WF—wheat flour, RFT—ramón flour tortilla, WFT—wheat flour tortilla. \* CAE—citric acid equivalent (0.064). Comparison between flours and between tortillas. Different letters indicate significant difference (*p* < 0.05).

Little information has been published on the nutritional composition of the *Brosimum alicastrum* Sw. seed. A study carried out by Carter [8] showed that the dietary fiber content in ramón's seeds from different countries (México, Honduras and Guatemala) was from 4.91 to 21.71 g/100 g (dry weight). In this study, RSF had a dietary fiber content of 13.0 g/100 g (fresh weight) or 14.9 g /100 g (dry weight), which is within the range reported by other authors [9,28]. Considering this fiber content, and according to the FDA criteria, RSF should be considered as food ingredient rich in dietary fiber (52.1% DRV/100 g) [29]. In comparing the mineral content of the flours, it is important to consider that commercial WF is enriched with folic acid, iron, and zinc, in accordance with Mexican regulations [11], which indicates that RSF is naturally rich in iron and would not need to be enriched to meet the legal iron requirements. This is particularly important for groups in poverty, for example iron deficiency is the main cause of anemia in Mexico, and is associated with a low intake of food from animal origin and a high intake of corn, with a high content of phytates that inhibit the iron absorption [30]. On the other hand, it is the first time that vitamin C or total carotenoid content of RSF has been reported. Both values were lower than those reported in lettuce using the same analytical technique [13], so we considered RSF was not a good source of these compounds.

Substitution of 25% WF by RSF had a positive impact in the mineral and dietary fiber content of RFT, so the new formulation showed a better nutrient profile and could have a positive impact on the consumer's health. Minerals have important biological roles in the organism; for example, copper is an essential cofactor involved in the maturation of connective tissue, the synthesis of neurotransmitters and the prevention of cardiovascular diseases [31] while potassium participates in insulin secretion, creatine phosphorylation, carbohydrate metabolism, protein synthesis, nerve transmission and muscle contraction [32,33]. The potassium content in RFT was similar to that of bananas (358 mg/100 g) and higher than tomatoes (237 mg/100 g), which are considered as foods rich in this mineral [34]. The dietary fiber content in RFT (14% DRV/100 g) was higher than in corn tortilla (3.1% DRV/ 100 g product) [34], so RST can be considered as a good source of dietary fiber [29]. Guevara-Arauza et al. [35] reported a high fiber content in a tortilla added with nopal fiber (16.7%); nevertheless, this tortilla would provide 45% less protein than RFT. The importance of dietary fiber in food is well known; its consumption has shown benefits throughout the digestive process. Different physiological and prebiotic effects at the colon level make this nutrient a key component of a healthy diet [36]. Finally,

the partial replacement of 25% WF with RSF did not provide a significant increase of Vitamin C or carotenoids in RFT.

#### *3.3. Sensory Attributes and Consumer Acceptance*

The sensory characterization of both tortillas (RFT and WFT) in different attributes is shown in Figure 1. For a better interpretation of the intensity linear scale (150 mm), five intensity levels were considered: low (L, 0 to 37 mm), medium low (ML, 38 to 74 mm), medium (M, 75 mm), medium-high (MH, 76 to 112 mm) and high (H, 113 to 150 mm).

**Figure 1.** Sensory profile of tortillas. RFT—ramón flour tortilla, WFT—wheat flour tortilla. Intensity linear scale (150 mm). \* Significant difference at *p* < 0.05.

Of all the attributes evaluated in both tortillas, only two were perceived as significantly different between them: color and rollability. In color, RFT showed a MH intensity level (100.4 ± 9.4 mm), significantly different from WFT that was situated in L intensity level (32.2 ± 10.1 mm) (*p* < 0.01). Pantone® scale was used to identify the color tones in both tortillas. RFT was located between Pantone codes # C 728C and 729C, indicating a light brown color, while WFT was classified with Pantone code # C 7401C, which corresponds to a light cream color. In tactile tests, RFT presented higher rollability (MH, 96.8 ± 23.1 mm) compared to WFT, which was in ML intensity (58.6 ± 24.8 mm) (*p* < 0.01). In the case of hardness, a trend was observed and WFT was perceived to be slightly harder (ML, 55.4 ± 24.0 mm) than RFT (L, 37.5 ± 14.7 mm) (*p* = 0.09). RSF color tones may be caused by the presence of tannins which are present in the maturation stages of certain seeds that change from green to brown due to morphological changes in the vascular tissue [37]; while WFT color tones are caused by Maillard reactions [38]. The increase of salt in RFT, added to promote hydration, was not detected in the salty taste. Cohesiveness and adhesiveness, which are directly related to the viscoelastic properties that gluten provides to create a support matrix capturing water and air molecules [39], were not affected by the water increase in RFT.

In the olfactory phase, no significant differences were identified between tortillas, although RFT was perceived with a higher odor to whole wheat (MH, 90.0 ± 38.0 mm) (*p* = 0.06) and a less intense flour odor (ML, 52.8 ± 26.9 mm) (*p* = 0.07) in comparison to WFT (ML, 54.8 ± 25.9 mm and MH, 79.1 ± 27.0 mm, respectively). The smell of dough (RFT, 72.8 ± 34.5 mm and WFT 61.1 ± 36.4 mm) (*p* = 0.52) and toast (RFT, 62.8 ± 28.4 mm and WFT, 44.4 ± 27.5 mm) (*p* = 0.20) was found in a ML intensity in both tortillas (Figure 2).

**Figure 2.** Sensory odor profile of tortillas. RFT—ramón flour tortilla, WFT—wheat flour tortilla. Intensity linear scale (150 mm).

The final consumer acceptance was tested on 120 participants. The 9-point hedonic scale was divided into three areas (like, neutral and dislike). RFT and WFT were both liked by the consumer (*p* > 0.05) (Figure 3). However, 36% of consumers showed dislike for RFT and 10% for WFT (*p* < 0.01). RFT and WTF were placed in the following categories on the 9-point hedonic scale: "Like extremely" 2.5 and 4%, respectively (*p* = 0.72), "Like very much" 10.8 and 15%, (*p* = 0.44); "Like moderately" 22.5 and 40.8%, (*p* < 0.01); "Like slightly" 20.8 and 15.8%, (*p* = 0.40); "Neither like nor dislike" 7.5 and 14.2%, (*p* = 0.14) and "Dislike slightly" 20% and 5%, (*p* < 0.01). In general, 64% of consumers accepted RFT. Some consumer comments indicated that RST was milder, with balanced salt content, and whole wheat odor. These comments confirm the attributes evaluated by the judges, describing RST as softer (*p* = 0.06) and having a whole wheat odor too (*p* = 0.09).

**Figure 3.** Consumer acceptance tortillas. RFT—ramón flour tortilla, WFT—wheat flour tortilla. \* Significant difference at *p* < 0.05.

RFT had a sensory acceptance similar to corn tortilla added with soy and amaranth (60%) [40] or corn tortilla added with *Brosimum alicastrum* [41]. Innovation in products with high cultural value could be an important factor for greater consumer acceptance. Studies in this regard are necessary in the development of functional foods appealing to different populations or ethnic groups.

#### *3.4. Rheological Characterization*

Data from the TPA test showed that there was a significant difference in gumminess (*p* < 0.01), adhesiveness (*p* = 0.01), elasticity (*p* = 0.01) and chewiness (*p* < 0.01) in the tortilla doughs (Table 2). RSF addition generated a less gummy, elastic and softer dough; requiring less force during mastication compared to the WF dough or other foods, such as white bread (10.8 Nm), lettuce (9.8 Nm) and carrot (12.7 Nm) [42]. It was also less adhesive than corn tortilla dough substituted with bean and amaranth flour (−0.36 N) and nopal flour and algae (−0.32 N) [43]. These results might be related to the extra water addition that participates in the dough structural changes and in the absorption capacity of flour starch granules, which affect the adhesiveness and hardness mass viscoelastic and tension properties [44]. In the same way, it has been reported that the starch granules of RSF are similar to the granules of potato, which possess superior gelling properties and better capacity of water retention than wheat starch [45].

**Table 2.** Rheological characteristics of tortillas dough samples.


Mean ± standard deviation (SD). N—Newton, J—joul, mm—millimeter, Nm—newton for meter, RFS—ramón seed flour, WF—wheat flour RSF—ramón seed flour, WF—wheat flour. Comparison between doughs. Different letters indicate significant difference (*p* < 0.05).

The cohesiveness test was also performed on the tortilla dough, where RFT was slightly less cohesive than WFT dough. This indicates that the particles of the dough are less strongly bound in the dough with RSF [46] and therefore less effort is required to deform or break it. Therefore, it is considered that these behaviors could be produced by the primary structure of the RSF proteins, that could be low in thiol groups. In this way, there is a possibility that the lack of thiol–thiol reactions to form disulfide bridges, would make the dough collapse [47].

Cut test for the RFT required less strength and work than for the WFT (*p* < 0.01). Thereby, RSF provides more softness to the tortilla (Table 3), which means that it does not require a great effort to cut it with the incisors unlike other functional tortillas added with bean flour with amaranth (5.22 J) or nopal with algae (6.51 J) [43]. Another factor that could modify the rheology of the tortilla is the size of the starch granules of RSF (6.5 to 15 μm), which are smaller than those provided by wheat flour (11 to 41 μm), so it can be retro-degraded quicker in the cooking process [48,49]. In the multi-directional elongation and extensibility test, the same behavior was found as in the previous tests. The WFT was 10 times more resistant than RFT (*p* < 0.01) and with a similar elongation (*p* = 0.74), whereas extensibility in RFT required less force during extension (*p* < 0.01), it was less cohesive (*p* = 0.01) and it needed half the work to extend (*p* = 0.01). All these rheological characteristics are due to the interactions between globular proteins and starch molecules. The lower cohesiveness and other related rheological characteristics of RFT in comparison with WFT could be explained by considering that RSF proteins have a higher molecular weight, which would affect their extension properties generated by the release of CO2 during cooking and, hence, reduce the force of intermolecular interactions in the RFT [50,51]. A complete characterization of RSF proteins could help to better explain these behaviors

and the rheological properties of RFT. Another factor that could be responsible for the lack of resistance in RFT may be the molecular rearrangement of polysaccharides with water during the cooking process, compromising the viscoelasticity and gelling of starch granules [52].


**Table 3.** Rheological characteristics in tortilla samples.

Mean ± SD. RFT—ramón flour tortilla, WFT—wheat flour tortilla. Comparison between tortillas. Different letters indicate significant difference (*p* < 0.05).

#### *3.5. Polyphenolic Quantification and Antioxidant Capacity*

RSF and RFT reported higher phenolic content than WF and WFT, respectively (*p* < 0.1). RSF showed higher total flavonoid content than WF (*p* < 0.01) and RFT and WFT were equal between them (*p* = 0.08) (Table 4). Compared with other studies, the total phenol content of RSF was higher than that reported by Tokpunar [53] for RSF (24.6 mg GAE/g), this could be due to genetic and environmental differences between *B. alicastrum* trees, which are wild trees with an extensive geographical distribution [9,24]. The phenolic content of RSF was also higher than that of other seeds, such as walnut (15.6–16.3 mg GAE/g sample), pecan nut (12.8–20.2 mg GAE/g sample), pistachio (8.7–16.6 mg GAE/g sample) or almond (2.4–4.2 mg GAE/g sample) [54] and 73 times higher than WF.

**Table 4.** Phytochemical content and antioxidant capacity of flour and tortillas samples.


Mean ± SD. All values are presented on dry weight basis. TPC—total phenolic compounds, TF—total flavonoids, DPPH -.2,2-diphenyl-1-picryl-hydrazyl-hydrate free radical assay, ABTS - 2,2'-azinobis-(3-ethylbenzothiazoline-6-sulfonate assay, FRAP -ferric ion reducing antioxidant power assay. GAE—gallic acid equivalents, CE—catechin equivalents, TEAC—trolox equivalents antioxidant capacity, RSF—ramón seed flour, WF—wheat flour, RFT—ramón flour tortilla, WFT—wheat flour tortilla. Comparison between flours and between tortillas. Different letters indicate significant difference (*p* < 0.05).

The content of phenolic compounds in RFT was 12 times higher than WFT and also exceeded the content of the previously mentioned nuts, and was similar to blackberry (27.1 mg GAE/g) [55]. In the same way, RFT was higher in flavonoids content than WFT, or other foods, such as blackberry (0.6 mg CE/g) [56], pistachio and almond (0.14 and 0.15 mg CE/g, respectively) [57]. In this sense, partial substitution of RSF in the tortilla samples provided a high phenolic and flavonoid contribution.

In relation to antioxidant capacity, RSF showed 48 times more activity than WF in the ABTS assay. Also, it was higher than blackberry (11.4 mmol TEAC/100g) [56] and equal to walnut (13.7 mmol TEAC/100 g) [58]. In the case of the tortillas, RFT reported twice as much antioxidant activity as WFT. The ABTS assay showed the highest values of antioxidant activity in all the samples. ABTS

is sensitive to the presence of hydrophilic and lipophilic compounds, and a good correlation can be usually observed between the content of total phenols and antioxidant capacity [59], so it is a suitable technique to clarify the impact of RSF on the potential functional properties of RFT.

#### *3.6. Identification of Individual Phenolic Compounds in RSF by UPLC-QTOF-MS*

In addition to the nutritional value of RSF, the presence of a high content and diversity of phenolic compounds, many of which are strong free-radical scavengers, provides this product with added potential as a functional food. Twenty phenolic compounds were tentatively (comparison of high-resolution m/z value and isotope distribution) or positively (MS data plus retention time of available standards) identified in RSF (Table 5).

Many of them are phenolic acids esterified with quinic acid. Compounds 3, 5 and 8 with Rt = 0.53, 0.94 and 1.54 min and a m/z = 353.0885, 353.0884, and 353.0882 were tentatively identified as isomers of caffeoylquinic acid, a derivate of caffeic acid esterified with quinic acid differing only in the position of esterification (C3, C4 or C5 of the aryl ring of quinic acid). Only compound **5** was positively identified as chlorogenic acid (Figure 4C). Compounds 16 and 17 were tentative identified as dicaffeoylquinic acid isomers with the same m/z = 515.1282 and 515.1206. These compounds have a structure of two moieties of caffeic acid and one of quinic acid linked through esterification at different positions into the aryl ring of quinic acid (Figure 4B). Compound 10 (Rt = 2.40 min, m/z = 367.1036) was tentatively identified as 3-O-feruloylquinic acid (Figure 4G). Compound **6** (m/z = 137.0244 and a Rt = 1.03 min) was tentatively identified as any of the possible isomers (*o*-,*m*-,*p*-) of hydroxybenzoic acid (Figure 4E). Compounds 9 and 12, wich eluted at Rt = 1.66 and 2.91 min respectively, with a m/z = 337.0936 and 337.0928, were tentatively identified as isomers of coumaroylquinic acid (Figure 4D). Compound 18 (Rt = 4.96 min, m/z = 499.1259) was tentatively identified as a coumaroyl-caffeoylquinic acid derivate (Figure 4A). Compound **1** (Rt = 0.47 and m/z = 329.0886) was tentatively identified as vanillic acid glucoside and was the only glycosylated phenolic acid (Figure 4F).

**Figure 4.** Structures of phenolic compounds tentatively or positively identified in RSF extract, by ultra-performance liquid chromatography quadrupole time of flight mass spectrometry (UPLC-QTOF-MS) analysis. Numbers in red indicate the possible isomers of each structure. Names of compounds (**A**–**P**) are provided in the figure and description of their identification by MS is given in the text.


 

**Table 5.**

Retention times, and

characteristic

ions

of

phenolic compounds

 found in RSF extracts.

Abbreviations: Rt. Retention time \* Identification confirmed by commercial standards.

#### *Foods* **2019**, *8*, 613

These compounds, linked by esterification with quinic acid, were the most numerous phenolic compounds in the RSF extract, representing half of the total compounds detected. Free quinic acid was also found (Compound 7, Figure 4J). This was expected since quinic acid is an organic acid abundant in plant tissues, that participates in metabolic routes as synthesis of lignin [60]. Succinic and cinnamic acids were also identified tentatively (Compounds 1 and 13, Figure 4I,H, respectively), both organic acids are precursors in the biosynthesis of flavonoids, isoflavones, and stilbenes [61].

Only a few flavonoids, stilbenes and their glycosylated derivates were identified in the RSF extract. Two stilbenes were tentatively identified, compound **20** (Rt = 7.06 and m/z = 227.0723) as cis or trans resveratrol (Figure 4K), and compound **19** (Rt = 5.58 and m/z = 243.0673) as piceatannol (Figure 4L), a more hydroxylated derivative of resveratrol. Compound 11 (Rt = 2.54 and m/z = 345.0628) was tentatively identified at as syringetin, a dimethylated flavonoid (Figure 4M). Compounds 14 and 15 (Rt = 3.73 and 3.86 with m/z = 609.1457 and 463.0894, respectively) were tentatively identified as glycosylated flavonoids: kaempferol 3-dihexoside (Figure 4N), whose structure contains two sugar moieties and isoquercetin, that is formed with quercetin and glucose (Figure 4O). Finally, one of the few free phenolics tentatively identified in RSF with Rt = 0.59 and m/z = 441.0813, was catechin gallate (Figure 4P). As described, the most abundant portion of phenolic compounds in the RSF extract consisted of phenolic acids esterified with quinic acid, while a minor portion was more heterogeneous containing flavonoids, stilbenes, and flavan-3-ols, mostly esterified with sugar moieties.

Only one study has been published identifying phenolic compounds in RSF, where hydroxycinnamic, gallic, vanillic, caffeic, and coumaric acids were identified in acid-hydrolyzed methanol extract and one flavonoid, epicatechin, was released by a continuous alkaline extraction [62]. In comparison with this previous study, a greater number of phenolic compounds were identified in RSF methanol extract in the present work. This was made possible by using the UPLC-QTOF-MS equipment whose main advantage is to provide a tentative identification of compounds for which no commercial standards are available, by using information about the molecular weight of individual phenolic compounds. To our knowledge, this is the first study to report a more detailed profile of phenolic compounds in ramón seeds using this technique, allowing for the identification of their native structures without alterations by hydrolytic reactions. This is important because acid and alkaline hydrolysis can degrade the glycosyl or ester linkages, but at the same time can fragment the structure of the phenolic compounds and alter their subsequent identification [63]. It is worth mentioning that no hydrolysable or condensed tannins were identified in RSF, which should be considered beneficial since tannins, despite their high antioxidant capacity are also known to interfere in nutrient absorption, which is not desirable in a product aimed at groups in poverty and/or malnutrition.

Derivatives of phenolic acids esterified with quinic acid, such as chlorogenic acid and the caffeoyl, coumaryl and feruloyl derivatives, found to be abundant in ramón seed, are typically found in coffee beans, which are recognized as a good source of healthy antioxidant compounds with health benefits like cardiovascular risk prevention. Special mention deserves those derivatives in which one quinic acid is esterified with two phenolic acid moieties; for example, dicaffeoil quinic acid, which has been found in medicinal plants and has a greater antioxidant activity than free phenolic acids [64]. The presence of stilbenes, such as resveratrol is also worth mentioning, although their abundance was low, and their identity should be confirmed. Resveratrol is a phenolic compound that demonstrates a wide range of health benefits as antioxidant, anti-inflammatory, and anti-proliferative in cancer cells. It has been identified in foods like peanuts, grapes and their products like roasted peanut butter and wine [65]. So, the fact that RSF can be considered a novel source of resveratrol for foods that regularly do not have it, such as tortilla or other food formulations, endows this seed with meaningful functional potential. The small number of flavonoids like syringetin, kaempferol and isoquercetin can further increase the healthy properties of RSF.

#### **4. Conclusions**

According to the results obtained in this study, *Brosimum alicastrum* Sw. seed flour (RSF) was a good source of protein, dietary fiber and minerals, such as copper and potassium, and a natural iron resource comparable to iron-fortified wheat flour (WF). In addition, RSF has a high antioxidant capacity (AC) and is rich in phenolic compounds, mainly chlorogenic acid and other phenolic acids esterified with quinic acid, although stilbenes and flavonoids were also present. The partial substitution of RSF (25%) in a wheat flour tortilla (RFT) increased dietary fiber, copper and potassium content. The sensory characteristics of RFT were like those of traditional flour tortilla (WFT), except for the light brown color and higher rollability. RFT was soft and less cohesive and it was accepted by 64% of consumers. Also, RFT increased 12 times its content of total polyphenolic compounds and twice its AC compared to WFT. Therefore, ramón seed flour improves the nutritional value of wheat flour tortilla and may provide potential functional properties that contribute to a healthier diet.

#### **5. Patents**

Patent request: MX/a/2018/011397. Dough of wheat flour and *Brosimum alicastrum* Sw. (ramón) flour for elaboration of food products, preferably tortilla.

**Author Contributions:** Conceptualization, N.R.M.-R.; Data curation, A.L.-S.; Formal analysis, R.S.-C., M.L.R.-V., L.A.d.l.R., A.A.V.-F. and N.R.M.-R.; Investigation, R.S.-C., A.L.-S., L.A.d.l.R. and N.R.M.-R.; Methodology, M.L.R.-V., L.E.S.-C., M.G.-M., A.A.V.-F., J.R.-G., J.A.N.-G. and N.R.M.-R.; Project administration, N.R.M.-R.; Resources, A.L.-S.; Supervision, L.E.S.-C., M.G.-M., J.R.-G. and A.Y.C.-A.; Writing—Original draft, R.S.-C.; Writing—Review and editing, L.A.d.l.R., A.Y.C.-A. and N.R.M.-R.

**Funding:** Part of this work was supported by the CONACyT (project CB-2016-286449).

**Acknowledgments:** The authors thank Consejo Nacional de Ciencia y Tecnología (CONACyT) for funding Rodrigo Subiria Cueto's master scholarship in Universidad Autónoma de Ciudad Juárez, Chihuahua, México.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### **Sorghum–Insect Composites for Healthier Cookies: Nutritional, Functional, and Technological Evaluation**

#### **Temitope D. Awobusuyi, Muthulisi Siwela \* and Kirthee Pillay**

Department of Dietetics and Human Nutrition, University of KwaZulu-Natal,

Pietermaritzburg 3209, South Africa; temmybusuyi@gmail.com (T.D.A.); pillayk@ukzn.ac.za (K.P.)

**\*** Correspondence: siwelam@ukzn.ac.za; Tel.: +27-33-260-5459

Received: 9 September 2020; Accepted: 25 September 2020; Published: 9 October 2020

**Abstract:** Protein-energy malnutrition (PEM) is a major health concern in sub-Saharan Africa (SSA). Relying on unexploited and regionally available rich sources of proteins such as insects and sorghum might contribute towards addressing PEM among at-risk populations. Insects are high in nutrients, especially protein, and are abundant in SSA. Sorghum is adapted to the tropical areas of SSA and as such it is an appropriate source of energy compared with temperate cereals like wheat. It is necessary to assess whether cookies fortified with sorghum and termite would be suitable for use in addressing PEM in SSA. Whole grain sorghum meal and termite meal were mixed at a 3:1 ratio (*w*/*w* sorghum:termite) to form a sorghum–termite meal blend. Composite cookies were prepared where the sorghum–termite blend partially substituted wheat flour at 20%, 40%, and 60% (sorghum–termite blend:wheat flour (*w*/*w*). The functional and nutritional qualities of the cookies were assessed. Compared with the control (100% wheat flour), the cookies fortified with sorghum and termite had about double the quantity of protein, minerals, and amino acids. However, with increased substitution level of the sorghum–termite blend, the spread factor of the cookies decreased. There is a potential to incorporate sorghum and termite in cookies for increased intake of several nutrients by communities that are vulnerable to nutrient deficiencies, especially PEM.

**Keywords:** protein energy malnutrition; insect; sorghum; wheat

#### **1. Introduction**

The World Health Organization (WHO) estimates that about 60% of all deaths occurring among children under five years of age in developing countries could be attributed to malnutrition [1]. Protein-energy malnutrition (PEM) results from deficiency in dietary protein and/or energy in varying proportions [2]. Sub-Saharan Africa (SSA) continues to lead in bearing the brunt of PEM and globally, the prevalence has continued to rise up to 47.3% with the worst increase occurring in regions of Africa [3,4]. Therefore, an improvement in nutrition is needed to decrease the high mortality and morbidity rates associated with PEM [5].

Sorghum, a drought-tolerant staple, contributes to the diet of over half a billion people in the regions, where maize struggles to grow if there are very limited agronomic intervention technologies [6,7]. Baked products, such as bread and cookies are part of the leading foods world-wide, including in the sub-Saharan African region. Therefore, they are the most appropriate vehicles to deliver vital nutrients, for example protein to vulnerable populations [8]. In SSA, the limitation of producing wheat due to the less conducive climatic conditions and the exorbitant demand on forex for its importation creates a necessity to try to substitute wheat with the well adapted cereals like sorghum in baked products [9]. However, sorghum grain is low in protein content, while the essential amino acids lysine, threonine, and tryptophan are limited [10,11]. Hence, if sorghum were to be used to partially substitute

wheat in baked goods, it would be necessary to find an accessible, yet high quality source of nutrients, including protein, to complement it.

Insects are a traditional source of food in several parts of the world. They are especially rich in protein, calcium, iron, and zinc [12,13]. The energy content of insects is comparable to that of meat, except for pork, because of its particularly high fat content [14]. Furthermore, given that food insecurity is prevalent in SSA, the use of insects in this region, where they are already being consumed, although not at a nutritionally significant level, should be promoted to serve as an alternative protein source, in particular. Thus, the incorporation of insect in popular, staple foods to complement staple cereal grains should be considered.

Cookies are an energy dense and shelf-stable, popular baked ready-to-eat snack, consumed by both children and adults globally [15,16]. The main ingredients in cookie baking include wheat flour, fat, sugar, butter, and water. Other added ingredients may be optional or added to improve organoleptic attributes [17]. With the afore-mentioned nutritional advantages of insects and the popularity of cookies, it might be advantageous to complement sorghum with insects in partial replacement of wheat flour in cookies to contribute to addressing PEM in developing regions, especially SSA.

Sorghum–legume cookies in which sorghum was combined with sunflower and peanut flours, to increase the protein content of the cookies have been reported [18]. In addition, Mridula et al. [19] reported that acceptable cookies could be developed with wheat–sorghum composite flours with up to 50% sorghum substitution level. Several studies have demonstrated the potential for supplementing wheat flour with sorghum in bread, and cookies, and other snacks [9,20,21]. The influence of finger millet flour [22], fibers from different cereals [23], maltodextrin, and guar gum [24] on the rheological properties of dough and quality of cookies has also been reported. However, it appears that the compositing of wheat, sorghum, and termite to make cookies has not been reported. Therefore, this study aimed to determine the effect of partially substituting wheat flour with a sorghum–termite blend on the nutritional composition and functional properties of cookies.

#### **2. Materials and Methods**

#### *2.1. Preparation of Sorghum and Termite Meal Blend*

Winged termites (*Macrotermes belliscosus*), harvested during the harmattan season were purchased from Oja-oba main market in Ondo State, Nigeria, and used in this study. The termites were de-winged and cleaned three times to remove soil and dirt. They were oven-dried at 40 ◦C for 8 h [25]. Dried insects were milled into a meal with a blender to a particle size of ≤1 mm, vacuum packed, labelled and stored at −4 ◦C until analysis. The termites were washed, dried in the oven, and milled into a meal. Sorghum grain was purchased and cleaned to make sure they were free of dirt. A mill fitted with a 0.4 mm screen was used to grind whole grain sorghum meal into a meal [9]. Both the sorghum and termite meal were used to substitute wheat flour at varying proportions (Table 1).


**Table 1.** Ratios of ingredients (wheat:sorghum and termite) for cookie formulation.

Sorghum:termite meal (ratio 3:1) replaced wheat flour at 0, 20, 40, and 60% (*w*/*w*) levels.

#### *2.2. Methods*

#### 2.2.1. Preparation of Cookie Samples

Cookies fortified with sorghum and termite as well as the control were prepared according to the method described by de Jager [26], with minimal modification. The sorghum meal and insect meal were mixed at a 3:1 (*w*/*w*) ratio to form a sorghum–termite blend. The ratio 3:1 was chosen after preliminary trials in the laboratory revealed that other substitution levels of sorghum and termite resulted in cookies that were too brittle. Experimental cookies were prepared where wheat flour was partially substituted with different proportions of the sorghum–termite blend, 20%, 40%, and 60% (*w*/*w*), separately. Cookies (100% wheat) in which no sorghum or insect was added served as the control (Table 1). About 200 g of sugar, 5 mL of vanilla essence, 5 mL of salt, 50 g of powdered milk, and 10 mL baking powder were sieved and mixed together with 480 g wheat flour for three to five minutes. About 250 g of margarine was added to the mixture and kneaded for two minutes to form a firm dough. The dough was rolled out, cut into desired shapes, and transferred into the oven. Cookies were baked at 150◦ C in a preheated oven for 20 min. Cookies were crushed to a particle size of ≤ 1 mm for chemical analyses and then stored.

#### 2.2.2. Nutritional Composition

The nutritional composition of cookies was determined by standard methods stated below.

#### Ash

Ash was determined using the Association of Official Analytical Chemists (AOAC) official method 923.03 [27].

#### Protein

Crude protein was measured using the AOAC official method 968.06 [28].

#### Glycemic (Available) Carbohydrate Content

Glycemic carbohydrates were calculated by difference.

#### Fat

Fat content was determined according to the AOAC Official Method 920.39C [29].

#### Fiber

Fiber was done based on the method reported by Saha et al. [22].

#### Gross Energy

Gross energy was determined according to the AOAC Official Method 935.42 [30].

#### Selected Minerals

Mineral content was determined by the AOAC Official Method 6.1.2 [31].

#### Amino Acids

The amino acid profile of the cookie samples was analysed by the Waters API Quattro Micro Method, which consists of a column C18, 1.7 μm, 2.1 × 100 mm and a binary solvent manager. Samples (400 mg) were subjected to AccQ-Tag Ultra Derivatization kit (WatersTM, Johannesburg, South Africa); 10 μL of the undiluted sample was added to the Waters AccQ-Tag kit constituents and placed in a heating block at a temperature of 55 ◦C for 10 min. Injection volume was 1 μL and gradient separation was performed using Solvents A and B from the Waters Accutag kit.

#### In Vitro Protein Digestibility

In vitro protein digestibility was determined using the method described by Hamaker et al. [32].

#### 2.2.3. Physical Characteristics

Physical quality parameters of the cookie samples, such as diameter, thickness, spread ratio, and spread factor were determined using the procedure of the American Association of Cereal Chemists [33].

#### Texture

Texture analysis of cookies fortified with sorghum and termite (Figure 1) was done using a TA-XT plus 100C model texture analyzer (Stable Micro Systems, Godalming, UK). The cookies were measured for hardness and fracturability using a three-point bend rig attachment at a 3.0 mm/s cross head speed for a 5 mm distance and a 5 kg load cell.

**Figure 1.** Cookies fortified with sorghum and termite. C0: control (100% wheat cookies); C20: 15% sorghum and 5% termite substitution level; C40: 30% sorghum and 10% termite substitution level; C60: 45% sorghum and 15% termite substitution level.

#### 2.2.4. Functional PropertiesWater and Oil Absorption Capacity

#### Water and Oil Absorption Capacity

The water and oil absorption capacities were determined by the method of Sosulski et al. [34].

#### Bulk Density

The process reported by Okaka and Potter [35] was used to determine the bulk density of the cookie flours.

#### 2.2.5. Statistical Analysis

The resulting data was analysed using the Statistical Package for Social Science (SPSS version 20.0 SPSS Inc., Chicago, IL, USA). One-way analysis of variance (ANOVA) was done; and separation of means was by Fisher Least Significance Difference (LSD) test. A *p*-value of ≤0.05 was considered significant.

#### **3. Results**

#### *3.1. Proximate Composition*

Protein, fat, and carbohydrates constitute the major nutrients in cookie samples (Figure 1). The proximate composition of the cookies fortified with the sorghum–termite blend is presented in Table 2. As expected, the highest protein content was observed in cookies containing 60% sorghum and termite substitution level, which contained the most insect. The results showed that fat content increased with increasing concentrations of the sorghum–termite blend. The fiber and ash content of cookie samples fortified with the sorghum–termite blend was also significantly higher than the

control. Carbohydrate content was higher in the control cookies than in the cookies fortified with the sorghum–termite blend. The gross energy of the cookies fortified with the sorghum–termite blend was higher than the control.

**Table 2.** The sorghum–termite blend on the proximate composition (g/100 g) and gross energy (kJ) of cookies 1.


Mean (±*SD*) of three determinations; CHO: glycemic carbohydrates; Means with different superscripts in a column vary significantly (*p* < 0.05); <sup>1</sup> Values are on dry matter basis. kJ; refers to Kilojoules; C0: control (100% wheat cookies); C20: 15% sorghum and 5% termite substitution level; C40: 30% sorghum and 10% termite substitution level; C60: 45% sorghum and 15% termite substitution level.

#### *3.2. Mineral Composition*

The minerals abundant in the cookie samples were iron, phosphorus, and magnesium (Table 3). The incorporation of the sorghum–termite blend substantially increased the mineral concentration of the cookies. Zinc and iron were abundant in the cookie samples fortified with the sorghum–termite blend. Cookies containing 60% of the sorghum–termite blend had the highest concentration of these minerals compared with the cookie samples with lower concentrations of the sorghum–termite blend.


**Table 3.** Selected mineral elements in cookies fortified with sorghum–termite blend (mg/100 g) 1.

Mean (±*SD*) of three determinations. Means with different superscripts in a column vary significantly (*<sup>p</sup>* <sup>&</sup>lt; 0.05). <sup>1</sup> Values are on dry matter basis. <sup>2</sup> Food and Agriculture Organization (FAO)/World Health Organization (WHO) 2006 Recommended daily intake for children and adults. C0: control (100% wheat cookies); C20: 15% sorghum and 5% termite substitution level; C40: 30% sorghum and 10% termite substitution level; C60: 45% sorghum and 15% termite substitution level.

#### *3.3. Amino Acid Content*

Table 4 shows that cookies containing the sorghum–termite blend had substantial amounts of amino acids and lysine and leucine were the major amino acids present. Cookies fortified with the sorghum–termite blend showed the highest increase across all amino acids when compared with the control.


**Table 4.** Essential amino acid composition of cookies fortified with sorghum–termite blend (mg/100 g) 1.

<sup>1</sup> Values are on dry matter basis; <sup>2</sup> Food and Agriculture Organization/World Health Organization (2007). C0: control (100% wheat cookies); C20: 15% sorghum and 5% termite substitution level; C40: 30% sorghum and 10% termite substitution level; C60: 45% sorghum and 15% termite substitution level.

#### *3.4. In Vitro Protein Digestibility*

The in vitro protein digestibility (IVPD) results are presented in Table 5. The addition of the sorghum–termite blend improved the digestibility of cookies, which increased by 23.8% with the incorporation of the sorghum–termite blend from 67% to 83%.

**Table 5.** In vitro protein digestibility of cookies fortified with sorghum–termite blend.


Mean (± *SD*) of three determinations. Means with different superscripts in a column vary significantly (*p* < 0.05). C0: control (100% wheat cookies); C20: 15% sorghum and 5% termite substitution level; C40: 30% sorghum and 10% termite substitution level; C60: 45% sorghum and 15% termite substitution level.

#### *3.5. Physical Characteristics*

The physical characteristics of cookies fortified with the sorghum–termite blend are shown in Table 6. The results showed that the addition of sorghum–termite blend reduced the weight of cookies, compared with the control. Cookies fortified with 60% sorghum–termite blend (24.5 g) recorded the highest weight loss when compared with the control (29.5 g). There was a progressive increase in the thickness of the cookies fortified with the sorghum–termite blend (7.5 mm, 7.7 mm, and 7.9 mm for sample C20, C40, and C60, respectively), when compared with the control (7.3 mm).

**Table 6.** Physical qualities of cookies fortified with sorghum–termite blend.


Mean (± *SD*) of three determinations. Means with different superscripts in a column vary significantly (*p* < 0.05). C0: control (100% wheat cookies); C20: 15% sorghum and 5% termite substitution level; C40: 30% sorghum and 10% termite substitution level; C60: 45% sorghum and 15% termite substitution level.

#### *3.6. Texture and Colour*

The effect of incorporating sorghum and termite on the texture and colour of cookies is shown in Table 7. The results showed that as the substitution level of the sorghum–termite blend increased, the cookie hardness decreased. Further, cookies became darker with increasing concentration of the sorghum–termite blend. The L\* (lightness) values decreased and the a\* (redness) values increased, while b\* (yellowness) values were similar across all cookie samples.


**Table 7.** Texture and colour of cookies fortified with sorghum–termite blend.

Mean (± *SD*) of three determinations. Means with different superscripts in a column vary significantly (*p* < 0.05). L\* = Lightness, a\* = Redness, b\* = Yellowness. C0: control (100% wheat cookies); C20: 15% sorghum and 5% termite substitution level; C40: 30% sorghum and 10% termite substitution level; C60: 45% sorghum and 15% termite substitution level.

#### *3.7. Water and Oil Absorption Properties of Cookie Flours*

The oil absorption capacity (OAC) varied between 1.43 and 1.65 g oil/g flour (Figure 2) while the water absorption (WAC) varied from 1.55 g to 1.78 g water/g flour. The results revealed that the higher the sorghum and termite substitution level, the higher the absorption capacities of the flours.

**Figure 2.** Water and oil absorption capacity of cookie flours. Water absorption is expressed as g water/g flour. Oil absorption is expressed as g oil/g flour. Error bar values are actual values obtained. C0: control (100% wheat cookies); C20: 15% sorghum and 5% termite substitution level; C40: 30% sorghum and 10% termite substitution level; C60: 45% sorghum and 15% termite substitution level.

#### *3.8. Bulk Density of Cookie Flours*

The packed bulk density (PBD) and loose bulk density (LBD) varied (Figure 3). The PBD was 0.69 g/mL for the control and 0.96 g/mL for the 60% composite flour containing the sorghum–termite blend, which was much higher than the control. The LBD ranged between 0.59 and 0.87 g/mL. All composite flours containing wheat, sorghum and termite had relatively higher LBD than the wheat flour (control).

**Figure 3.** Bulk density of cookie flours. Error bar values are actual values obtained. C0: control (100% wheat cookies); C20: 15% sorghum and 5% termite substitution level; C40: 30% sorghum and 10% termite substitution level; C60: 45% sorghum and 15% termite substitution level.

#### **4. Discussion**

#### *4.1. Proximate Composition*

Protein content was significantly higher in all cookies fortified with the sorghum–termite blend (36.4 to 41.0 g/100 g), compared with the control (10.5 g/100 g). This result agrees with Koffi-Niaba et al. [36], who reported that supplementing sorghum with termite flour significantly improved the protein content from 9.6% in the control to 21.7%. A report by Kinyuru et al. [37] working on the nutritional quality of wheat buns enriched with edible termites also found that the buns showed a significant increase in protein content (47.5%), when compared with buns without termite substitution. It has been reported that fortifying sorghum with defatted soy flour significantly enhanced the protein quality and content of cookies [21]. Similar results have also been reported by Omoba and Omogbemile [38]. Further, as the concentration of the sorghum–termite blend increases, a substantial increase was observed in the fat content. This is expected as termites, the insect used in this study ranked among the highest in fat concentration [39,40]. Fibre also recorded a significant increase across all cookies containing sorghum–termite blend due to the addition of insects (10.2 to 13.5 g/100 g). Fiber is beneficial in the human diet to reduce the risk of heart disease, blood pressure, obesity, and lower cholesterol levels [41].

The ash content of the cookie samples fortified with the sorghum–termite blend (3.5 to 4.2 g/100 g) was significantly higher than the control (1.7 g/100 g). However, C0 (control), had the highest carbohydrate content (54%). Cookies containing 60% sorghum–termite blend had the highest energy value (2217.6 kJ/100 g) and the lowest was recorded in the control (C0) (gross energy: 1180.2 kJ/100 g). This result could be attributed to the high fat content of the experimental cookies. The energy value of edible insects has been reported to depend mainly on their fat content [38]. Similar results were reported by Koffi-Niaba et al. [36], they obtained between 1626 kJ/100 g and 1712 kJ/100 g in sorghum-based cookies fortified with termites. Overall, these are promising results and developing countries, where PEM remains a major problem, could benefit from the consumption of these cookies.

#### *4.2. Mineral Composition*

Cookie samples fortified with the sorghum–termite blend had increased mineral content compared with the control (Table 3). The calcium content was significantly higher (5.5 to 10.8 mg/100 g) when compared with the control (2.5 mg/100 g). The iron content varied significantly, with the highest level (37.4 mg/100 g) found in sample C60 (45% sorghum, 15% insect substitution level). Although insects are known to have high levels of minerals [40], the higher levels noted in this study could also be partially attributed to the high iron content in sorghum [42]. Zinc levels were also significantly high across all cookies fortified with the sorghum–termite blend (8.4 to 14.8 mg/100 g), compared with the control (2.5 mg/100 g). This supports Yhoungaree et al. [43], who stated that insects are a valuable source of iron and zinc. A previous study used caterpillar cereal to prevent anemia and stunting in infants. The authors found that the caterpillar cereal produced had appropriate macro and micronutrient contents and concluded that it could be used for complementary feeding [44]. This further supports the results in this study, which showed that consumption of insects, could prove to be a valuable measure to combat deficiencies of iron and zinc in developing countries.

In this study, phosphorous content was highest in the cookie sample containing 60% sorghum– termite blend (37.6 mg/100 g) and lowest in the control (0.8 mg/100 g). Magnesium ranged from 24.3 to 33.5 mg/100 g in cookies fortified with the sorghum–termite blend, which was significantly higher than the control (1.8 mg/100 g). Copper ranged from 0.8 mg/100 g in the control to 3.8 mg/100 g in sample C60 (45% sorghum and 15% termite substitution level). Manganese ranged between 13.2 and 24.5 mg/100 g in cookies fortified with the sorghum–termite blend, which was higher than the control (1.2 mg/100 g). Overall, the addition of sorghum and termite substantially increased the mineral profile of the cookies. This study demonstrates the potential of edible insects to increase the intake of minerals, which are well reported to be deficient and causing severe public health problems in poor populations [45]. Given the worldwide deficiencies of these minerals among human population groups [46], insect-fortified cookies would supply the amount of iron and zinc required for basic body functions. Further, the bioavailability of minerals from insects is likely to be higher than that from plant foods because their nutrients are easily assimilated by the human body and there are no antinutritional factors such as phytic acid and oxalic acid in insects [47,48]. Previous studies found an appreciable concentration of minerals: calcium, iron, magnesium, copper, potassium, sodium, and zinc, and a particularly high iron content present in termites (*syntermes* soldiers) [49]. Chakravorty et al. [50] also reported that insects; *Oecophylla smaragdina* and *Odontotermes* sp. can serve as a source of micronutrients such as Fe, Zn, Cu, and Mn. Therefore, the consumption of insects should be encouraged, especially among rural communities with low animal protein intake, to contribute to meeting their nutritional requirements.

#### *4.3. Amino Acid Profile*

Lysine was significantly higher in cookies fortified with the sorghum–termite blend. The increase in lysine could also be attributed to the addition of the insect (termite) meal. It has been reported that the most concentrated essential amino acid found in termites was lysine [51]. As stated earlier, cereal grains are important staples in diets globally but are generally lysine deficient [39]. Therefore, supplementing cereal-based foods with insect is recommended as it would improve the lysine content of the foods. The lysine content reported in this study accounted for 80% of the recommended intake for children and 100% requirements for adults. The fact that insects are a traditional food in most developing regions, including SSA [52] is an advantage. Tryptophan and threonine known to be deficient in cereal proteins were also significantly higher in cookies fortified with the sorghum and termite meal. Tryptophan content was lowest in the control (10 mg/100 g) and ranged between 18 mg/100 g and 32 mg/100 g in cookies fortified with the sorghum–termite blend. Threonine content in cookies fortified with the sorghum–termite blend was between 30 mg/100 g and 46 mg/100 g, compared with 21 mg/100 g in the control. High concentrations of methionine and cysteine were also found (20 to 29 mg/100 g and 22 to 33 mg/100 g, respectively), in comparison to the control (18 mg/100 g and 18 mg/100 g methionine and cysteine, respectively). Histidine content ranged from 21 to 43 mg/100 g. Histidine is a precursor of histamine, which is present in small quantities in cells. Histamine communicates messages to the brain, triggers the release of stomach acid to aid digestion, and is released after an injury or allergic reaction as part of the body's immune response [53]. Further, children grow poorly if there is an absence of histidine in their diet [53]. Therefore, cookies fortified with sorghum–termite blend developed in this study could be a good source of histidine required by children. Isoleucine (35 to 46 mg/100 g), leucine (48 to 63 mg/100 g), phenylalanine (33 to 44 mg/100 g), tyrosine (32 to 42 mg/100 g), and valine

(42 to 47 mg/100 g) were also present in abundance. Overall, the concentrations of amino acids obtained in this study are higher compared with the concentrations found in meat sources such as beef, pork, and chicken meat [54]. Although it has been reported that the nutritional composition of insects may vary due to their feeding habits and harvesting season [51], values reported in this study could be largely dependent on the environmental factors from where the termites were purchased. As compared with the 2007 Food and Agriculture Organization (FAO)/World Health Organization (WHO) standard, the concentrations of the essential amino acids in all cookies fortified with the sorghum–termite blend were generally higher than the pattern of amino acid requirements for both children and adults [55] (Table 4). Hence, the developed cookies would be able to contribute to the essential amino acids in the human diet. Although most of the amino acids reported exceeded the requirements, they can be viewed as beneficial, especially for population groups whose staple diet consists of maize and wheat, which may lack some of the important amino acids [56].

#### *4.4. In Vitro Protein Digestibility*

In vitro protein digestibility (IVPD) ranged from 67% to 83% (Table 5). The reason for the higher IVPD in the cookies fortified with the sorghum–termite blend compared with the control is likely due to the higher digestibility of insect protein. Insect protein is highly digestible, and a range of 77% to 98% digestibility has been reported [39,57]. The results of this study are similar to the results reported by Ajayi [58], who reported high digestibility for winged termites (83.41%) and soldier termites (81.10%).

#### *4.5. Physical Characteristics*

The results show that as the sorghum–termite blend substitution level increased, cookie diameter decreased. Cookies containing 60% of the sorghum–termite blend had the lowest spread factor. The experimental cookies had a lower spread factor relative to the control most likely due to dilution of gluten, which is essential in the expansion of baked products [59–61].

#### *4.6. Texture and Colour*

The decrease in hardness and high fracturability of cookies fortified with the sorghum–termite blend may be attributed to dilution of gluten in the experimental cookies, because sorghum and termite do not contain gluten proteins. Gluten, which is formed during the dough mixing process and coagulated into a fiber-like foam, is responsible for the mechanical structure of baked products [61]. Furthermore, coarse particles may have been introduced by the increased fiber content of cookies, interfering with the homogeneity of the dough and cookie structure, thereby resulting in lower hardness values [62]. Additionally, the high crumbliness and fragility of the experimental cookies could also be due to high level of bran and the absence of gluten [63]. Table 7 shows that the 100% wheat cookies (control) were the hardest and least fracturable of all the cookies. The C0 (control) recorded the lowest fracturability value, while the maximum fracturability value was recorded for C60 (45% sorghum and 15% termite substitution level). Higher fracturability for cookies enriched with fiber has also been reported [64]. A previous study by Awobusuyi et al. [65], reported that fortifying wheat with a sorghum–insect meal did not compromise the product quality or acceptability, as the texture of the cookie samples containing the sorghum–termite meal was liked and rated the same as that of the control (100% wheat cookies).

The cookies darkened (decreasing Hunter L\* values) with increased concentration of the sorghum–termite blend. This is likely due to the darker colour of the sorghum insect blend compared with wheat flour. Awobusuyi et al. [65], reported that the acceptability of sorghum–insect cookies and the colour acceptability of cookie samples supplemented with 5% sorghum–termite meal was higher when compared with the control (100% wheat biscuits) and cookies with higher concentrations of termites. It has also been reported that cookies made with whole meal sorghum resulted in cookies with a darker colour [63]. In addition, the increased protein content of the experimental cookies would result in production of higher levels of Maillard reaction products, the majority of which are brown pigments [66].

#### *4.7. Water and Oil Absorption Properties of Cookie Flours*

The results revealed that the higher the sorghum–termite blend substitution level, the higher the water absorption capacity (WAC). Water absorption capacity is a product's ability to interact with water under restricted conditions [67]. Previous reports have suggested that flours with high water absorption capacity as the composite flours of this study would be beneficial in bakery products, as this could prevent staling by reducing moisture loss [68]. Similarly, oil absorption capacity (OAC) increased with increasing substitution level of sorghum–termite blend. Oil absorption capacity (OAC) refers to the capability of flour to absorb oil [69]. This is vital because oil acts as a flavour retainer and improves the mouth feel of cookies [70]. The observed trend of an increase in OAC with an increase in the concentration of the termite meal in biscuits may be attributed to the high protein content of the termite meal. The main chemical component affecting OAC in foods is protein, which is composed of both hydrophilic and hydrophobic parts. Hydrophobic proteins possess superior binding of lipids—non-polar amino acid side chains predominant in hydrophobic proteins can form a hydrophobic interaction with hydrocarbon chains of lipids, and thereby increase OAC [71,72]. The blends in this study are potentially valuable in the structural interaction in food, which is also important in developing new food products and the extension of shelf life, particularly in baked foods or other food products where fat absorption is desired [73,74].

#### *4.8. Bulk Density of Cookie Flours*

The results show that all flours containing the sorghum–termite blend had relatively higher packed bulk density (PBD) and loose bulk density (LBD) than the control. This indicated that the sorghum–termite blend had a higher bulk density than wheat flour. As explained earlier, bulk density provides information on the porosity of a product and can affect the choice and design of the packaging materials [42].

#### **5. Conclusions**

Cookies fortified with a sorghum–termite blend have the potential to serve as a protein, energy, and nutrient-rich supplementary food to address PEM. The results of the present study suggest that a sorghum–termite blend can be successfully incorporated into cookies up to a level of 60% (15% insect) and yield cookies of high nutritional value. The contribution of the experimental cookies to dietary iron, zinc, and lysine would be of particular significance as deficiencies of these nutrients remain a problem in Africa, especially in countries in SSA.

**Author Contributions:** Conceptualization, T.D.A., M.S. and K.P; methodology, T.D.A., M.S. and K.P.; investigation; T.D.A., M.S. and K.P.; resources, M.S. and K.P.; data curation, T.D.A.; writing—original draft preparation, T.D.A.; and writing—review and editing, M.S. and K.P. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

#### *Article*

### **Determination of the Sensory Characteristics of Traditional and Novel Fortified Blended Foods Used in Supplementary Feeding Programs**

#### **Sirichat Chanadang 1,2 and Edgar Chambers IV 1,\***


Received: 30 June 2019; Accepted: 15 July 2019; Published: 17 July 2019

**Abstract:** Despite the wide use of traditional non-extruded fortified blended foods (FBFs), such as corn soy blend plus (CSB+), in supplementary feeding programs, there is limited evidence of its effectiveness on improving nutritional outcomes and little information on actual sensory properties. Fifteen novel extruded FBFs were developed with variations in processing and ingredients in order to improve the quality of food aid products based on the Food Aid Quality Review (FAQR) recommendations. Descriptive sensory analysis was performed to determine the effects of the processing parameters and ingredients on the sensory properties of traditional and novel FBFs. The extrusion process affected the aroma and flavor of the tested products. Novel FBFs from the extrusion process had more pronounced toasted characteristics, probably because of the high temperature used during extrusion. The ingredient composition of the FBFs also had a significant impact on the sensory properties of the products. The addition of sugar to novel FBFs leads to a significant increase in sweetness, which could improve acceptance. The level of lipids in binary blends appeared to be mainly responsible for the bitterness of the product. In addition, legumes, which were a primary ingredient, contributed to the beany characteristics of the products. The higher amounts of legume used in the formulations led to beany characteristics that could be perceived from the products and could be a negative trait depending on consumers' prior use of legume-based products.

**Keywords:** fortified blended foods (FBFs); sensory; food aid; extrusion; cereal; legume; infant; child; porridge

#### **1. Introduction**

Food insecurity around the world is always increasing due to many causes, including growing populations, poverty, and natural disasters [1]. The State of Food Insecurity in the World, 2015 reported that approximately 795 million people in the world were undernourished in 2014–2016 [2]. Fortified blended foods (FBFs) were developed in the 1960s by the United States Agency for International Development (USAID, Washington, DC, USA) to provide a supplement for young children who suffered from moderate acute malnutrition in many developing countries around the world [3,4]. The most commonly distributed cereal based FBF by USAID is a corn-soy blend (CSB) which consists of corn (75–80%) as a source of carbohydrate and soy (20–25%) as a source of protein. Although FBFs form an important part of the food aid ration, there is limited evidence of their abilities in treating young children with malnutrition [3–5] and little information on their sensory properties.

The Food Aid Quality Review (FAQR) in 2011 by Webb et al. [6] recommended changing the formulation of existing FBFs in order to improve their nutritional quality. These recommendations

included adding animal-source protein to promote linear growth of children, increasing fat content through the addition of vegetable oil, adding a flavor enhancer to formulations to improve the acceptability of FBFs, and upgrading micronutrient compositions in FBFs. In addition, the decortication of cereals and legumes used in FBFs is recommended in order to reduce the fiber content and eliminate phenolic compounds that can reduce the energy intake and lower protein digestibility and mineral absorption [5].

Another recommendation from Webb et al. [6] was to increase the solids content of FBFs to 20% to increase the nutrient content. However, porridge prepared from the current FBFs at this concentration is too viscous for consumption by infants and young children [7]. Mothers normally add more water into porridge to make it more drinkable before feeding it to their children, which results in a low nutritional value and energy density [5]. Extrusion cooking of starchy ingredients for FBFs can result in less viscous cooked porridge, making them more ideal for delivering higher density energy meals at lower viscosities for infants and young children [8]. Extruded products also require short cooking times and less fuel [5], which makes them more valuable to people with limited time and energy sources.

Webb et al. [6] also encouraged the exploration of new grains or legumes that could be used beyond the traditional FBFs, including CSBs and wheat-soy blends (WSB). Corn has been used as the main staple for current FBFs because it is a good source of starch, plant-based protein, dietary fiber, B vitamins, and is available in bulk for the food aid program [9,10]. However, the high demand of corn for many uses, especially for fuel production, makes the prices increase [11] and this directly affects food aid commodities. Heat-treated soy in full fat form or defatted flour is primarily used as a source of protein in FBFs. However, soy may contain high levels of anti-nutritional factors, such as phytate and phytoestrogen, which have unknown long-tern health effects [9].

Sorghum has been examined as a potential alternative ingredient in FBFs with a number of advantages over corn, including higher levels of protein, fat, and some micronutrients when processed properly [12,13]. Cowpea has also been considered as an alternative legume that can be used in FBFs because of the high levels of protein, energy, and other nutrients [14]. Sorghum and cowpea are cultivated and consumed as part of human foods in many parts of developing countries [14,15]. Therefore, populations in these areas should be familiar with the flavor of sorghum and cowpea, which makes these good candidates for use in FBFs. Moreover, both sorghum and cowpea are mostly non-genetically modified organism (GMO) crops, which allows them to be used in many countries that have banned the use of GMO products.

Recent work has shown that various types of extruded FBFs made with sorghum or corn and cowpea or soy are at least as preferred as CSB+ by children in Tanzania [16]. In addition, data has shown that the shelf-life of such products is generally 24 months or greater [17], far exceeding the required shelf-life for such products.

Based on the recommendations of the FAQR, fifteen newly formulated, extruded FBFs with varied processing techniques and ingredients were developed. The objective of this study was to determine the effects of processing techniques (extrusion vs. non-extrusion, milling type, decortication process, and the step of adding antioxidant to the product) and ingredients on the sensory properties of traditional and novel FBFs.

#### **2. Materials and Methods**

#### *2.1. Samples*

Fifteen novel extruded FBFs and one current-non extruded FBF were used in this study. These products were potential variations in a large feeding trial in Tanzania to test sorghum cowpea blends against other products [18].

#### 2.1.1. Novel Extruded Fortified Blended Foods

Fifteen possible extruded FBFs varied in milling type, decortication process, the order of adding antioxidant to the blends, and ingredients are shown in Table 1.

The whole grains—sorghum varieties V1 (Fontanelle 4525), V2 (738Y), V3 (217X Burgundy) (Nu Life Market, Scott City, KS, USA), and corn (Agronomy Foundation Seed, Kansas State University, Manhattan, KS, USA) were used for pilot milling at Hall Ross Flour Mill (Kansas State University, Manhattan, KS, USA) to obtain whole and decorticated flours. Soybeans (Kansas River Valley Experiment Field, Kansas State University, Manhattan, KS, USA) and cowpea grains (LPD Enterprises LLC, Olathe, KS, USA) were milled at Hall Ross Flour Mill (Kansas State University, Manhattan, KS, USA). Commercially milled whole and decorticated sorghum flour variety V1 were obtained from Nu Life Market, Scott City, KS, USA. Commercially milled degermed corn flour and whole corn flour were purchased from Agricor, Marion, Indiana, USA. Defatted soy flour was purchased from American Natural Soy, Cherokee, IA, USA.

The cereal/legume flours were blended. For seven sorghum-cowpea blends, one of the three sorghum varieties of flour, whole or decorticated, was mixed with cowpea flour. For five sorghum-soy blends, sorghum variety V1, whole or decorticated, was mixed with low fat (1.85%), medium fat (6.94%), or full fat (16.93%) soybean flour. For three corn-soy blends, whole or degermed corn flour with medium fat and full fat soybean flour were used. All binary blends were extruded on a single screw extruder X-20 (Wenger Manufacturing Inc., Sabetha, KS, USA) at a screw speed ranging from 500–550 rpm with 18–24% process moisture. The extrudates were cut at the die exit with a face-mounted five blade rotary knife, and dried in a Wenger double pass Dryer/Cooler (Series 4800, Wenger Manufacturing Inc., Sabetha, KS, USA) at 104 ◦C for 10 min.

The dried extrudates were ground using a Schutte Buffalo Hammer mill (Buffalo, NY, USA). The ground binary blends were then mixed with sugar (Domino Foods, Inc., Yonkers, NY, USA), whey protein concentrate WPC80 (Davisco Foods International, Inc., Eden Prarie, MN, USA), antioxidant (BHA, butylated hydroxyanisole and BHT, butylated hydroxytoluene), vitamins and minerals (Research Products Company, Salina, KS, USA), and non-GMO soybean oil (Zeeland Farm Services, Inc., Zeeland, MI, USA). The composition of all blends is shown in Table 2.

#### 2.1.2. Current Non-Extruded Fortified Blended Food

Corn soy blend plus (CSB+) was produced by Bunge Milling (St. Louis, MO, USA) according to the USDA commodity requirements [19] (Table 2).

#### *2.2. Sample Preparation*

All products were prepared into porridges, which are the most common dishes prepared from cereal-based commodities for children in developing countries [20–22], with 20% solids content according to the recommendation from [6].

A weighted FBF flour (200 g) was mixed with cold water (400 mL) to prevent the formation of lumps. The mixture was then added to boiling water (400 mL), brought back to a boil, cooked with continuous stirring with a wooden spoon for 2 min for extruded FBFs and 10 min for non-extruded FBFs. The sample was removed from the stovetop and cooled to a temperature of 45 ◦C, which is the typical consumption temperature by infant and young children [23].




**Table 2.** Composition of extruded FBFs and non-extruded FBFs.

<sup>1</sup> For extruded FBFs with full-fat soy, WPC80 was increased from 9.5 to 13.0%, and soybean oil was decreased from 9 to 5.5%. <sup>2</sup> Antioxidant was a mixture of 50% butylated hydroxyanisole (BHA) and 50% butylated hydroxytoluene (BHT).

#### *2.3. Descriptive Sensory Analysis*

Descriptive sensory analysis was conducted by six highly-trained panelists at the Center for Sensory Analysis and Consumer Behavior, Manhattan, Kansas USA. All of these panelists had completed 120 h of general descriptive analysis panel training, and had over 2000 h of evaluation experience with a wide array of food products, including cereal-based products.

Sixteen sensory attributes, including 6 aroma and 10 flavor, were evaluated in all samples (Table 3). Some of the same attributes were used in Chanadang et al. [23].

Fifty grams of each prepared porridge was served in a 4 oz styrofoam cup (Dart container corporation, Mason, MI, USA) and labeled with a three-digit code for each panelist. All samples were evaluated on a numerical scale of 0–15 with 0.5 increments, where 0 represents none and 15 represents extremely high. The samples were prepared and evaluated in triplicate in a randomized order.

#### *2.4. Data Analysis*

Sixteen sensory attributes were evaluated for all porridge samples, however, panelists did not detect rancid or painty characteristics in any samples. Therefore, twelve sensory attributes, besides rancid and painty characteristics, were reported and analyzed in this study.

Data for each sensory attribute was analyzed by a one-way ANOVA mixed effect model (SAS version 9.4, The SAS Institute Inc., Cary, NC, USA) using PROC GLIMMIX to determine significant differences (*p* ≤ 0.05) among porridge samples. Tukey's HSD test was used at the 5% level of significance to locate significant effects of the sample on each sensory property. Principal component analysis (PCA) was performed in order to visualize the relationship among sensory attributes and samples using Unscrambler® X 10.5 (Camo, Magnolia, TX, USA).


*Foods* **2019**, *8*, 261


**Table3.***Cont*.

#### *Foods* **2019**, *8*, 261

#### **3. Results and Discussion**

The results showed that six out of twelve sensory attributes were significantly different among porridge samples (*p* ≤ 0.05), including toasted and beany aroma and flavor, sweetness, and bitterness (Table 4).

Porridges prepared from novel extruded FBFs appeared to be higher in toasted aroma and flavor than non-extruded FBF (CSB+), although, not all novel extruded FBFs were significantly different from CSB+ in this sensory characteristic (*p* > 0.05). The high temperature used in the extrusion process might be the main reason for the increased toasted characteristic in extruded FBFs. Extrusion cooking of cereal normally involves thermally induced reactions, including the Maillard reaction, that could generate chemical compounds that correspond to a desirable aroma and flavor of the products [24,25], including such aspects as "toasted" sensory properties. Parker et al. [26] reported that extruded cereal samples with high levels of Maillard reaction products, such as pyrazines and sulfur-containing alicyclic compounds, were generally described as having a desirable toasted or roasted cereal aroma and flavor. Besides the extrusion process, other processing parameters, including types of milling, decortication process, and the step of adding antioxidant to the blends, did not show significant effects on sensory properties of FBFs in this study.

The composition of FBFs seemed to be another important factor that affected the sensory properties of the products. Porridges prepared from sorghum-cowpea blends, especially WSCB-V3, had significantly higher intensity in beany aroma and flavor (*p* ≤ 0.05) than the ones prepared from sorghum-soy and corn-soy blends. Beany characteristics are often found in legume-containing products and are attributed to the action of the lipoxygenase enzyme, which catalyzes the lipid oxidation of linolenic and linoleic fatty acids [27,28]. Since all of the products in this study contained legumes (either soybeans or cowpea), the difference in intensity in beany characteristics among products was primarily due to the amount of legume used in each blend. This probably explains why sorghum-cowpea blends with higher amounts of legume (38.6% cowpea) were higher in beany aroma and flavor.

The variety of sorghum used in FBFs might be another factor that affected the beany property of the products. The blend containing whole red sorghum flour (WSCB-V3) was significantly higher in beany flavor than the rest of the FBFs, except for the one that contained decorticated red sorghum flour (SCB-V3). Vara-Ubol et al. [29] indicated that beany was considered as a combination of attributes, including musty/dusty, musty/earthy, sour aromatics, and characterizing attributes such as green/pea pod, nutty or brown. Red sorghum varieties were reported to have higher dusty flavor [30] and porridges made with red sorghum were also reported to have higher overall flavor intensity [31]. FBFs with red sorghum variety in this study might be higher in dusty flavor or overall intensity, and that resulted in an increased intensity of beany characteristics.

Porridges prepared from various FBFs were also significantly different in sweetness (*p* ≤ 0.05). As expected, novel extruded FBFs with the addition of 15% sugar were significantly higher in sweetness than the traditional non-extruded FBF (CSB+) (*p* ≤ 0.05). The addition of sugar into the FBFs formulation was not only to provide energy, but could also to increase the palatability and consumption of the products [6]. Iuel-Brockdorf et al. [32] also found that products with a sweeter flavor received better ratings in terms of child and caregiver acceptability.

Salt was significantly different among the FBFs porridges (*p* ≤ 0.05), however, it was only a small difference (lower than 0.5 points on a 15 point scale). The higher intensity of salt in novel extruded FBFs was probably due to the higher amount of vitamin and mineral premix that had been added into the formulation. Gilbertson et al. [33] indicated that the taste system plays important roles in nutrient identification and salty taste reflects the recognition of minerals in foods. The study by Teillet et al. [34] also found that a more salty taste was found in waters with higher mineral contents.


**Table 4.** Mean scores 1 (standard error) of sensory attributes for porridges prepared from FBFs.

 =

.

 parameter

 a

 same

 were

 treatments.

 ≤

= White variety of sorghum, V3 = Red variety of sorghum, com = Commercial milling, (pre-anti) = Antioxidant had been added to the binary blend before extrusion process. 3 (a) = Aroma,(f)Flavor4Averageforeachwithdifferentletterinthecolumnsignificantlydifferent(*<sup>p</sup>*0.05)between

*Foods* **2019**, *8*, 261

Porridge prepared from binary blends with higher levels of lipids, e.g., whole corn with full-fat soybean blend (WCS"B), was significantly higher in bitterness than most of the FBFs porridges (*p* ≤ 0.05). The high temperature used in the extrusion process could have accelerated the degradation of lipids, and the degraded lipids appeared to be associated with unpleasant flavors, such as astringent, bitter, and rancid [24,35,36]. WCS"B, which had high levels of lipid, was more likely to have a higher amount of degraded lipid after the extrusion process, and this could result in the higher bitter taste of the cooked porridge.

Principal component analysis (PCA) of twelve sensory attributes helped to visualize the differences among porridge samples (Figure 1). PC1 accounted for 39% of the variation, and seemed to differentiate among samples according to beany, toasted, grain, musty, and bitter attributes. PC2 accounted for 25% of the variation, and seemed to differentiate among samples according to flavor attributes, including astringency, sweetness, and saltiness. Current non-extruded FBF (CSB+) was separated from novel extruded FBFs due to the lower intensity in sweetness, saltiness, and astringency. Extruded corn-soy blends and extruded sorghum-soy blends were grouped together and had more pronounced bitter and musty attributes. As previously mentioned, the extruded products containing higher amount of lipids were more bitter (*p* ≤ 0.05) because of the high possibility of having more degraded lipids. However, it must be noted that the lipids certainly were not degraded enough to produce marked changes in shelf-life [17]. Phenolic compounds, which can be found in sorghum, are responsible for the bitterness of many similar foods and may cause a negative effect on products' acceptability [35,37]. Therefore, the higher amount of sorghum (47.6% sorghum) used in sorghum-soy blend formulations was another reason that made those blends higher in bitter taste. This effect also was found in 20% solids FBFs made of sorghum without added sugar [38].

**Figure 1.** Principal component analysis of the porridges prepared from FBFs and sensory attributes (**a**) Score plot. (**b**) Correlation loading plot. For the FBFs, W = Whole, first S = Sorghum flour, first C = Degermed corn flour, second S = Low-fat soy flour, S' = Medium-fat soy flour, S" = Full-fat soy flour, second C = Cowpea flour, V1 and V2 = White variety of sorghum, V3 = Red variety of sorghum, com = Commercial milling, (pre-anti) = Antioxidant had been added to the binary blend before extrusion process. CSB+ represents the control sample (current non-extruded FBF).

All extruded sorghum-cowpea blends were grouped together. They were mainly characterized by toasted, grain, and beany attributes. The sorghum-cowpea binary blend that was used to make extruded sorghum-cowpea blends had lower levels of lipids compared to sorghum-soy and corn-soy binary blends [39]. Feng and Lee [40] reported that during extrusion, the lipid worked as a lubricant, and decreased the temperature in the extruder barrel. The lower amount of lipids in the sorghum-cowpea blend contributed to higher friction between the particles in the mix and the screw surface, and directly related to a higher temperature in the extruder barrel. The higher temperature during the extrusion

process could probably generate higher levels of chemical compounds from the Maillard reaction, which were responsible for desirable attributes, such as cereal-like, toasted, or roasted aromas [24,26].

#### **4. Conclusions**

The results from this study clearly identified the effects of the extrusion process and ingredients used on the sensory properties of the products. Novel FBFs from the extrusion process had more pronounced toasted characteristics due to the higher temperature during extrusion. The type of milling, decortication process, and the step of adding antioxidant to the blends did not show effects on the sensory properties of FBFs in this study. Adding sugar and increasing the amount of vitamin-mineral premix in the novel FBFs formulation increased the sweetness and saltiness of the products, respectively, as expected, which is not surprising given that caregivers have been shown to add sugar to current unsweetended FBFs. The level of lipids in binary blends was mainly responsible for the bitterness of the product. In addition, legumes, such as soybeans and cowpeas, were the main ingredient that contributed to the beany characteristics of the products. The higher amount of legume used in the formulations, the more beany characteristics that could be perceived from the products.

**Author Contributions:** Conceptualization, E.C.; Data curation, E.C. and S.C.I.; Formal analysis, S.C.I. Funding acquisition, E.C.; Investigation, S.C.I.; Methodology, E.C. and S.C.I.; Project administration, E.C.; Supervision, S.C.I.; Writing-original draft, S.C.I.; Writing-review & editing, E.C. Both authors contributed to the design, conduct, and writing of the manuscript.

**Funding:** Agricultural Research Service: FFE-621-2012/033-00.

**Acknowledgments:** This work was funded, in part, by the USDA Foreign Agricultural Service under the Micronutrient Fortified Food Aid Products Pilot (MFFAPP) program, contract number #FFE-621-2012/033-00. The authors would like to thank the staff of the Center for Sensory Analysis and Consumer Behavior at Kansas State University who assisted in conducting this study. The authors wish to acknowledge Sajid Alavi and Brian Lindshield for their work in planning the larger project and for help in selecting the products that would be tested in this study.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

#### *Article*

### **Use of Almond Skins to Improve Nutritional and Functional Properties of Biscuits: An Example of Upcycling**

#### **Antonella Pasqualone 1,\*, Barbara Laddomada 2, Fatma Boukid 3, Davide De Angelis <sup>1</sup> and Carmine Summo <sup>1</sup>**


Received: 23 October 2020; Accepted: 16 November 2020; Published: 20 November 2020 -

**Abstract:** Upcycling food industry by-products has become a topic of interest within the framework of the circular economy, to minimize environmental impact and the waste of resources. This research aimed at verifying the effectiveness of using almond skins, a by-product of the confectionery industry, in the preparation of functional biscuits with improved nutritional properties. Almond skins were added at 10 g/100 g (AS10) and 20 g/100 g (AS20) to a wheat flour basis. The protein content was not influenced, whereas lipids and dietary fiber significantly increased (*p* < 0.05), the latter meeting the requirements for applying "source of fiber" and "high in fiber" claims to AS10 and AS20 biscuits, respectively. The addition of almond skins altered biscuit color, lowering *L\** and *b\** and increasing *a\**, but improved friability. The biscuits showed sensory differences in color, odor and textural descriptors. The total sum of single phenolic compounds, determined by HPLC, was higher (*p* < 0.05) in AS10 (97.84 μg/g) and AS20 (132.18 μg/g) than in control (73.97 μg/g). The antioxidant activity showed the same trend as the phenolic. The *p*-hydroxy benzoic and protocatechuic acids showed the largest increase. The suggested strategy is a practical example of upcycling when preparing a health-oriented food product.

**Keywords:** almond skins; by-product; upcycling; biscuits; health claims; fiber; nutritional composition; sensory properties; phenolic compounds

#### **1. Introduction**

Recently, the reuse of food industry by-products has become a particularly important research topic, in order to develop systems capable of minimizing environmental impact and the waste of resources. The confectionery industry, in the production of blanched almonds, generates large quantities of almond skins as a by-product, which are mostly destined to cattle feeding [1] and composting [2]. However, almond skins can be considered functional food ingredients because they contain several bioactive phenolic compounds, namely flavonoids, phenolic acids, and tannins, the latter both hydrolysable and condensed [3–7]. The phenolic content of fresh almond skins comprises between 11.1 and 17.7 mg/g, depending on the extraction protocol [7], whereas 0.25–0.85 mg/g d.m. (dry matter) were quantified in dried almond skins, with the lowest amount in sun-dried skins and the highest in skins oven-dried at a temperature of 45–60 ◦C [7].

The polyphenols of almond skins are bioavailable and possess in vitro and in vivo antioxidant activity, able to reduce plasmatic oxidative stress [8] and to protect LDL (low-density lipoprotein) from oxidation [4,9]. The bioactive compounds of almond skins display also antibacterial and antiviral effects [10,11]. Recently, an extract of almond skins has been proposed for use in intestinal inflammatory diseases [12]. Furthermore, almond skins are also a rich source of fiber and therefore have a prebiotic effect, favorably influencing the gut microbiome [13,14]. The recommended daily intake of fiber ranges from 18 g to 38 g for adults and it varies among different countries, but many people do not reach this threshold [15]. Almond skins could hence be used to functionalize foods and to improve their nutritional profile in terms of fiber content. The reuse of almond skins in food products would represent an example of upcycling [16], responding to the need to increase sustainability in the food industries within the framework of the principles of a circular economy [17].

Functional ingredients, such as almond skins, could be easily added to cereal-based products, but any modification of the physico-chemical and sensory characteristics of the end-products should be carefully evaluated so as to fulfill consumer expectations for healthy but pleasant foods. The potential use of almond skins in composite dough with wheat flour was evaluated in a previous study, highlighting significant alterations of alveograph and farinograph indices due to the presence of fibers, which interfere with the gluten network [7]. Therefore, almond skins could be used in those cereal-based products which better tolerate a weak gluten network, such as biscuits.

Biscuits are popular baked goods, eaten daily and characterized by a long shelf-life. These features make biscuits a good recipient for the addition of functional ingredients. To date, however, almond skins are still an underexploited resource and no study has considered their introduction in biscuit formulation, despite many researchers having reformulated biscuits by incorporating an array of new ingredients, mostly of vegetable origin, such as apple peel powder [18], acorn flour [19], grape marc extract [20,21], purple wheat flour [22], inulin [23], soy protein isolate [24], blue berry by-product [25], and green tea extract [26].

Within this framework, the aim of this research has been to verify the effectiveness of almond skin addition in the preparation of functional biscuits with improved nutritional properties.

#### **2. Materials and Methods**

#### *2.1. Raw Materials*

The ingredients used for preparing the experimental biscuits were: refined wheat flour (0.52 g/100 g ashes) (Molini Spigadoro, Bastia Umbra, Italy), sucrose (Eridania, Bologna, Italy), extra virgin olive oil (Olearia De Santis, Bitonto, Italy), baking powder (sodium bicarbonate and potassium bitartrate, 'Belbake', Lidl Stiftung & Co. KG, Neckarsulm, Germany), all purchased at local retailers, and almond skins. The latter were collected from an almond processing industry (Calafiore S.r.l., Floridia, Italy), then dried at 60 ◦C for 30 min by a rotary air drier (mod. Scirocco, Società Italiana Essiccatoi, Milano, Italy), milled (Cutting Mill SM 100, Retsch, Haan, Germany) and sieved on a sieve with 0.6 mm holes. Moisture, aw, phenolic compounds, antioxidant activity, color, and odor notes of almond skins are reported in a previous paper [7].

#### *2.2. Preparation of Biscuits*

The formulation of biscuits is reported in Table 1. Two levels of addition of almond skins were considered: 10 g/100 g (AS10) and 20 g/100 g (AS20) on a wheat flour basis, which were compared with control biscuits prepared without adding almond skins. The amount of water was defined in preliminary trials in order to achieve the same dough workability in the three types of biscuits. The process consisted in: kneading for 3 min sucrose, extra virgin olive oil and baking powder by an electric mixer with flat beater (Kitchen Aid, Antwerp, Belgium), then adding flour (pure wheat flour or blended with almond skin powder as in Table 1) and kneading for 3 min, finally adding water and kneading for about 10 min to form a homogeneous dough. The dough was then rolled out

with a rolling pin to a thickness of 4 mm and cut into 6 cm diameter disks with the aid of a circular biscuit cutter with scalloped edges. The disks of dough were placed on a baking tray, mixing them in a randomized block pattern to minimize any effect of tray location during baking, then were baked in an electric oven (mod. Ignis ACF961IX, Whirlpool Italia S.r.l., Pero, Italy) at 175 ◦C for 15 min. Two independent production trials were carried out. Biscuits were finely crushed for analysis, except for the textural, colorimetric and sensory analyses.

**Table 1.** Formulation of the experimental biscuits (per 100 g of flour). Control = Biscuits without Almond Skins; AS10 and AS20 = Biscuits prepared by adding 10 g and 20 g Almond Skin Powder per 100 g of Wheat Flour, respectively.


#### *2.3. Determination of Nutritional Composition*

Protein (N × 5.7) and moisture content were determined according to the American Association of Cereal Chemists (AACC) Methods 46–11.02 and 08–01, respectively [27]. The lipid fraction was extracted according to ICC Standard Method no. 136 [28]. Total dietary fiber was determined by the enzymatic-gravimetric procedure according to the AOAC Official Method 991.43 [29]. Carbohydrates were calculated by difference: 100 – (moisture + proteins + lipids + fiber + ash). Energy value (kJ), calculated by using the Atwater general conversion factors, also considered the contribution of 8 kJ/g from total dietary fiber, according to Annex XIV of Regulation (EC) No 1169/2011 [30]. All analyses were carried out in triplicate.

#### *2.4. Determination of Physical Properties*

The *a\** (red/green balance), *b\** (yellow/blue balance), and *L\** (lightness) coordinates of the CIELAB color space were determined by a colorimeter (CM-600d Chromameter, Konica Minolta, Tokyo, Japan) under illuminant D65. Five replicated analyses were carried out. Total color difference (ΔE) was calculated as follows [31]:

$$
\Delta \mathbf{E} = \left[ \left( \Delta L^\* \right)^2 + \left( \Delta a^\* \right)^2 + \left( \Delta b^\* \right)^2 \right]^{1/2}
$$

The following scale was considered: ΔE = 0–0.5, very low difference; 0.5–1.5; slight difference; 1.5–3.0, noticeable difference; 3.0–6.0, appreciable difference; 6.0–12.0, large difference; and >12.0, very obvious difference [32].

Water activity (aw) was analyzed in triplicate by a water activity meter (mod. Aqualab 4TE, Meter group, Pullman, WA, USA).

Textural properties, in terms of breaking strength (N mm<sup>−</sup>2), were determined by a three-point bending test ("snap test") using a ZI.0 TN texture analyzer (ZwickRoell GmbH & Co. KG, Ulm, Germany), equipped with 1 kN load-cell. The biscuits were placed on the analyzer supports with their top surface down. The distance between the support bars was 4 cm. The downward movement of the probe, set at a speed of 5 mm min<sup>−</sup>1, was continued until the biscuit was broken. Eight replicated analyses were carried out.

#### *2.5. Baking Induced Variations of Dimensional Parameters and Weight*

The weight (W) of biscuits before and after baking was assessed by a balance (Gibertini, Novate Milanese, Italy). The diameter (D) and thickness (T) of biscuits before and after baking

were determined by a caliper. The spread factor was calculated as the ratio between D and T of baked biscuits, according to the AACC Method 10-50.05 [27]. The percentage variations in W, D, and T were calculated as follows:

% variation of W (or D, T) = (W (or D, T) after baking—W (or D, T) before baking)/W (or D, T) before baking × 100. Six replicated analyses were carried out.

#### *2.6. HPLC analysis of Phenolic Compounds*

The phenolic compounds were extracted from 1 g biscuits according to the procedure reported in Laddomada et al. [33], which involved defatting, alkaline hydrolysis, acidification and double ethyl acetate extraction. The extracts were lyophilized and dissolved in 400 μL of a solution of methanol diluted with 200 mL/L distilled water, then 50 μL were filtered on 0.45 μm polytetrafluoroethylene (PTFE) filters (Teknokroma, Barcelona, Spain) and analyzed by HPLC-DAD (Agilent 1100 Series, Agilent Technologies, Santa Clara, CA, USA) with a reversed phase C18(2) Luna column (Phenomenex, Torrance, CA, USA) (5 μm, 250 × 4.6 mm), as in Pasqualone et al. [7]. Identification of peaks was made by comparison of their UV-Vis spectra, and their retention times to those of authentic phenolic standards. Phenolic acids were quantified via a ratio of 3,5-dichloro-4-hydroxybenzoic acid, used as internal standard, and calibration curves of phenolic acid standards. Other phenolics (flavan-3-ols, flavonol and flavonone glycosides and aglycones) were quantified using calibration curves according to the external standard method [6]. The linear range, correlation coefficient, limit of detection (LOD) and limit of quantification (LOQ) for the phenolic compounds quantified are reported in Table S1.

#### *2.7. Determination of Antioxidant Activity*

An amount of1gsample, mixed with 10 mL of methanol and shaken at 250 rpm for 2 h in the dark, was centrifuged for 5 min at 5000 × g. The supernatant was submitted to the assessment of the antioxidant activity by the 2,2-diphenyl-1-picrylhydrazyl (DPPH) radical scavenging capacity assay, as in Pasqualone et al. [22]. A calibration curve was prepared with 0.1–100 μM solutions of 6-hydroxy-2,5,7,8-tetramethylchroman-2-carboxylic acid (Trolox) (Sigma–Aldrich Chemical Co., St. Louis, MO, USA) (y = −0.008x + 0.6087; *R2* = 0.9971).

#### *2.8. Determination of Sensory Properties*

Quantitative Descriptive Analysis (QDA) of biscuits was performed by a trained sensory panel of eight people, following the ethical guidelines of the laboratory of Food Science and Technology of the Department of Soil, Plant and Food Science (DISSPA), Dept. of Bari University (Italy). Panelists, regular consumers of biscuits and almonds and free of food intolerances or allergies, were informed about the study aims, and signed an individual written informed consent. Pre-test sessions were carried out, as in Pasqualone et al. [34]. Eight sensory descriptors, defined in Table 2, were rated on a 0–9 score range (0 = minimum; 9 = maximum intensity). The analyses were carried out in triplicate.

#### *2.9. Statistical Analysis*

One-way analysis of variance (ANOVA) followed by Tukey's HSD test, was made using the XLSTAT software (Addinsoft SARL, New York, NY, USA).


#### **Table 2.** Descriptive terms used for the sensory profiling of biscuits.

#### **3. Results and Discussion**

#### *3.1. Nutritional and Technological Characteristics*

Almond skin powder is particularly rich in fiber (52.6 g/100 g), as shown by the analysis of its nutritional characteristics (Table 3).

**Table 3.** Nutritional composition of dried almond skin powder and wheat flour used in the preparation of experimental biscuits. Values per 100 g, expressed on fresh weight basis.


This by-product of almond processing also showed a relevant presence of lipids (21.3 g/100 g). The lipid fraction of almond skins, however, is particularly healthy, being composed mainly of mono and polyunsaturated fatty acids (mostly oleic and linoleic acids) [6] associated with high amounts of vitamin E [6]. The composition of the lipid fraction of skins parallels the lipid composition of the whole seed [35]. The protein content of almond skins accounted for about 11 g/100 g, and low amounts of carbohydrates were observed. The overall composition of almond skin powder agreed with the current literature [6,36]. The composition of wheat flour was quite different than that of almond skins, being rich in carbohydrates and poor in fiber, with negligible levels of lipids.

The analysis of the nutritional features of biscuits (Table 4) shows that the protein content was not significantly influenced by the addition of almond skins, the latter having a protein content similar to wheat flour. However, AS20 biscuits had a significantly (*p* < 0.05) higher lipid content than control, due to the relevant contribution of almond skins. The lipid content of all biscuits was in the range of those commonly marketed [37].


**Table 4.** Nutritional features (values per 100 g, expressed on fresh weight basis) of biscuits enriched by increasing levels of almond skins. Control = biscuits without almond skins; AS10 and AS20 = biscuits prepared by adding 10 g and 20 g of almond skin powder per 100 g of wheat flour, respectively.

Different letters in row indicate significant differences (*p* < 0.05).

As for the content of dietary fiber, it progressively increased with the increase of almond skin addition. EC Regulation n. 1924/2006 [38], relating to nutrition and health claims made on food products, defines that a food is a "source of fiber" only if contains at least 3 g/100 g fiber, or at least 1.5 g/100 kcal fiber, while "high in fiber" applies only if a food contains at least 6 g/ 100 g fiber, or at least 3 g/100 kcal fiber. The level of fiber ascertained in AS10 and AS20 biscuits met the requirements for applying the "source of fiber" and the "high in fiber" claims, respectively.

Moisture content increased, but not significantly, after the addition of almond skins due to their contribution of fiber. The higher the protein and fiber content, the higher the water absorption by the dough and moisture retention are found of the final product [39].

As a consequence of the increase in fats and fiber, the level of carbohydrates significantly decreased in almond skin-added biscuits compared to control. The energy value did not vary significantly by adding almond skins, because the increase of lipids was compensated for by an increase of fiber and a decrease in carbohydrates.

As for the main physical characteristics (Table 5), the aw of AS10 and AS20 was slightly higher than control, but without a significant difference. The aw values observed in all biscuits agreed with moisture content and showed that they were conveniently dry and stable from the microbiological point of view (aw < 0.6).


**Table 5.** Physical characteristics of biscuits enriched by increasing levels of almond skins. Control = biscuits without almond skins; AS10 and AS20 = biscuits prepared by adding 10 g and 20 g of almond skin powder per 100 g of wheat flour, respectively.

Different letters in row indicate significant differences (*p* < 0.05).

The addition of almond skins, which were brown colored, resulted in an expected substantial alteration of biscuit color (Figure 1), with a significant decrease of *L\** and *b\**, and an increase of *a\** in AS10 and AS20 compared to the control (Table 5). The total color difference (ΔE) of AS10 and AS20 biscuits compared to the control was greater for AS20 than for AS10, but in both cases with very high values, confirming that the control had a distinct color [40]. ΔE values >12.0, in fact, indicate a very obvious color difference [32].

**Figure 1.** Biscuits enriched by increasing levels of almond skins. From left to right: Control = biscuits prepared without adding almond skins; biscuits prepared by adding 10 g of almond skin powder per 100 g of wheat flour (AS10); biscuits prepared by adding 20 g of almond skin powder per 100 g of wheat flour (AS20).

The textural analysis showed that the addition of almond skins caused a decrease in the strength necessary to break the biscuits, i.e., an increase of friability, which is a particularly important characteristic. Friability is a salient textural characteristic for biscuits [41,42]. Tough, non-crumbly biscuits have low acceptance values in consumer tests [43]. This variation of breaking strength was significant when comparing control with AS20 and was due to the high presence of fiber in the almond skins. Fibers are highly hygroscopic and interfere with the formation of a strong and complete gluten network [44]. Preliminary work, in fact, showed that the rheological properties of the dough [7] significantly worsened after the addition of almond skin powder. However, among baked goods, biscuits are the most suitable for being reformulated with the addition of fibrous raw materials, since for their production a weak gluten network is not only sufficient but even necessary. In addition, although the difference in lipid content with control was significant only for AS20, the lipid fraction deriving from almond skins could have positively influenced the friability [45]. Therefore, the addition of almond skins did not harden biscuits at all; on the contrary, it gave a crumblier texture.

As for the dimensional variations induced by baking (Table 6), due to the thermal expansion of gases (carbon dioxide developed by the baking powder, dough moisture, and air entrapped during kneading), all biscuits increased more in thickness than in diameter. This result, commonly observed in biscuit baking [19], is due to the retaining effect of gluten, which tends to limit enlargement, whereas the upward thrust of the oven heat (oven rise) is less opposed [45]. AS10 and AS20 showed a greater diameter increase than control, which was significantly different for AS20, but had a lower increase in thickness. The easier enlargement observed in almond-skin added biscuits was due to the coupled effect of the dilution of gluten by a non-gluten raw material and the interference with gluten formation by the fiber and lipids of the same material. These findings agreed with studies where other fibrous and gluten-free ingredients were added to biscuit dough [18,19]. In addition, better expanded biscuits are usually less compact and more friable than those which expand less, in agreement with the observed textural data.

**Table 6.** Baking induced variations of dimensional parameters of biscuits enriched by increasing levels of almond skins. Control = biscuits without almond skins; AS10 and AS20 = biscuits prepared by adding 10 g and 20 g of almond skin powder per 100 g of wheat flour, respectively.


Different letters in row indicate significant differences (*p* < 0.05).

The spread factor increased progressively as the amount of almond skins increased, with a significant difference between control and AS20. A higher spread factor indicates a better quality and is linked to an increase in consumer acceptability [46]. The observed values were higher than those reported for biscuits enriched with pure fiber of various cereals [47].

Weight loss, primarily due to the moisture loss from dough during baking, decreased by increasing the amount of almond skins as a consequence of the greater hygroscopicity of fibers, which limited water migration. The values ascertained were in the range of other researches [48–50].

The sensory profiles of the biscuits showed significant differences in odor, color and textural descriptors (Table 7). As for taste, the bitter note was negligible in the biscuits investigated, while sweetness was moderately intense, both without significant difference among formulations.


**Table 7.** Sensory properties of biscuits enriched by increasing levels of almond skins. Control = biscuits without almond skins; AS10 and AS20 = biscuits prepared by adding 10 g and 20 g of almond skin powder per 100 g of wheat flour, respectively.

Different letters in row indicate significant differences (*p* < 0.05).

A slight odor note of caramel, derived from sugar caramelization and Maillard reaction, was perceived by the panelists in all biscuit types, without statistically significant differences between them. Instead, differences between the samples were found in the intensity of leafy odor. This odor note, absent in the control, was perceived with low intensity in biscuits formulated with almond skins, with the highest perception in AS20 and with an intermediate value in AS10. In previous research [7] this characteristic smell note was observed in the dried almond skins used in biscuit-making, albeit much more pronounced than in the finished product.

The color of biscuits became progressively and significantly darker as the level of addition of almond skin powder increased, as already indicated by colorimeter determinations.

As for friability, evaluated as the way biscuit fractured when broken by finger, the sensorial results were similar to those obtained instrumentally by the texture analyzer (snap test). AS20 was significantly more friable than control.

Dryness and graininess, on the other hand, were evaluated during chewing. Dryness did not show significant differences, whereas graininess was scored higher in almond-skin added biscuits, due to their granular and fibrous crumbles.

#### *3.2. Functional Characteristics*

Almond skins are rich in phenolic compounds [7], therefore the content of these bio-actives was evaluated in biscuits, as well as antioxidant activity (Table 8). The total sum of phenolic compounds, determined by HPLC, was significantly higher in AS10 and AS20 than in control.


**Table 8.** Phenolic compounds and antioxidant activity of biscuits enriched by increasing levels of almond skins. Control = biscuits without almond skins; AS10 and AS20 = biscuits prepared by adding 10 g and 20 g almond skin powder per 100 g of wheat flour, respectively.

AA = antioxidant activity; TE = Trolox equivalents; DPPH = 2,2-diphenyl-1-picrylhydrazyl radical. Different letters in row indicate significant differences (*p* < 0.05).

In more detail, the variation of the single phenolics did not show the same trend for all the compounds, which showed different behavior according to the phenolic composition of the raw materials. In particular, among the phenolic acids, the *p*-hydroxy benzoic and protocatechuic acids showed a relevant increase after the addition of almond skins. The flavan-3-ols catechin and epicatechin also followed the same trend, being not detectable in control and showing a concentration-effect increment between the AS10 and AS20. In fact, these phenolic compounds are the most abundant in almond skins [7]. A smaller increase, but always statistically significant, was observed for syringic acid, vanillic and *p*-coumaric acids.

Instead, the most abundant phenolic acid, namely the ferulic acid, followed by the sinapic acid, decreased when comparing control biscuits with the almond-skin added, because these phenolic acids are typically present in wheat [33,51], but not in almond.

The flavonol glycosides and their aglycones, as well as the flavanone glycosides and their aglycones, despite their presence in almond skins [7], were not detected in biscuits. Probably, since their starting quantity was not remarkably high, they became undetectable in the biscuits, due to the dilution effect of wheat flour. In addition, oxidation and other degradation phenomena could not be excluded during processing (kneading and baking) since a decrease in phenolic compounds had already been observed when raw almond skins were thermally dried [7]. In any case, the total phenolic compounds of AS20 were approximately double that of the control, indicating that the addition of almond skins in the formulation can concretely contribute to enhance the nutritional value and the potential health benefits of the end products.

The antioxidant activity followed the same trend as the phenolic and showed higher values in the almond skin supplemented biscuits, compared to the control, also highlighting a concentration effect. Indeed, in the AS20, the antioxidant activity was about five times higher than the control. The observed values of antioxidant activity were consistent with those of the almond skins added [7].

#### **4. Conclusions**

The increasing sensibility of modern consumers towards the potential benefits of food on human health has led to a strong demand for functional products.

To date, almond skins, in spite of having high fiber content and antioxidant substances, are a by-product of almond processing usually addressed to animal feed and/or composting. This study, instead, demonstrates that almond skins can be effectively used for the production of functional biscuits, addressing the needs of both producers, who require the reduction of waste production,

and consumers, who increasingly demand healthier food. For this latter purpose, the nutritional claims "source of fiber" and "high in fiber", defined in EC Regulation n. 1924/2006, were applicable to the AS10 and AS20 biscuits, respectively.

Therefore, using almond skins in biscuit-making is a feasible way to convert a low-value by-product into a valuable resource, providing to the almond processing industry an efficient and environment-friendly solution for waste disposal. This is a practical example of upcycling while preparing a health-oriented food product.

**Supplementary Materials:** The following is available online at http://www.mdpi.com/2304-8158/9/11/1705/s1, Table S1: Regression equation, linear range, correlation coefficient (*R*2), limit of detection (LOD), and limit of quantification (LOQ) of the HPLC analysis of phenolic compounds quantified in the experimental biscuits.

**Author Contributions:** Conceptualization, A.P.; Formal analysis, A.P., B.L., C.S.; Funding acquisition, C.S.; Methodology, A.P., B.L., D.D.A.; Writing—original draft, A.P.; Writing—review & editing, A.P., B.L., F.B., D.D.A. and C.S. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the Regione Puglia, Project "Almond Management Innovations – AMI", within the call "PSR 2014-2020, MIS 12, Cooperazione. Sottomisura 16.2: Sostegno a progetti pilota e allo sviluppo di nuovi prodotti, pratiche, processi e tecnologie".

**Acknowledgments:** The authors acknowledge Virgilio Giannone for helpful discussions and for kindly furnishing the samples of almond skins. The authors acknowledge also Leone D'Amico for skillful assistance in HPLC analyses.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Article*

### **Evaluation of the Use of a Co**ff**ee Industry By-Product in a Cereal-Based Extruded Food Product**

### **Elisa A. Beltrán-Medina 1, Guadalupe M. Guatemala-Morales 1, Eduardo Padilla-Camberos 1,\*, Rosa I. Corona-González 2, Pedro M. Mondragón-Cortez <sup>1</sup> and Enrique Arriola-Guevara 2,\***


Received: 23 June 2020; Accepted: 23 July 2020; Published: 27 July 2020

**Abstract:** The evaluation of by-products to be added to food products is complex, as the residues must be analyzed to demonstrate their potential use as safe foods, as well as to propose the appropriate process and product for recycling. Since coffee is a very popular beverage worldwide, the coffee industry is responsible for generating large amounts of by-products, which include the coffee silverskin (CS), the only by-product of the roasting process. In this work, its characterization and food safety were evaluated by chemical composition assays, microbiological determinations, aflatoxin measurements and acute toxicity tests. The results showed that CS is safe for use in food, in addition to providing dietary fiber, protein and bioactive compounds. An extruded cereal-based ready-to-eat food product was developed through an extreme vertices mixture design, producing an extruded food product being a source of protein and with a high fiber content. Up to 15% of CS was incorporated in the extruded product. This work contributes to the establishment of routes for the valorization of CS; nevertheless, further research is necessary to demonstrate the sustainability of this food industry by-product.

**Keywords:** coffee silverskin; chemical characterization; toxicological analysis; extrusion; extreme vertices mixture design; product development

#### **1. Introduction**

Today, there is a considerable emphasis on the recovery, recycling and upgrading of wastes, particularly in the food processing industry, in which wastes, effluents, residues and by-products can be recovered. They can often be upgraded into useful products and value-added food supplements that can provide dietary fiber and bioactive compounds [1,2]. The possibility of the utilization of these food processing by-products for manufacturing various human foods has created enormous scope for waste reduction, indirect income generation, the reduction of raw material costs [1,3] and even the potential for them to be considered as novel foods with beneficial properties [4]. However, for the development of future sustainable industrial processes centered on the valorization of food waste, aspects such as technical feasibility, an analysis of their-economic potential and a life-cycle-based environmental assessment need to be considered [5].

Coffee is one of the most consumed beverages in the world and is the second largest traded commodity after petroleum [6]. The coffee production chain begins with the harvest of the ripe coffee berries that are to be treated in order to separate the pulp from the coffee bean by one of two processes—(a) a wet process or (b) a dry process—where the green coffee bean is obtained. Finally, the bean is heat treated by a process called roasting, thus producing the coffee that will be used for the preparation of the drink [4,7]. In the world, during the 2018/2019 season, 10.3 million tons of green coffee was produced [8]. Since coffee is a very popular and appreciated beverage around the world, the coffee industry is responsible for generating large amounts of wastes, which include the coffee silverskin (CS), the only waste obtained during the roasting process [7,9]. The CS represents about 4.2% (w/w) of the coffee beans [9]. Despite the produced quantity being low compared to that of other coffee by-products, it has been reported that for 120 tons of roasted coffee, about 1 ton of CS is produced [10]. It can be considered that if all the green coffee produced worldwide during the 2018/19 season had been roasted, it would be equivalent to having produced around 71,822 tons of CS (conversion factor: 1.19 tons of green coffee = 1 ton of roasted coffee [8]). CS is a yellowish transparent endosperm that covers each green coffee bean (Figure 1) [4,7,9] and is currently used as a fuel and fertilizer [11]. However, coffee wastes have been reported to possess bioactive compounds, mainly secondary metabolites such as phenolic acids, for example, hydroxycinnamic acids and flavonoids, desired for their beneficial antioxidant properties [12,13]. 5-caffeoylquinic acid (5CQA) belongs to the family of the chlorogenic acids (hydroxycinnamic acids). It is one of the most abundant polyphenolic compounds in the human diet and is produced by certain plant species; it is an important component in coffee and in the CS [14,15]. 5CQA is of special interest due to the wide spectrum of its potentially beneficial effects on health, including antidiabetic, anti-obesity, antioxidant, anti-hypertension, anti-inflammatory and antibacterial effects [16,17].

**Figure 1.** Coffee silverskin from coffee roasting process.

CS has been reported as a source of chlorogenic acids; however, to date, there are few reports concerning the content of 5CQA in CS, and those that exist show controversial results, since the reported concentrations are in the range of 1000 to 11,678 mg of 5CQA/kg of CS [11,18–20]. Different studies have shown the functional properties of CS such as a high dietary fiber content (54.11 to 74.15 g/100 g of CS) [9,21,22] and a total phenolic content in the range of 4.6 to 46.65 mg/g, depending on the extraction method employed [11,21,23,24]. The principal constituents of its fibrous tissues are cellulose (24%) and hemicellulose (17%). It is a source of minerals such as potassium (21,100 mg/kg dry basis (db)), iron (843 mg/kg db), sodium (57 mg/kg db), manganese (50 mg/kg db) and zinc (22 mg/kg db), among others [9]. The enzyme inhibitory properties of CS extracts and peptide composition of CS protein hydrolysates have been investigated, from the perspective of their application in the pharmaceutical and nutraceutical industry [24,25].

The holistic concept of food production tries to connect differing goals, such as the highest product quality and safety, highest production efficiency and the integration of environmental aspects into product development and food production. Vegetable residues mostly contain considerable amounts of potentially interesting compounds [1]. However, the benefits of recycling should not be undermined by the environmental impacts caused by new production processes [26]. Food extrusion is a versatile process in food engineering as it combines various unit operations such as transport, thermomechanical and degradation changes, mixing and molding. It is a technology widely used in the food industry due to its versatility, high productivity and energy efficiency [27,28]. Extrusion cooking is increasingly used in the food industries for the development of new cereal-based snacks, baby foods and breakfast cereals [29]. The incorporation of by-products from different fruit and vegetable processing industries into extruded products has led to hope for their utilization as well as the development of nutritionally healthy extruded products [30]. The extrusion process has been used to develop new products in which 2% to 20% of various by-products of the agri-food industry have been incorporated, such as barley-fruit and cauliflower by-products and red lentil-carrot pomace, among others [30–33]. The by-product incorporation in extruded food has been reported at the lab scale, and no industrial-level studies have been shown. Nevertheless, extrusion is a mature and scalable technology; even scale-up considerations and mathematical models for extrusion cooking are available [34,35].

Cereal-based food products have been the basis of the human diet since ancient times. Cereals contain all the macronutrients (protein, fats and carbohydrates) we need for support and maintenance [36]. They contain only low levels of micronutrients, most of which are lost during processing for food [37], bringing the possibility to incorporate new raw materials that provide these micronutrients. To the best of our knowledge, there are few reports on the use of CS in a cereal-based product. A treatment of CS with alkaline hydrogen peroxide before being added to Barbari bread to improve its properties has been described [38]. In another study, CS was added to cake as a fat replacer [39]. Some authors investigated the use of CS in biscuits, where it was incorporated as a sugar replacer or to enhance the phenolic content and antioxidant capacity of the product, employing a standardized formulation [40,41]. However, no studies regarding the formulation design under official standards to create a product with specific requirements have been reported. Hence, the development of a cereal-based food product adding CS using extrusion technology is proposed.

In all those articles in which CS was added in cereal-based products, wheat flour was used as the cereal basis [38–41]; nonetheless, in present work, corn and popped amaranth were chosen as the cereal product base. Corn is undoubtedly part of the identity of Mexico; it is present in the daily life of its inhabitants [42], which will allow the obtaining of a familiar flavor and better acceptance of the developed product. Amaranth is a popular snack in Mexico of pre-Hispanic origin [42] and is characterized by an excellent nutritional composition (*A. hypochondriacus,* protein, 15.9%; lysine, 4.9 g/100 g of protein; fat, 6.1%; tocopherols, 5.5 mg/100 g; starch, 62.4%; sucrose, 1.4%; ash, 3.3% [43]); nevertheless, it is difficult to produce expanded products directly by the extrusion cooking of amaranth grain alone because of its high fat content. Therefore, the extrusion cooking of amaranth flour in combination with other cereals produces well-accepted forms of expanded extrudates [43]. The combination of these cereals with the CS could allow the obtaining of a product with good nutritional quality.

Therefore, as CS appears to be a potential new low-cost ingredient, the aim of this work was to evaluate whether its consumption was harmless to humans, demonstrating the food safety of CS by microbiological tests and the determination of aflatoxins and the Lethal Dose (LD50) by the acute oral toxicity test. Its potential use as a food ingredient was evaluated by determining its nutritional contribution and by developing a cereal-based extruded food product with this new ingredient added.

#### **2. Materials and Methods**

#### *2.1. Materials*

CS produced by roasting coffee beans (*C. Arabica* 100%) was obtained from two states of Mexico (Chiapas and Jalisco). Popped amaranth was purchased from Nutriactivate Company (Puebla, Puebla, Mexico), and white corn was obtained from the food market of Guadalajara city (Jalisco, Mexico). CS, popped amaranth and white corn were milled prior to the extraction and extrusion process (Average Particle Size = 0.28 ± 0.01 mm).

5-caffeoylquinic acid powder reference standard (USP 12601), 2,2-diphenyl-1-picrylhydrazil (DPPH), 2,2 -azino-bis(3-ethylbenzthiazoline-6-sulphonic acid) (ABTS) and 6-hydroxy-2,5,7,8 tetramethyl chroman-2-carboxylic acid (Trolox) were purchased from Sigma-Aldrich (St. Louis, MO, USA). Methanol (MeOH) was obtained at HPLC grade (Sigma-Aldrich, St. Louis, MO, USA). Phosphoric acid was obtained at reagent grade (Karal, Leon, Guanajuato, Mexico).

The standard of 5CQA was diluted in MeOH to obtain a stock solution at 1000 μg/mL, from which the calibration curve was prepared. All the solutions remained refrigerated at 4 ◦C in amber vials.

#### *2.2. CS Characterization and Microbiological Quality*

#### 2.2.1. Bromatological and Microbiological Analysis

The bromatological analyses were performed according to the following Mexican Regulations: moisture, NMX-F-083-1986 [44]; protein, NMX-F-608-NORMEX-2011 [45]; ashes, NMX-F-607-NORMEX-2013 [46]; fats (ethereal extract), NOM-086-SSA1-1994 (Regulatory Appendix C, Number 1) [47]; and, carbohydrates, method 986.25 A.O.A.C. Volume 1 [48].

The microbiological analysis was performed according the Official Mexican Regulations: aerobic mesophilic bacteria, NOM-092-SSA1-1994 [49]; total coliforms, NOM-113-SSA1-1994 [50]; molds and yeasts, NOM-111-SSA1-1994 [51]; *Salmonella*, NOM-114-SSA1-1994 [52]; *Escherichia coli*, CCAYAC-M-004 [53]; and *Staphylococcus aureus*, NOM-115-SSA1-1994 [54].

#### 2.2.2. Total Dietary Fiber (TDF)

The TDF was estimated by the enzymatic gravimetric method according to the Mexican Regulation NMX-F-622-NORMEX-2008 [55]. Briefly, one gram of sample suspended in phosphate buffer solution was sequentially digested by heat stable α-amylase for 30 min in a boiling water bath, after which a 0.275 M NaOH solution and protease were added and incubated for 30 min at 60 ◦C. Then, a 0.325 M HCl solution and amyloglucosidase were added and incubated for 30 min at 60 ◦C. After filtration, the insoluble dietary fiber was recovered from enzyme digestate, dried at room temperature and then weighed. Soluble dietary fiber in the filtrate was precipitated with ethanol and filtered. The precipitate was dried and weighed. Insoluble and soluble dietary fiber contents were corrected for residual protein and ash content. The TDF content was the sum of both fibers.

#### 2.2.3. Extraction Method

This extraction method was used for DPPH, ABTS, total polyphenol and HPLC analysis. The extraction method was adapted from Del Río et al. (2014) [56]; 0.5 g of sample (CS or extruded product) was weighed, and 5 mL of MeOH/water 3:1 (v/v) was added. The mixture was sonicated (Branson 5800, Dansbury, CT, USA) at a 40 kHz frequency for 15 min, removed and stirred in a Vortex-Genie (Scientific Industries, Bohemia, NY, USA) for another 15 min; afterwards, it was centrifuged at 3400 rpm for 10 min. The supernatant was transferred to another container, and the residue was re-extracted. The second extract was added to the first, and it was filtered (0.45 μm). All the extracts were kept in amber vials, under refrigeration, until analysis.

#### 2.2.4. Total Polyphenol Determination

The quantification of total polyphenols was carried out by the Folin–Ciocalteu method proposed by Singleton and Rossi (1965) [57]. The extracts (30 μL) were mixed with 150 μL of Folin–Ciocalteu reagent (1:10), followed by the addition of 120 μL of 20% (w/v) sodium carbonate. After 1 h, the absorbance at 760 nm was read in the spectrophotometer. The results are expressed as g gallic acid equivalents (GAE)/100 g sample.

#### 2.2.5. Antioxidant Activity

The antioxidant activity of the extracts was determined by two methods: the ABTS and DPPH (free radical scavenging) assays. The ABTS assay was based on a method developed by Nenadis et al. (2004) [58]. A solution of 7 mM ABTS, 2.5 mM potassium persulfate and 10 mL of distilled water were mixed and incubated in the dark at room temperature for 16 h before use. This solution was diluted with MeOH to an absorbance of 0.7 ± 0.02 at 734 nm. After the addition of 20 μL of extract or Trolox standard to 200 μL of diluted ABTS solution, the absorbances were recorded at 6 min after mixing. Methanolic solutions of known Trolox concentrations were used for calibration. The results are expressed as mg Trolox equivalents (eq)/g sample.

The DPPH antioxidant activity assay was performed by the Brand-Williams et al. (1995) [59] method with slight modifications. A MeOH solution containing 500 μmol of DPPH was prepared. After adjusting the blank with MeOH, an aliquot of 20 μL of extract was added to 200 μL of this solution. After 30 min in the dark, the absorbance at 515 nm was read with the spectrophotometer. The results are expressed as mg Trolox eq/g sample.

#### 2.2.6. HPLC Analysis

Sample analysis was performed on a liquid chromatograph Alliance 2695, equipped with a 2998 Diode Array Detector (Waters, Milford, MA, USA) and Software Empower 3. The separation was carried out on a 5 micron (100 Å, 250 × 4.6 mm) C-18 reverse phase Kromasil column (Ale, Bohus, Sweden) at room temperature. The mobile phase was phosphoric acid at 5 mM (solvent A) and MeOH (solvent B), at a flow rate of 1 mL/min. The elution gradient was as follows: a linear gradient of 85–80% solvent B (0–5 min), 60% B (6–10 min), 70% B (11–20 min), 80% B (21–25 min) and, finally, 85% B (26–30 min). The injection volume was 20 μL, and the 5CQA was detected at a wavelength of 325 nm. This method was adapted from Fujioka and Shibamoto (2008) [60]. Sample chromatograms were compared with those of the 5CQA standard for identification. The measurements were carried out in triplicate. Instrumental calibration: Eight different levels of concentration were employed for 5CQA. The Pearson correlation coefficient (r) was calculated to estimate the type of adjustment of the experimental points in the calibration curve, and subsequently, statistical analyses with Student's t-test [61] and variance analysis were performed, to verify its significance.

#### *2.3. CS Toxicological Analysis*

Aflatoxins B1, B2, G1 and G2 were quantified by the method of QuEChERS extraction and ultra-high liquid chromatography tandem mass spectrometry (UPLC-MS/MS) detection [62].

An acute oral toxicity test was performed following the procedure described in the OECD 425 guidelines [63]. Briefly, five female mice, Balb-c strain, 9 weeks old, were used. They were administered 2000 mg/kg of body weight of the aqueous extract of CS, in a single dose, with a cannula, witha4h food fast but not water fast. Under the conditions of a temperature of 24 ± 1 ◦C and photoperiod of 12 h light/12 h darkness, mortality and toxicity signs were registered daily, and weight was measured weekly. Animal experimentation was carried out in accordance with the Official Mexican Method NOM-062-ZOO-1999 [64]; in addition, the protocol was authorized and reviewed by the Internal Committee for the Care and Use of Laboratory Animals of CIATEJ (code 2019-002A).

#### *2.4. Product Development*

#### 2.4.1. Extrusion Cooking

Ingredient mixes of cornmeal (CM), amaranth flour (AF) and CS were weighed and then mixed in a Kitchen Aid mixer (St. Joseph, Michigan, MI, USA). The mixtures were conditioned to adjust them to 21.0 ± 1.0% of moisture content, placed into plastic bags and maintained under refrigeration for 48 h before processing. Sixteen samples in total were prepared. In each treatment, 300 g of sample was used.

Extrusion trials were performed using a Brabender single screw extruder (Plasti-Corder 815808, Duisburg, Germany). The barrel diameter and D/L ratio were 475 mm and 19/25, respectively. A screw configuration with a 3:1 compression ratio was used. The exit diameter of the circular die was 2 mm. A vertical dosing screw feeder (628456, Duisburg, Germany) was used for feeding the conditioned mixtures. The process conditions were set as follows: a feed rate of 40 g/min, a screw speed of 80 rpm, and three barrel temperatures—120 ◦C at the feed entry, 130 ◦C at the middle and 140 ◦C at the die exit. The pressure, material temperature and torque were monitored during the extrusion runs. The extrusion conditions were obtained by preliminary tests (data not shown).

Extrudates were left to cool at room temperature for about 30 min. Moisture content was determined [44]. The extruded products were subsequently baked at 60 ◦C for 2 h, until a moisture content of 5.4 ± 0.3% was achieved, packaged in plastic bags and stored at room temperature until analysis.

#### 2.4.2. Water Solubility Index (WSI)

The method of Anderson et al. (1970) [65] was used. In brief, 2.5 g of sample was added to 30 mL of distilled water at 30 ◦C, in centrifuge tubes, and shaken on a rotary shaker (Roto-Shake Genie, Bohemia, New York, NY, USA) for 30 min. They were then placed in a centrifuge (Universal 320 R Hettich, Tuttlingen, Germany) run at 4000 rpm for 10 min. The supernatant liquor from each tube was transferred into aluminum trays to be oven dried at 80 ◦C for 24 h. As the WSI, the amount of dried solids recovered by evaporating the supernatant from the water absorption test just described is expressed as the percentage of dry solids. Analyses were carried out in triplicate.

#### 2.4.3. Experimental Design

An extreme vertices mixture design [66] was used, varying the amounts of AF (50–98%), CM (0–45%) and CS (2–15%). The proposed range of CS was determined according to previous studies in which agri-food industry by-products were incorporated into extruded foods [30–33]. Figure 2 shows the experimental region of the extreme vertices mixture design, where the 16 points of the experimental runs are indicated.

**Figure 2.** Experimental region of the extreme vertices mixture design.

#### *2.5. Statistical Analysis*

The STATGRAPHICS Centurion XV package (Statpoint Technologies; Warrengton, VA, USA) was used for the data analysis of the analytical method, as well as for design and analysis of the extreme vertices mixture design. Statistically significant differences between values were determined

at the *p* < 0.05 level [66]. The results are expressed as the mean values ± standard errors of the three separate determinations.

#### **3. Results and Discussion**

#### *3.1. CS Characterization and Microbiological Quality*

#### 3.1.1. Bromatological Composition

The results of the bromatological analyses for CS are shown in Table 1. The protein content in CS (15.09% *w*/*w*) was minor compared to previous values reported, 18.6% *w*/*w* [22], and 18.69% *w*/*w* [9]. The low fat content in CS (1.99% *w*/*w*) was similar to a value cited before, 2.2% *w*/*w* [22], and lower than those presented by other authors, 3.78% *w*/*w* [9]. The ash, carbohydrate and moisture values obtained were similar to those published before [6,9,22]. The dietary fiber content was 67.6% *w*/*w*, which is superior to that published, 54.11% *w*/*w* [9] and 62.4% *w*/*w* [22]. The differences in the contents of the nutrients could be due to the origins of the coffee beans.

**Table 1.** Chemical composition of coffee silverskin.


#### 3.1.2. Microbiological Determinations

There is no regulation for the microbiological parameters for CS, as it is a coffee industry by-product; however, in the Mexican Regulation for roasted coffee [67], less than 3 CFU/g of *E. coli* is specified; thus, by comparison, the CS result shows an acceptable level, and when comparing the results obtained for the remaining microbiological determinations with the Official Mexican Regulation for cereals and their products [68]—because the chemical composition of CS is similar to that of cereals—the results obtained are within the parameters, as shown in Table 2.



<sup>1</sup> Maximum level suggested for roasted coffee for Mexico by the Secretaría de Economía. <sup>2</sup> Maximum level for cereal additives for Mexico by the Secretaría de Salud. ns—not specified.

#### 3.1.3. Antioxidant Capacity and Total Polyphenol Content

The DPPH assay is based on the change of the blue-violet color towards pale yellow due to the reaction with antioxidant substances. The antioxidant capacity of the sample according to the DPPH method was 33.23 ± 0.02 μM Trolox/g of CS (dry basis, db). In another study, 21.35 ± 0.39 μM eq Trolox/g (db) was reported [9]. The antioxidant capacity according to the ABTS method was 3.45 ± 0.02 mM Trolox/100 g of CS (db). Some authors have published values for CS of 1.92 mM Trolox/100 g [22], and 2.12 ± 0.4 mM Trolox/100 g dry matter [15], which are consistent with results found in this work.

The total polyphenol assay provides an approximation of the total amount of polyphenols in the sample. In present work was obtained 16.48 ± 6.6 mg GAE/g of CS (db). This was similar to what has been already reported, 16.1 ± 1.2 mg/g of CS [11]. These results suggest the possibility of recycling CS in a new food product as a contribution of bioactive compounds.

#### 3.1.4. Quantification of 5CQA

#### Method Performance

According to Regulation (EC) No. 333/2007 [69], if an analytical method includes an extraction step, the result of the analysis must be corrected based on the recovery, so the level of recovery was calculated. The efficiency of the extraction of 5CQA was 87.01% of recovery. The determination coefficient (r2) was 0.99, which demonstrates the linearity of the calibration curve for the 5CQA at eight concentration levels in the range of 10–500 μg/mL. The instrumental detection (LOD) and quantification (LOQ) limits for 5CQA were determined based on the signal-to-noise ratios of 3 and 10, respectively, using the weighted parameters [61], thus obtaining an LOD of 3.311 μg/mL and LOQ of 11.037 μg/mL.

#### 5CQA Content in CS

The concentration of 5CQA extracted from CS was 499.03 ± 7.45 mg of 5CQA/kg of CS (db). An amount of 198.9 ± 6.6 mg of chlorogenic acid/100 g of CS was reported [11], which is four times higher than the concentration obtained in this work. Meanwhile, others studies have shown contents of 1.0 ± 0.0 to 1.7 ± 0.1 mg of chlorogenic acid/g of CS [18], 9.4 ± 2.6 mg of 5CQA/g extract of CS [19] and 89.83 ± 0.64 mg of 5CQA/g of dry extract of CS [20]. The difference in the contents of 5CQA in the CS could be due to the nature of the coffee beans, their origins, the extraction methods, and the processes of coffee roasting. During this process, when the temperature is higher than 160 ◦C, a series of exothermic and endothermic reactions take place; the bean become light brown, its volume increases considerably and the detachment of CS occurs. The chemical reactions responsible for the aroma and flavor of roasted coffee are triggered at approximately 190 ◦C. These reactions are interrupted at the desired point based on the bean color or programmed time [70–72]. At temperatures between 150 ◦C and 170 ◦C the decrease in 5CQA content starts to speed up [72]. Therefore, as the beans (and the CS) stay longer in the roaster, where high temperatures are present, the content of 5CQA considerably diminishes. This could explain the concentration of 5CQA obtained in the CS.

#### *3.2. Toxicological Aspects*

The negative impact on human health of aflatoxins, especially because of their carcinogenicity, shows the importance of carrying out their quantification [73]. The aflatoxin quantification yielded the following results: aflatoxin B1 < 0.20 ppb, B2 < 0.06 ppb, G1 < 0.20 ppb and G2 < 0.06 ppb. The maximum admissible levels in food, in the European Union, for the sum of the four aflatoxins (B1, B2, G1 and G2) have been set from 4 μg/mL to 15 μg/mL, depending on the type of food (peanuts, nuts, dried fruits and their by-products, and cereals and their by-products) [73]; thus, the sum of the four aflatoxins for CS was below these limits.

For the acute oral toxicity test [63], a single dose administration of aqueous extract (CS) at 2000 mg/kg, was provided by esopharingeal cannulation. Normal behavior was recorded daily in the mice, with normal postural reflex and hygiene habits as well as food and water consumption as appropriate for the species. There were no clinical abnormalities. During the test period (14 days), no signs of evident toxicity or mortality of the experimental mice were observed. The results obtained allow us to affirm that the LD50 is above 2000 mg of CS/kg body weight.

The characterization of the CS and its toxicological evaluation allowed the evaluation of its potential as an ingredient for the food industry, confirming that it is a source of bioactive compounds (including 5CQA), dietary fiber and protein, and low in fat, and that its consumption is safe, so it can be considered for food development.

#### *3.3. Product Development*

The CS was totally incorporated into a food product to recycle this by-product without generating a new by-product derived from the subsequent process; therefore, an extruded ready-to-eat cereal-based food was developed.

#### 3.3.1. Product Formulation

The product formulation was developed using the parameters obtained from the bromatological analyses of the three raw materials. The composition obtained for corn was 7.57% protein, 1.24% ashes, 2.22% fats, 77.46% carbohydrates and 7.49% corrected dietary fiber [74], and that for popped amaranth was 15.60% protein, 2.88% ashes, 7.97% fats, 73.55% carbohydrates and 9.41% dietary fiber; the CS composition is described in Section 3.1.1.

An extreme vertices mixture design was used to determine the best combination of the three raw materials—CS, CM and AF—that minimized the WSI and maximized the 5CQA content. The proposed formulations were designed to be classified as foods with high fiber contents and sources of protein, in accordance with the established Regulation (EC) No. 1924/2006 [75], where a food with a high fiber content is one that has a minimum of 6 g of fiber/100 g of product, and a food source of protein is one in which protein contributes at least 12% of the total energy value, which was verified by performing a theoretical calculation using the values obtained from the bromatological analysis of the raw materials, according to the formulations obtained through the mixture design. Table 3 shows the formulations proposed by the mixture design and the results for the dietary fiber and corresponding percentage of energy contributed by proteins, satisfying both requirements.


**Table 3.** Extreme vertices mixture design. Theoretical values for dietary fiber and protein energy contribution. WSI and 5CQA concentration determined in extruded product.

<sup>1</sup> CS:CM:AF, coffee silverskin/cornmeal/amaranth flour; <sup>2</sup> calculated values for each mixture; <sup>3</sup> TEV, Total Energy Value; <sup>4</sup> WSI, Water Solubility Index; <sup>5</sup> 5CQA, 5-caffeoylquinic acid; mg/kg, mg of 5CQA/kg extruded product; <sup>6</sup> Measured values.

#### 3.3.2. WSI and 5CQA Content in Extruded Products

Extrusion cooking was accomplished. The WSI and 5CQA content were determined for the extrudates; the results are exhibited in Table 3.

The WSI is related to the quantity of soluble molecules, which is related to dextrinization. Thus, the WSI can be used as an indicator for the degradation of molecular compounds and measures the degree of starch conversion during extrusion [29]. The WSI of the extrudates was influenced by the quadratic effect of the raw materials. The adjusted R-square value was 0.90. The CM effect was more important for the decrease in this property. In Figure 3, a decrease in the WSI with an increase in the CM content can be observed. The reduction in starch degradation lowers the WSI, which increases the bowl life of breakfast cereals and reduces the undesirable powdery mouthfeel of extruded snacks [28].

**Figure 3.** Contour plot for WSI of extruded product formulated with corn meal (CM), amaranth flour (AF) and coffee silverskin (CS).

The 5CQA content of extrudates was influenced by the quadratic effect of the raw materials. The adjusted R-square value was 0.97. The CS effect was more important for the increase in this property. In Figure 4, an increase in the 5CQA content with an increase in the CS content can be observed, as expected since this ingredient is the source of this bioactive compound, as shown in Section 3.1.4.

**Figure 4.** Contour plot for 5CQA content of extruded product formulated with corn meal (CM), amaranth flour (AF) and coffee silverskin (CS).

The optimization of the formulation with the response variables WSI and 5CQA content, using the desirability function [66] in the indicated region, maximizing the 5CQA content and minimizing the WSI response, showed that the optimal values for the studied components were 35% CM, 50% AF and 15% CS. The overall desirability was 0.937. The desirability function predicts the response values of the WSI at 24.67 and 5CQA content at 63.41 mg of 5CQA/kg of extruded product, which are similar to the values determined in Table 3 for this formulation. Figure 5 shows the optimized extruded product.

**Figure 5.** Optimized extruded product.

Further research is suggested in order to study the texture properties and sensory evaluation of the extrudates; also, consumer acceptability needs to be explored.

#### **4. Conclusions**

The CS can be considered as a new food ingredient, which can increase the protein and dietary fiber content of food, in addition to providing bioactive compounds such as 5CQA, which has been shown to exert benefits in the human organism. Up to 15% of CS was incorporated into the extruded product.

To ensure the safety of CS as a food ingredient, the application of good manufacturing and storage practices are recommended, from its collection in the coffee industry. A grinding process is suggested for easy handling.

Although this study contributes to establishing routes for the valorization of CS, it is important to note that studies are still needed to demonstrate the sustainability of this food industry by-product. To incorporate CS in the food production chain, it is suggested to carry out shelf-life tests, studies on logistics issues and/or business opportunity studies for coffee roasters. An assessment of consumer acceptability is also necessary. Furthermore, economic feasibility studies are required.

**Author Contributions:** Conceptualization, G.M.G.-M., E.A.-G., R.I.C.-G. and E.A.B.-M.; methodology, E.A.B.-M. and R.I.C.-G.; validation, E.A.B.-M., R.I.C.-G. and P.M.M.-C.; formal analysis, G.M.G.-M., E.A.-G., E.A.B.-M., E.P.-C. and R.I.C.-G.; investigation, G.M.G.-M., E.A.B.-M., E.P.-C., P.M.M.-C. and R.I.C.-G.; resources, G.M.G.-M., E.P.-C. and E.A.-G.; writing—original draft preparation, E.A.B.-M. and P.M.M.-C.; writing—review and editing, G.M.G.-M. and E.A.-G.; visualization, G.M.G.-M.; supervision, E.A.-G.; project administration, G.M.G.-M.; funding acquisition, G.M.G.-M. and E.A.-G. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by CONACYT, FORDECYT 292474, the Centro de Investigación y Asistencia en Tecnología y Diseño del Estado de Jalisco, A.C. (CIATEJ), and the Universidad de Guadalajara (UDG).

**Acknowledgments:** The authors would like to thank CONACYT for scholarship No. 262766.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

*Article*

### **Agave Syrup as an Alternative to Sucrose in Mu**ffi**ns: Impacts on Rheological, Microstructural, Physical, and Sensorial Properties**

**César Ozuna 1,2,\*, Eugenia Trueba-Vázquez 1, Gemma Moraga 3, Empar Llorca <sup>3</sup> and Isabel Hernando 3,\***


Received: 19 May 2020; Accepted: 19 June 2020; Published: 8 July 2020

**Abstract:** Natural sweeteners, such as agave syrup, might be a healthy alternative to sucrose used in sweet bakery products linked to obesity. We evaluated the effect of sucrose replacement by agave syrup on rheological and microstructural properties of muffin batter and on physical and sensorial properties of the baked product. Muffins were formulated by replacing 25%, 50%, 75%, and 100% of sucrose by agave syrup (AS) and partially hydrolyzed agave syrup (PHAS), and by adding xanthan gum and doubled quantities of leavening agents. Rheological and microstructural properties of batter during baking were analyzed over the range of 25–100 ◦C. In the muffins, the structure, texture, color, and sensory acceptance were studied. The combination of agave syrup with xanthan gum and doubled quantities of leavening agents affected (*p* < 0.05) rheological and microstructural properties of the batters and textural properties of the low-sucrose muffins compared to the controls. The increase in agave syrup levels resulted in a darker crumb and crust. Sensory evaluation showed that AS-75 and PHAS-75 were the best alternatives to the control samples. Our results suggest a plausible substitution of up to 75% of sucrose by agave syrup in preparation of muffins, with physical and sensorial characteristics similar to those of their sucrose-containing counterparts.

**Keywords:** inulin; bakery products; xanthan gum; leavening agent

#### **1. Introduction**

Nowadays, the eating habits of the world's population include high ingestion of foods rich in sugars and fats, such as sweet bakery products [1,2]. Due to their practicality and pleasant taste, muffins are widely consumed by people of all ages, although mainly children [3,4]. The excessive consumption of these products has contributed to an increase in a series of non-communicable diseases and comorbidities, such as overweight, obesity [5], and dental caries [6]. According to the World Health Organization, almost 40% of the adult world population was overweight in 2016 and 13% were obese, whereas 40 million children under 5 years old were overweight or obese in 2018 [7].

In addition to providing sweetness to bakery products, sucrose contributes to the development of their structure, texture, and color [8]. Therefore, the replacement of sucrose content by artificial or natural sweeteners in the production of bakery products represents a challenge for the food industry [1,9,10]. Intense or non-caloric sweeteners (sucralose, saccharin, cyclamate, stevia, etc.) have great sweetening power; however, they do not contribute to the formation of the body of the bakery product [11]. On the other hand, energy sweeteners (monosaccharides, disaccharides, polyalcohols, etc.) can give rise to bakery products with stable structure, but they tend to lack in flavor [11]. Nevertheless, some natural agents may combine the best qualities of both groups of sweeteners: good sweetening power and a stable structure of the final bakery product; this group of sweeteners includes agave syrup (AS) [12–14].

Agave syrup is the sweet substance obtained by the hydrolysis of fructans present in the *Agave* spp. heads. In Mexico, where *Agave* spp. is endemic, there are about 205 species; however, agave syrup is obtained mostly from *Agave tequilana* Weber var. blue [15]. This natural sweetener, composed of fructose and fructooligosaccharides, has proven to have beneficial properties on human health, such as high prebiotic capacity and a low glycemic index score, and could prevent obesity and type II diabetes mellitus [16,17].

In bakery products, the use of agave syrup as a partial or total sucrose replacer has been employed in the elaboration of cookies [12], gluten-free cakes [13], and cereal bars [14]. However, the effects of agave syrup on the rheological, microstructural, and sensorial properties of bakery products remain unknown, both in the batter and in the final product.

The aim of this research was to evaluate the effects of sucrose replacement by agave syrup on rheological and microstructural properties of muffin batter, as well as on physical and sensorial properties of the baked product. Additionally, our study aimed to find added value in a natural sweetener that is currently underused in bakery products due to the lack of knowledge on its behavior during the baking process.

#### **2. Materials and Methods**

#### *2.1. Mu*ffi*n Batter Ingredients*

The muffin batter ingredients included wheat flour (Cerealien Bischheim GmbH, Bischheim, Germany; composition provided by the supplier: 15% moisture, 12% protein); sugar (Pfeifer & Langen GmbH and Co., Cologne, Germany); whole liquid egg (Huevos Guillen S. L., Valencia, Spain); skimmed milk powder (Capsa Food, Asturias, Spain); refined sunflower oil (Sovena España S.A., Sevilla, Spain); agave syrup (Mieles Campos Azules S. A. de C. V., Amatitlan, Mexico; specifications of moisture and total sugars provided by the supplier: 23.20% moisture, 92.86% fructose, 0.15% glucose, 0.12% sucrose, 6.71% inulin, and 0.16% other carbohydrates); partially hydrolyzed agave syrup (PHAS) (Mieles Campos Azules S. A. de C. V., Amatitlan, Mexico; specifications of moisture and total sugars provided by the supplier: 23.30% moisture, 85.52% fructose, 0.40% glucose, 0.25% sucrose, 13.58% inulin, and 0.25% other carbohydrates); xanthan gum (Cargill France SAS, Puteaux, France); leavening agents, including sodium bicarbonate, malic acid, and tartaric acid (Hacendado, Valencia, Spain); and salt (Salinas Del Odiel, S. L., Huelva, Spain). The sucrose equivalent (SE) of agave syrup (AS) and partially hydrolyzed agave syrup (PHAS) was calculated using the equation proposed by Koehler and Kays [18] and Belšˇcak-Cvitanovi´c et al. [19]; the SE values for AS and PHAS were 1.61 and 1.49, respectively.

#### *2.2. Batter and Mu*ffi*n Preparation*

Nine muffin batters were prepared according to batter formulations in Table 1. In the case of batters elaborated with sucrose replacement by AS and PHAS (25% replacement: AS-25 and PHAS-25; 50% replacement: AS-50 and PHAS-50; 75% replacement: AS-75 and PHAS-75; and 100% replacement: AS-100 and PHAS-100), the concentration of leavening agents was doubled and xanthan gum was used as a loading agent according to Martínez-Cervera et al. [3].


**Table 1.** Formulations of the different muffin batters (percentage on wheat flour basis).

AS: Agave syrup; PHAS: Partially hydrolyzed agave syrup.

The all-in mixing procedure reported by Rodríguez-García et al. [20] was employed in batter preparation, with some modifications. First, the liquid ingredients (including 0.5 g of xanthan gum dissolved in 30 g of water), except for the sunflower oil, were introduced into the commercial kneader (Thermomix, TM31, Wuppertal, Germany) and mixed for 1 min at a speed of 200 rpm. Then, the solid ingredients were added into the same container and mixed for an additional 2 min at the same speed. Finally, sunflower oil was added and mixed in for 3 min at 500 rpm. Once the smooth batter was obtained, 45 g were placed in paper molds and subsequently deposited in silicone trays. Finally, they were introduced into an electric oven (Electrolux, EOC3430DOX, Stockholm, Sweden) that had been preheated to 180 ◦C. Samples were baked for 30 min at 180 ◦C. Each baking batch consisted of 12 muffins and each formulation was carried out in triplicate, representing 36 samples per formulation. All analyses were performed 24 h after baking in order to ensure stability in the samples.

#### *2.3. Rheology and Microstructure of Batters*

Rheological and microstructural analyses were performed in batters with 0% (control), 50%, and 100% sucrose replacement by both AS and PHAS. Rheological measurements were carried out using a rotational rheometer (Kinexus Pro+, Malvern Panalytical, Malvern, UK) equipped with a Peltier plate cartridge. A series of tests were performed at 20 ◦C with parallel plate geometry (Φ = 40 mm), with the geometry gap set at 1500 μm. Before the rheological test, the batters were all kept at 25 ◦C for 60 min post-preparation. Flow measurements were conducted (shear rate 1 s−<sup>1</sup> to 100 s−1, temperature = 25 ◦C), along with frequency sweeps (stress = 10 Pa, frequency = 0.1–10 Hz, temperature = 25 ◦C) and temperature sweeps in the linear viscoelastic region (frequency = 1 Hz, stress = 100 Pa, temperature = 25–100 ◦C). Vaseline oil was applied to the exposed surfaces of the samples to prevent sample drying during testing.

Microscopical examination of muffin batters during simulated micro-baking was carried out following the methodology proposed by Rodríguez-García et al. [20]. One drop of the sample was placed in the concavity of a glass slide and placed into a temperature-controlled vault. The temperature in the vault rose steadily from 25 ◦C to 105 ◦C at the rate of 1.5 ◦C/min. Batter samples were observed at 4× magnification (objective lens 4 × /0.13∞/ − WD 17.1, Nikon, Tokyo, Japan). A camera (ExWaveHAD, model no. DXC-190, Sony, Tokyo, Japan) was attached to the microscope and connected to the video entry port of a computer. Images were captured and stored in the format of 640 × 540 pixels using the microscope software (Linksys 32, Linkam, Surrey, UK).

#### *2.4. Mu*ffi*n Height and Crumb Structure*

Muffins were cut vertically with a stainless steel knife and scanned by means of a conventional scanner (Epson Perfection 1250; Epson America, Inc., Long Beach, CA, USA) at a resolution of 300 dpi. The maximum muffin height and the crumb structure were measured using ImageJ software (National

Institute of Health, Bethesda, MA, USA). Each image was cropped to a 30 × 30 mm section on which the crumb analysis was performed. The image was split into color channels, the contrast was enhanced, and the image was binarized at the grey scale, resulting in air cells being colored in black and the rest of the crumb in white. Four macroscopic characteristics of the crumb cell structure were calculated, namely the cell area (mm2), cell circularity, cell density, and total cell area within the crumb (%). The data for the muffin height and crumb structure were obtained from 18 different images for each formulation.

#### *2.5. Mu*ffi*n Texture*

Muffin textural properties were evaluated using a texture analyzer (Stable Micro System, TA-XT plus, Godalming, UK) and the Texture Exponent Lite 32 program (version 6.1.4.0, Stable Micro Systems, Godalming, UK). For texture profile analysis (TPA), cubes were cut from the central area of the muffin (1.5 cm per side). Double compressions of 40% deformation at a speed of 1 mm/s were performed, with a resting time of 5 s between the two compressions. Compression was performed with a 35-mm diameter aluminum plate. After the two compression cycles, the following parameters were recorded: hardness, elasticity, cohesiveness, and chewiness [21].

#### *2.6. Mu*ffi*n Crust and Crumb Color*

Color measurements of both the muffin crust and crumb were carried out with a CR-400 chroma meter (Konica Minolta Sensing Americas, Inc., Ramsey, MN, USA). The results were expressed in accordance with the International Commission on Illumination (CIELAB) system with reference to illuminant C and a visual angle of 2◦. The determined parameters were *L*\* (lightness, *L*\* = 0 (black) and *L*\* = 100 (white)), *a*\* ((*a*\* negative values = greenness and *a*\* positive values = redness)), and *b*\* ((*b*\* negative values = blueness and *b*\* positive values = yellowness)). The chroma (*C*\**ab*), hue angle (*h*\**ab*), and total color difference (Δ*E*\*) were calculated using Equations (1)–(3):

$$\mathcal{C}\_{ab}^\* = \sqrt{a^{\*2} + b^{\*2}} \tag{1}$$

$$h\_{ab}^{\*} = \tan^{-1} \frac{(b^{\*})}{(a^{\*})} \tag{2}$$

$$
\Delta E^\* = \sqrt{\left(L^\* - L\_0^\*\right)^2 + \left(a^\* - a\_0^\*\right)^2 + \left(b^\* - b\_0^\*\right)^2} \tag{3}
$$

where *L*∗ <sup>0</sup>, *a*<sup>∗</sup> <sup>0</sup>, and *b*<sup>∗</sup> <sup>0</sup> represent the values of the chromatic coordinates of the control muffin.

#### *2.7. Sensory Evaluation of Final Product*

Seventy untrained judges (18–80 years old, recruited from the Life Sciences Division of the University of Guanajuato, Irapuato, Mexico) simultaneously evaluated the sensory characteristics (appearance, flavor, texture, color, and general acceptability) of the control and the eight muffin formulations made with AS and PHAS. The product acceptability was determined using a nine-point hedonic scale (9 = like; 1 = dislike) [22].

#### *2.8. Statistical Analysis*

Two-way analyses of variance (ANOVAs) with sucrose replacement (0%, 25%, 50%, 75%, and 100%) and agave syrup type (AS and PHAS) as between-subject factors were carried out on all dependent variables in SPSS 18.0 software (SPSS Inc., Chicago, IL, USA). Tukey post-hoc tests were performed for all significant main effects and interactions.

#### **3. Results and Discussion**

#### *3.1. Rheological Properties of Mu*ffi*n Batters*

Figure 1 shows the effect of sucrose replacement (at 0%, 50%, and 100% reduction levels) by AS and PHAS and a combination of xanthan gum and doubled quantities of leavening agents on flow curves of the muffin batters.

**Figure 1.** The impact of sucrose replacement by agave syrup (AS) and partially hydrolyzed agave syrup (PHAS) on flow curves of the muffin batters: control (blue triangles), PHAS-50 (black crosses), AS-50 (yellow diamonds), PHAS-100 (green circles), and AS-100 (red squares).

A typical shear thinning behavior was observed in all batters (Figure 1). Despite having used a combination of xanthan gum and doubled quantities of leavening agents, the batters were less viscous when AS and PHAS were employed as sucrose replacers. This behavior is quite possibly related to the moisture content of agave syrups (23.20% and 23.30% for AS and PHAS, respectively) which would have induced the decrease in AS and PHAS batter viscosities. At a shear rate of 60 s<sup>−</sup>1, the apparent viscosity significantly (*p* < 0.05) decreased as more sucrose was replaced by AS and PHAS, from 3.35 (0.01; the values in parentheses refer to the standard deviation of the mean) Pa·s in the control sample to 2.48 (0.04) Pa·s and 1.98 (0.04) Pa·s in 50% and 100% sucrose-replaced batters, respectively. There was no significant (*p* > 0.05) effect of agave syrup type, nor an interaction between the two factors. Other authors also stated the decrease in the batter viscosity due to the sucrose replacement by syrups resulting in a low product volume and poor cell structure [2].

Figure 2A,B show the frequency dependence of the different batters on the elastic modulus (G ), the viscous modulus (G"), and the phase angle (δ) at 25 ◦C. At low frequencies (0.1–1 Hz), the control batter behaved as a solid (G > G"; δ < 45), while at high frequencies (>1 Hz) this batter exhibited a liquid-like behavior (G < G"; δ > 45). On the other hand, batters formulated with AS (AS-50 and AS-100) and PHAS (PHAS-50 and PHAS-100), using a combination of xanthan gum and doubled quantities of leavening agents, showed a solid-like behavior (G > G"; δ < 45), regardless of the frequencies tested, caused by the gelling effect of xanthan gum in the batter. At 1 Hz of frequency, sucrose replacement significantly (*p* < 0.05) affected the G modulus of batters, with higher sucrose substitution levels producing higher G values. However, there was no significant effect of agave syrup type (*p* > 0.05), nor an interaction between the two factors.

The competition for the available water between sucrose and xanthan gum led to a less elastic semi-solid network, with the G values obtained for batters with 0, 50, and 100% of sucrose replacement being 42.54 (5.43), 84.24 (4.90), and 98.42 (10.55) Pa, respectively. On the other hand, the G" modulus was not significantly affected by the factors studied or their interaction (all *p*s > 0.05). At 1 Hz, the G" values ranged from 45.26 (2.09) to 46.63 (1.95) Pa for the control and the 100% sucrose replacement batter, respectively. As for the δ values, these were significantly affected by sucrose substitution (*p* < 0.05), with higher sucrose replacement levels producing lower δ values, indicating a more solid-like

behavior. The actual δ values obtained at 1 Hz for 0, 50, and 100% sucrose replacement were 46.98◦ (2.35), 28.31◦ (0.45), and 25.50◦ (1.68), respectively. There was no significant main effect of agave syrup type (*p* > 0.05), nor an interaction between the two factors.

**Figure 2.** The impact of sucrose replacement by AS and PHAS on frequency (**A**,**B**) and temperature (**C**,**D**) sweeps of the muffin batters: control (blue triangles), PHAS-50 (black crosses), AS-50 (yellow diamonds), PHAS-100 (green circles), and AS-100 (red squares). In Figure 2A, two variables are shown: G (full data points) and G" (empty data points).

To analyze the structural changes provoked by heat in muffin batters, linear viscoelastic properties were studied in the range of 25–100 ◦C in order to simulate batter behavior during baking. The effects of temperature on the complex modulus G\* and δ during batter heating are shown in Figure 2C,D, respectively. The G\* modulus, a measure of batter stiffness, was lower in the control sample compared to the samples with sucrose replacement along all the sweep temperatures tested. As the temperature increased, batter stiffness increased as well; this was probably caused primarily by starch gelatinization and protein denaturation [23]. The temperature at which the G\* value started to increase was 60 ◦C for the control batter, however it shifted to 75 ◦C and 85 ◦C as sucrose was replaced at 100% and 50%, respectively. In this sense, a synergic effect between sucrose and xanthan gum was observed, delaying the starch gelatinization, and the thermosetting temperature increased in batters with 50% replacement. Most probably, the moisture content of the agave syrups contributed to higher values of the thermosetting temperature in replaced muffins than in the control, since a higher water content needed to be evaporated for the structure of the replaced muffins to be built. A decrease in δ values above those temperatures was also detected, reflecting the predominance of the elastic behavior versus the viscous behavior during starch gelatinization and protein denaturation [24]. In summary, the sucrose replacement by AS and PHAS and a combination of xanthan gum and doubled quantities of leavening agents increased the thermosetting temperature in batters. This fact is important for the correct formation of water vapor and CO2 in the batter, as well as their diffusion and expansion into occluded air cells during the baking process [3,25].

#### *3.2. Microstructural Properties of Mu*ffi*n Batters*

Figure 3A shows the effect of sucrose replacement (at 0%, 50%, and 100% reduction levels) by AS and PHAS in combination with the addition of xanthan gum and doubled quantities of leavening agents on batter microstructures during simulated micro-baking. These microstructural images were analyzed to quantify the bubble area (Figure 3B).

**Figure 3.** Changes in batters during micro-baking by baking temperature: (**A**) light microscopy images (4×) of bubble expansion; (**B**) bubble size distribution histograms.

When the temperature increased, there was an expansion of the bubbles in all samples and the bubble size distribution tended to widen (Figure 3A). In the control batter, the number of CO2 bubbles was higher if compared to replaced batters and the bubble size increased at a controlled and uniform rate. This behavior may have been due to the high viscosity of this batter (as observed previously in Figure 1), which may have reduced bubble movement in the batter and slowed down the coalescence phenomena. Thus, regardless of temperature, the control batters showed a higher frequency of small bubble sizes (0–10,000 μm2) in comparison with replaced batters (Figure 3B).

On the other hand, a lower viscosity of the replaced batters could have aided in bubble coalescence, giving place to bigger bubbles. When sucrose was totally replaced, a high frequency of big bubble sizes (over 60,000 μm2) could be observed for AS-100 and PHAS-100 when compared to the other batters. Regarding the influence of agave syrup type, PHAS samples showed a higher frequency of big bubbles (over 140,000 μm2) than AS samples at both 50% and 100% sucrose replacement (Figure 3B).

#### *3.3. Mu*ffi*n Crumb Structure*

Figure 4 shows the effect of sucrose replacement by AS and PHAS on the macroscopic characteristics of the crumb cell structure, measured by image analysis.

**Figure 4.** The impact of sucrose replacement by AS and PHAS on the muffin crumb structure, namely cell density (**A**), cell area (**B**), total cell area (**C**), and circularity (**D**). Error bars represent 95% confidence intervals. The different subscript letters of the same color represent Tukey's homogeneous groups, with significant (*p* < 0.05) differences between different sucrose replacement levels for AS (blue letters), PHAS (red letters), or for both types of agave syrup when there was no interaction between the sucrose replacement and agave syrup type (black letters).

Regarding cell density, the main effect of the sucrose replacement was significant (*p* < 0.05), with lower cell density for samples with 25% and 50% sucrose substitution in comparison to control

and samples with 75% and 100% sucrose substitution (Figure 4A). There was a significant main effect (*p* < 0.05) of agave syrup type, which did not interact with sucrose replacement (*p* > 0.05). The main effect of sucrose replacement on cell area was significant (*p* < 0.05). As sucrose substitution increased, the average air cell size was higher than the control muffin (Figure 4B). There was also a significant main effect (*p* < 0.05) of agave syrup type, which did not interact with sucrose replacement (*p* > 0.05). A higher average cell area was obtained when substituting sucrose by PHAS than by AS.

As for total cell area (%), there were significant main effects (*p* < 0.05) of sucrose replacement and agave syrup type, with higher levels of sucrose replacement and PHAS reaching higher percentages of total cell area than control muffin (Figure 4C). There was a significant interaction (*p* < 0.05) between the effects of the two variables, caused by an inversion in the general trend at 75% of sucrose substitution by PHAS. Sucrose replacement affected cell circularity, but a significant (*p* < 0.05) reduction was only observed for the formulations with 100% substitution (Figure 4D). However, there was no main effect of agave syrup type on cell circularity, nor an interaction between the two factors (*p* > 0.05).

The obtained results are in accordance with observations of the bubble expansion during micro-baking (Figure 3), where a high frequency of small bubble sizes in control batters could be observed. The variations that occurred in the muffin crumb as sucrose substitution increased are due mainly to the fact that with the addition of xanthan gum and doubled quantity of leavening agents, the batters with sucrose substitution comprised a greater amount of gas. In addition, the low viscosity of the batters allowed for the gas bubbles to have more mobility, provoking their coalescence. The gases did not reach the surface since their exit was hindered by an early formation of the crust, resulting in the setting of the bubbles in the form of diffusion pathways. Thus, the samples with over 25% sucrose replacement by both AS and PHAS reached a height of over 57.33 (2.10) mm. When compared to the average height of 50.93 (2.27) mm obtained from the control samples, significant (*p* < 0.05) differences in height can be observed for all samples with sucrose replacement by AS and PHAS.

#### *3.4. Mu*ffi*n Texture*

Figure 5 shows the effect of sucrose replacement by AS and PHAS on the texture profile analysis of the final products. Since the textural properties of muffins depend greatly on their crumb structure, all samples with over 25% sucrose substitution by either AS or PHAS presented significant differences (*p* < 0.05) with the control for all textural properties evaluated. As a result, all the replaced muffins considerably differed from the control regardless of syrup type or replacement level.

Regarding hardness, the main effect of sucrose replacement was significant (*p* < 0.05), with lower hardness values recorded for samples with sucrose replacement than control, decreasing from 1.27 (0.30) N in the control sample to a range of 0.65 (0.27) to 0.85 (0.32) N for the rest of the samples. Even though agave syrup type did not have a significant main effect (*p* > 0.05) on this attribute, it did interact with sucrose replacement (*p* < 0.05) (Figure 5A).

The main effect of sucrose replacement on springiness was significant (*p* < 0.05). As sucrose substitution increased, the average values tended to increase as well. There was also a significant main effect of agave syrup type on springiness (*p* < 0.05), which did not interact with sucrose replacement (*p* > 0.05). Moreover, a significant difference (*p* < 0.05) was observed between the control sample and the formulations with over 25% sucrose substitution, where the values increased from 0.32 (0.03) in the control formulation to a range of values between 0.38 (0.02) and 0.41 (0.02) for the other formulations. On the other hand, slightly higher springiness values were obtained when substituting sucrose with AS than PHAS (Figure 5B).

Cohesiveness was significantly affected (*p* < 0.05) by sucrose replacement, as well as by agave syrup type. However, there was no significant interaction between these variables (*p* > 0.05). As sucrose substitution levels escalated, cohesiveness significantly increased (*p* > 0.05) as well, from 0.71 (0.02) in the control formulation to at least a value of 0.74 (0.02) for the other samples. Similarly to springiness, slightly higher cohesiveness values were obtained when substituting sucrose with AS than PHAS (Figure 5C). As for chewiness, the results are very similar to hardness. The main effect of sucrose

replacement was significant (*p* < 0.05), while the main effect of agave syrup type was not (*p* > 0.05). There was a significant interaction between the effect of sucrose replacement and agave syrup type (*p* < 0.05) caused by an inversion in the general trend at 50% sucrose substitution by AS (Figure 5D).

**Figure 5.** The impact of sucrose replacement by AS and PHAS on the textural profile analysis of muffins: (**A**) hardness, (**B**) springiness, (**C**) cohesiveness, and (**D**) chewiness. The different subscript letters of the same color represent Tukey's homogeneous groups, with significant (*p* < 0.05) differences between different sucrose replacement levels for AS (blue letters), PHAS (red letters), or for both types of agave syrup when there was no interaction between sucrose replacement and agave syrup type (black letters).

In replaced batters, the gases were not able to fully exit the product due to the increase in the amount of gas production during baking, which was caused by the doubling of leavening agents and the low permeability of the crust. Along with this, the low viscosity of the batter allowed for the air bubbles to move and coalesce, causing the muffin crumb to contain large air cells resembling diffusion pathways, as well as compact areas [26]. This particular structure of the crumb led to increases in cohesiveness and springiness, due to the fact that the presence of air cells allowed the muffins to easily recover their original shape and size after being compressed.

Regarding the decrease in hardness shown by the sucrose-substituted muffins, this was probably caused by the increase in total cell area (%). The products became softer because the air cells did not give any resistance during their first compression. Since chewiness refers to the difficulty to chew and create food bolus, its value depends on the product hardness, which is why the products were easier to chew as they became less hard [3]. The decrease in muffin chewiness could be also attributed to the increase in cohesiveness, since this characteristic allows for the formation of a bolus instead of fracturing or crumbling upon mastication [27]. According to Gao et al. [28], partial and total replacement of sucrose in sweetened baked products usually results in higher hardness and chewiness values, as well as in lower springiness. This is credited to the ability of sucrose to delay starch gelatinization. Nonetheless, due to the incorporation of double quantities of leavening agents, our muffins showed opposite results [29].

#### *3.5. Mu*ffi*n Crust and Crumb Color*

In general terms, the control muffin showed a golden-brown crust and a yellow crumb, which are both characteristic of bakery products. However, as the percentage of sucrose substitution increased, the muffin color in general tended to be darker than in control samples, with the muffin crust turning towards red hues and the crumb losing some of its yellowness.

The main effect of sucrose replacement on *L*\* was significant (*p* < 0.05) in both the muffin crust and crumb. As sucrose substitution increased, the *L*\* values (ranging from 46.27 to 42.50 for the crust and from 61.34 to 53.23 for the crumb) significantly decreased (*p* < 0.05) in comparison with the control samples (58.81 and 69.16 for the crust and crumb, respectively).

Regarding the *C*\**ab* of the crust, significant effects (*p* < 0.05) were observed for sucrose replacement, as well as for agave syrup type, with a significant interaction (*p* < 0.05) between these factors. As for the crumb *C*\**ab*, only sucrose substitution had a significant effect (*p* < 0.05). As sucrose replacement increased, the color intensity of both the crust and crumb of the muffins showed different tendencies; while the crusts values decreased (40.59 for control and 35.99–32.22 for replaced muffins), the crumb values increased (21.87 for control and 30.27–27.50 for replaced muffins). Both presented significant differences (*p* < 0.05) between the control and the rest of the samples.

The main effect of sucrose substitution on *h*\**ab* was significant (*p* < 0.05) on both the muffin crust and crumb. Replaced muffins exhibited lower *h*\**ab* values of both the crumb and crust (ranging from 65.02 to 62.47 for the crust and from 84.07 to 76.10 for the crumb) than control samples (77.55 and 94.79 for the crust and crumb, respectively). Moreover, as percentages of sucrose replacement increased, the crumb *h*\**ab* values decreased. This shows that sucrose replacement by agave syrup provokes hue changes from yellow to orange-red, both in the muffin crumb and crust.

As for Δ*E*\*, the main effect of the two factors was significant (*p* < 0.05) for the crust and the crumb. In general, the samples in which sucrose was replaced with PHAS syrup showed higher Δ*E*\* values in the crust (20.45), whereas crumb Δ*E*\* values were similar for both AS and PHAS (ranging from 14.66 to 18.50). According to Bodart et al. [30], when the values obtained for total color difference are over three, the color variations between the analyzed samples are visible to the human eye. In this study, all samples with sucrose replacement showed Δ*E*\* values of over three when compared to the control formulation, meaning the variations in color were easily noticeable to the naked eye.

The differences in color exhibited by the different formulations can mainly be attributed to the intensification of non-enzymatic browning (Maillard reaction) that happened as the amount of incorporated agave syrup increased [31,32]. Similar results were reported by Kocer et al. [33]—as the amount of sucrose replaced by polydextrose increased, the baked products became darker. When substituting sucrose by other sweeteners such as erythritol, sorbitol, maltitol, xylitol, and mannitol, the opposite tends to occur, as these sweeteners do not contribute to the Maillard reaction [3,34]. On a smaller scale, the muffin color was also affected by the dark color of the syrups in comparison to sucrose. Temperature fulfills an important role regarding the color of the crust and crumb of bakery products. Since the Maillard reaction is temperature-dependent, its effect will vary according to the maximum temperature reached. The crust tends to present a greater change due to the fact that it is exposed to higher temperatures, while the crumb's exposure to the heat is more limited [32].

#### *3.6. Sensory Evaluation of Final Product*

Figure 6 shows the effects of sucrose replacement by AS and PHAS on sensory acceptance of muffins. The main effect of sucrose replacement on the appearance and flavor of the final products (Figure 6A,B, respectively) was significant (*p* < 0.05). There was also a significant main effect of agave syrup type (*p* < 0.05), which interacted with sucrose replacement (*p* < 0.05). Significant differences (*p* < 0.05) between AS and PHAS were observed at 100% sucrose substitution. Throughout all samples, PHAS maintained higher average scores than AS. The interaction in the case of appearance was caused by the decrease of scores that occurred at 50% sucrose substitution by PHAS, the point at which the tendency for the samples with AS kept rising.

Muffin texture was significantly affected by sucrose substitution (*p* < 0.05), as well as by agave syrup type (*p* < 0.05). However, there was no significant interaction between these factors (*p* > 0.05). The texture of samples with AS and PHAS had a similar tendency for all substitution levels, with muffins made with 25% and 50% sucrose substitution yielding significantly (*p* < 0.05) better texture acceptance than the control and 75% and 100% substitution levels (Figure 6C). In general, PHAS had better texture acceptance than AS.

Regarding color and general acceptance, the main effects of both sucrose replacement and agave syrup type were significant (*p* < 0.05). There was a significant interaction between these factors (*p* < 0.05). Although the samples where sucrose was substituted by PHAS did not show significant differences (*p* < 0.05) between replacement levels, the muffins with AS showed less consistency in their scores, presenting significant differences (*p* < 0.05) for the samples with 100% sucrose substitution.

As sucrose substitution increased, the muffin color became darker. This factor favored the color score up to 75% of sucrose substitution by AS. Some judges singled out the fact that the darker color gave the muffins a more "rustic" and "home-made" appearance, which they liked. However, once the color became too dark, the change became unfavorable. On the other hand, when PHAS was used as sucrose replacement, there were no significant differences in the color score of the different samples (Figure 6D). Taking all aspects into consideration, it is possible to say that the samples made by substituting 75% of the sucrose with either AS or PHAS syrups were the best alternative to the control samples (Figure 6E).

#### **4. Conclusions**

Based on the results of our study, both AS and PHAS can be used as alternatives for sucrose in muffin formulations in combination with the addition of xanthan gum and doubled quantities of leavening agents. The sucrose replacement by AS and PHAS reduced the batter viscosity and increased the thermosetting temperature in batters from 60 ◦C for the control to 75–85 ◦C for replaced batters. The textural properties of muffins depend greatly on their crumb structure; all the replaced muffins considerably differed from the control regardless of syrup type or replacement level. As for color properties, as the percentage of sucrose substitution increased, muffin color in general tended to be darker than in control samples, with the muffin crust turning towards red hues and the crumb losing some of its yellowness. Even though the percentage of sucrose replacement affected the rheological and microstructural properties of the batters and physical parameters analyzed in the baked products, the sensory evaluation of the muffins suggested that both types of agave syrup can be good alternatives up to 75% sucrose substitution. When 100% of the sucrose was replaced, muffins formulated with PHAS showed better texture, flavor, color, and general acceptability than those formulated with AS.

**Author Contributions:** Conceptualization, I.H. and C.O.; methodology, I.H., G.M., and E.L.; investigation, C.O. and E.T.-V.; writing—original draft preparation, C.O. and E.T.-V.; writing—review and editing, I.H., G.M., and E.L. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by Dirección de Apoyo a la Investigación y Posgrado (Universidad de Guanajuato, Mexico), grant number 1366/2019.

**Acknowledgments:** The authors acknowledge to the Mexican company Mieles Campos Azules S.A. de C.V. for supplying the agave syrups used in this work. The authors would also like to thank Stanislav Mulík, MA (Applied Linguistics), for his valuable contribution in writing the English version of this paper.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

#### *Article*

### **Development of Durum Wheat Breads Low in Sodium Using a Natural Low-Sodium Sea Salt**

#### **Elena Arena 1, Serena Muccilli 2, Agata Mazzaglia 1, Virgilio Giannone 3, Selina Brighina 1, Paolo Rapisarda 2, Biagio Fallico 1, Maria Allegra <sup>2</sup> and Alfio Spina 4,\***


Received: 11 May 2020; Accepted: 4 June 2020; Published: 5 June 2020

**Abstract:** Durum wheat is widespread in the Mediterranean area, mainly in southern Italy, where traditional durum wheat breadmaking is consolidated. Bread is often prepared by adding a lot of salt to the dough. However, evidence suggests that excessive salt in a diet is a disease risk factor. The aim of this work is to study the effect of a natural low-sodium sea salt (Saltwell®) on bread-quality parameters and shelf-life. Bread samples were prepared using different levels of traditional sea salt and Saltwell®. The loaves were packaged in modified atmosphere conditions (MAPs) and monitored over 90 days of storage. No significant differences (*p* ≤ 0.05) were found in specific volumes and bread yield between the breads and over storage times, regardless of the type and quantity of salt used. Textural data, however, showed some significant differences (*p* ≤ 0.01) between the breads and storage times. 5-hydroxymethylfurfural (HMF) is considered, nowadays, as an emerging ubiquitous processing contaminant; bread with the lowest level of Saltwell® had the lowest HMF content, and during storage, a decrease content was highlighted. Sensory data showed that the loaves had a similar rating (*p* ≤ 0.05) and differed only in salt content before storage. This study has found that durum wheat bread can make a nutritional claim of being "low in sodium" and "very low in sodium".

**Keywords:** *Triticum turgidum* L. subsp. *durum* Desf.; bread; NaCl; low-sodium sea salt; Na<sup>+</sup> reduction; physico-chemical and textural attributes; sensory evaluation

#### **1. Introduction**

There is much evidence suggesting that excessive salt intake endangers our health [1–3], and reducing its consumption is one of the first steps to preventing noncommunicable diseases [4]. Dietary habits are often developed during childhood [5–7], so nutritional education towards a low-sodium diet with adequate potassium intake should be encouraged [8,9]. In Italy, salt consumption by children and adolescents suggests that the average daily sodium consumption exceeds the official recommendations [10].

Natural foods contain modest amounts of sodium [11], and approximately two-thirds of salt intake come from its addition during food preparation [12]. Eighty food categories were identified

as significant contributors to salt intake, and targets were set for the food industry to meet in each category within a certain period [13].

The WHO member states have agreed to reduce the global population's intake of salt by a relative 30 % by 2025, and several strategies have been undertaken to improve the consumer's understanding of healthy eating recommendations [14–19].

Nutrition claims of "low sodium/salt", "very low sodium/salt", and "sodium/salt-free" for foods containing 1.2, 0.4, and 0.05 g kg−<sup>1</sup> of sodium, respectively, (or the equivalent value for salt) on food labels, informs consumers about salt content [20–22].

Salt is an essential ingredient in breadmaking: it retards gas production, enhances bread flavor, affects the rheological properties of dough, controls fermentation (decreasing yeast activity in the dough), and it can affect the quality parameters of bread [23,24]. Furthermore, NaCl has a strengthening effect on gluten, increasing its resistance or elasticity, and decreasing the extensibility of the dough [25,26].

The strategies to reduce sodium in bread include the use of reduced-sodium sea salt [27], the partial replacement of sodium chloride with potassium chloride and yeast extract [28], the use of a salt substitute with 57% of sodium chloride [29], and heterogeneous NaCl distribution, leading to enhanced saltiness by taste contrast [30].

In bread wheat (*Triticum aestivum* L.), salt is generally used at levels of about 1–2% based on flour weight [31]. A survey of salt content in artisan and industrial bread produced in all Italian regions was conducted in 2009/2010, its data having been recently published [32]. Artisan breads contained between 0.7% and 2.3% g/100 g of salt, while industrial bread, on average, contained 1.6% salt, most samples (56%) having a very high content. In the Mediterranean area, the cultivation of durum wheat (*Triticum turgidum* L. subsp. *durum* Desf.) is widespread compared to that of bread wheat [33] as it has a greater tolerance to drought, high temperatures, and fungal diseases, but less resistance to winter and spring cold. According to traditional uses, mainly in Southern Italy, bread is prepared from remilled durum wheat semolina [34]. Durum wheat milling products are characterized by peculiar chemical, rheological, colorimetric, and baking properties [35–39]. The bread has a compact texture, being in some cases excessively dense, with lower specific volume and harder crumbs than white bread [39], and the characteristic taste and flavor are generally enhanced by adding a high percentage of salt, from 20 to 25 g kg−<sup>1</sup> [28].

In the last few years, the use of low-sodium salts in foods has been recommended.

However, almost all the commercially available low-sodium salts are produced by blending purified potassium chloride with ordinary table salts to achieve a reduced sodium content.

Natural low-sodium sea salt not only provides less sodium and does not affect the taste profile, but contains lots of essential trace minerals such as magnesium, potassium, calcium, and other nutrients the body requires.

Therefore, the aim of this paper is to evaluate the effect of substituting salt with low-sodium sea salt to measure the quality parameters of durum wheat bread over long storage.

#### **2. Materials and Methods**

#### *2.1. Materials*

Durum wheat (*Triticum turgidum* L. subsp. *durum* Desf.) remilled semolina for breadmaking [39] was provided by "Valle del Dittaino Società Cooperativa Agricola" (Assoro, Italy), an industrial bakery with a durum wheat mill.

The bread ingredients were food grade. Compressed yeast (AB Mauri, Casteggio, Italy) and traditional sea salt (99.5% NaCl; Mulino S. Giuseppe, Catenanuova, Italy) employed in the breadmaking process were purchased in a local retailer. Saltwell® (Salinity Group, Saltwell AB Göteborg Sweden) is a natural low-sodium sea salt (less than 35%) extracted from an underground sea below the Atacama desert (Chile). This natural sea salt contains 65 ± 1% NaCl, 30 ± 1.5% KCl, 1.0 ± 0.1% of

MgSO4, 0.5 <sup>±</sup> 0.1% of CaSO4, and traces of other salts and minerals. Saltwell® was kindly donated by Medsalt—Mediterranean Salt Company S.r.l. (Rome, Italy).

Various levels (1.70%, 0.35%, 0.15% on semolina basis) of traditional sea salt and Saltwell® were used in dough formulations, as listed in Table 1.


**Table 1.** Bread type code and percentage of two salts on remilled semolina basis.

#### *2.2. Methods*

2.2.1. Physico-Chemical and Rheological Analyses of Remilled Semolina

The physico-chemical analyses of remilled semolina were carried out following the methods indicated by Giannone et al. (2018) and Palumbo et al. (2002) [39–41]: moisture content was determined according to the AACC 08-01 method (AACC, 2000) [42]. Protein content was determined by means of the Infratec 1241 Grain Analyzer (Foss Tecator, Höganäs, Sweden), based on near infrared transmittance. Ash content was determined according to the AACC 44-19 method (AACC, 2000) [42]. The particle size distribution was determined by a LabSifter (KBF7SN, Buhler, Switzerland). Remilled semolina for breadmaking was sieved for exactly 5 min on sieves with openings of 300, 200, 180, and 160 μm. Wet and dry gluten and gluten index were obtained by using a Glutomatic System (Glutomatic 2200, Centrifuge 2015, Glutork 2020; Perten Instruments AB, Huddinge, Sweden), according to the UNI 10690 method (UNI, 1979) [43]. The α-amylase activity was obtained by using the Falling Number 1500 apparatus (Perten Instruments AB, Huddinge, Sweden), according to the ISO 3093:2009 method (ISO, 2009) [44].

The CIELAB color parameters (*L\**, *a\**, *b\**) were determined by Chromameter CR-300 (Minolta, Osaka, Japan), using the illuminant D65. Alveograph indices were determined according to the AACC method 54-30A (AACC, 2000) [42] using an alveograph model MA 87 equipped with the software Alveolink NG (Tripette et Renaud, Villeneuve-la-Garenne, France). Farinograph parameters were obtained according to the AACC 54-21 method (AACC, 2000) [42] using a Farinograph, equipped with the software Farinograph® (Brabender instrument, Duisburg, Germany).

#### 2.2.2. Bread Sample Production and Packaging

Bread samples were produced in a local breadmaking company ("Valle del Dittaino Società Cooperativa Agricola", Assoro, Italy), according to a proven industrial formulation: remilled durum wheat semolina (65 kg), compressed yeast (0.9% on semolina basis), water (66.0% on semolina basis) and the corresponding amount of salt. Six bread formulations containing different levels of traditional sea salt and low-sodium sea salt were produced (Table 1). The dough was mixed for 17 min in a high-speed mixer (San Cassiano, Italy). The final dough temperature was 26 ± 1 ◦C. The dough was left to rest in bulk for 15 min, divided into 980 ± 20 g portions (100 loaves for each production), proofed for 150 min at 32 ± 1 ◦C and 66 ± 2% relative humidity (RH) and baked at 220 ◦C for 60 min, in industrial tunnel ovens measuring 33 × 3 m (Pavailler Engineering, Galliate, Italy). The baked loaves, with an approximate weight of 1 kg each, were automatically transported to a cooling chamber (Tecnopool, Italy) set at 20 ± 2 ◦C for 120 min. After cooling, the loaves were sliced by means of an automatic slicing machine (Brevetti Gasparin, Marano Vicentino, Italy) to 11 ± 1 mm thickness. About 450 g of sliced bread per loaf was packaged under modified atmosphere conditions (MAPs) using inert gas

(70:30 N2:CO2). The bread packaging materials consisted of two plastic films provided by Cryovac Sealed Air (Elmwood Park, NJ, USA).

The samples were stored for up to 90 days at 20 ± 2 ◦C and 60 ± 2% RH. The quality parameters were determined at regular intervals in triplicate for each batch.

The following parameters and properties were tested for each bread sample during each sampling: volume, height, weight, diameter basis, crumb porosity, internal structure, top and base crust thickness, texture profile analysis, water activity, moisture, pH, 5-hydroxymethylfurfural (HMF) content, crust and crumb color, and sensory evaluation.

#### 2.2.3. Bread Quality Evaluation

Determination of the Physico-Chemical Properties of the Breads

The volume was determined in a loaf volume meter by measuring the volume of rapeseed displaced by the bread, according to the AACC method 10.05.01 (AACC, 2000) [42]. The specific volume (mL/g) was calculated as a ratio of the loaf volume and the bread weight. The specific weight was calculated as the ratio of the loaf weight and bread volume. The h/d ratio was obtained as the ratio of the bread height and bread diameter of the loaf base. The crumb porosity was estimated using the Mohs scale. The CIELAB space *L\* a\* b\** color parameters were measured for the crumb, in the transversely cut bread, and on the crust surface, averaging ten distinct points in each case, using a chromameter (CR-200, Konica Minolta, Osaka, Japan) with illuminant D65.

Bread samples were analyzed for Na<sup>+</sup> (mg Kg<sup>−</sup>1) content by inductively coupled plasma optical emission spectrometry (ICP-OES Optima 2000DV, Perkin Elmer, Italy). The samples were first ground to a powder, and oven-dried at 105 ◦C for 4 h until constant weight, then an aliquot equal to 0.5 g was weighed and placed in a muffle furnace at 600 ◦C for 12 h. After mineralization, the ashes were dissolved in 4 mL of distilled water and 0.5 mL of nitric acid at 69.5% (Superpure; Merck, Darmastadt, Germany). The solutions were poured into 50 mL flasks and brought to volume with distilled water before the analyses.

Water activity (aw) was determined by a Hygropalm 40 AW (Rotronic Instruments Ltd., Crawley, UK) according to the manufacturer's instructions. Three bread slices (11 ± 1 mm thickness) were used, after removal of the crust. For each set of determinations, separate loaves were used.

The moisture content of bread crumb was determined by oven drying at 105 ◦C until constant weight, according to AOAC method no. 945.15 [45]. The pH was measured according to [46] using a pH meter (Mettler Toledo, MP 220).

#### *2.3. Texture Profile Analysis of Breads*

The texture profile analysis (TPA) of bread was determined using a Universal testing machine (model 3344, Instron, Norwood, MA, USA.) equipped with a cylindrical probe of 50 mm of diameter and a 2000 N load cell. Data were acquired through Bluehill® 2 software (Instron, Norwood, MA, USA). Cyclic compression tests (a 30-s gap between first and second compression) were set up: the crosshead speed was 3.3 mm/s, the force required to compress the samples by 40% was recorded on 5-cm side square portions of 24-mm thick slices, and the average value of five replicates was taken. The TPA profile recorded four primary parameters: hardness (N), springiness (mm), resilience, gumminess, and one derived parameters (chewiness, N mm).

#### *2.4. HMF Extraction and HPLC Analysis*

HMF was extracted and determined following the methodology proposed by [28]. Ground bread samples (5 g; La Moulinette, Moulinex, 2002) and 25 mL of water (J.T. Baker, Deventer, Holland) were put into a volumetric flask (50 mL) and stirred for 10 min. Then the sample was diluted up to 50 mL with water (JT. Baker, Deventer, Holland) and centrifuged for 45 min at 5000 rpm. An aliquot of the supernatant was filtered through a 0.45-μm filter (Albet) and injected into an HPLC system (Shimadzu Class VP LC-10ADvp) equipped with a DAD (Shimadzu SPD-M10Avp). The column was a Gemini NX C18 (150 × 4.6 mm, 5 μm; Phenomenex) fitted with a guard cartridge packed with the same stationary phase. The HPLC conditions were the following: isocratic mobile phase, 90% water (J.T. Baker) at 1% acetic acid (Merck), and 10% methanol (Merck); flow rate, 0.7 mL/min; injection volume, 20 μL. The wavelength range was 220–660 nm, and the chromatograms were monitored at 283 nm. HMF was identified by splitting the peak of the HMF from the bread-solution sample with a standard of HMF (*p* > 98% Sigma-Aldrich, St. Louis, MO, USA) and by comparing the UV spectra of the HMF standard with that of the bread samples. All analyses were performed in duplicate, including the extraction procedure, and the reported HMF concentration was, therefore, the average of four values. The results were expressed as mg of HMF per kilogram of dry matter.

#### *2.5. Sensory Evaluation*

The sensory profile [28,47] was defined by a trained [48] panel of 12 judges (six females and six males, 28–40 years old). The judges, recruited for their individual abilities, had more than five years of experience in the sensory analysis of bread and bakery products, and they were submitted to further training over 4 weeks to generate attributes using handmade and industrial breads and to familiarize themselves with the scales and procedures. The judges, using a discontinuous scale between 1 (absence of the sensation) and 9 (extremely intense), have evaluated the intensity of the 11 sensory attributes selected on the basis of frequency (≥60%), following the definitions given by [49–51] (Table 2).



The evaluation sessions, performed at 0, 15, 30, 60, and 90 days of storage, were conducted in the sensory laboratory [52] of Di3A (University of Catania, Italy) from 11:00 a.m. to 12:00 a.m. in individual booths illuminated with white light. The sliced bread samples were served on plates, coded with three-digit numbers, and water was provided to judges for rinsing between samples. The order presentation was randomized among judges and sessions using a randomized complete block. All data were acquired by a direct computerized registration system (FIZZ Biosystems. ver. 2.00 M, Couternon, France).

#### *2.6. Statistical Analysis*

The statistical analysis was performed using the Statgraphics® *Centurion XVI* software package (Statpoint Technologies, INC.). A two-way analysis of variance (ANOVA), followed by Tukey's HSD test (*p* ≤ 0.001; *p* ≤ 0.01; *p* ≤ 0.05), was carried out on physico-chemical and textural attributes. The data were expressed as means ± standard deviations. The sensory data for each attribute were submitted to one-way ANOVA. The significance was tested by means of the F-test. A principal component analysis (PCA) was performed using PAST, Paleontological Statistics software package, 2011 [53].

#### **3. Results and Discussion**

#### *3.1. Physico-Chemical and Rheological Characterization of the Durum Wheat Remilled Semolina*

Physico-chemical characteristics of remilled semolina were moisture 13.8 ± 0.07 g/100 g, protein 12.2 ± 0.10 g/100 g, and ash 0.87 ± 0.01 g/100 g. These quality parameters met the Italian legal requirements [54]. Particle size distribution was >300 μm: 11.0 ± 1.73 g/100 g; between 200–300 μm: 26.3 ± 1.15 g/100 g; between 180–200 μm: 22.0 ± 2.00 g/100 g; between 160–180 μm: 20.0 ± 1.00 g/100 g; <160 μm: 20.7 ± 4.04 g/100 g. These findings agreed with those reported by other authors for remilled semolina [39]. Dry gluten content was 10.0 ± 0.1 g/100 g. The gluten index value was 80.7 ± 4.0, and the value of amylase activity at the falling number was low (577 ± 3.0 s). Regarding dry gluten content and relative qualitative index, the sample exhibited regular gluten quantities and high gluten tenacity. Similar values were reported by [28,39,55].

As regards color parameters, the values were lightness (*L\**) 71.0 ± 0.3, red index (*a\**) 2.12 ± 0.02, yellow index (*b\**) 18.52 ± 0.05.

Rheological behavior was ev ergy (W) was 209 <sup>±</sup> 4 10−<sup>4</sup> <sup>×</sup> J, while the tenacity/extensibility (P/L) value showed a tenacious dough (value = 2.5). Strong gluten is expected in remilled durum wheat semolina [24].

Mixing behavior was evaluated by a farinograph apparatus. The semolina sample indicated the quantity of water absorbed at 500 BU (Brabender Unit), and the dough consistency was 60.6 ± 0.04% due to high protein content. The values of dough development time (1 min, 48 s ± 3.0 s), dough stability (4 min ± 12 s), and softening index (58 ± 1 BU) agreed with those reported by other authors on remilled semolina [28,38,39,55].

#### *3.2. Sodium Content in Bread*

The levels of the two salts used in the loaves, the sodium content, and the minimum limits established by EU regulations [20,21] applying to nutritional claims are shown in Table 3.

**Table 3.** Percentage of two salts in bread, sodium content and limits established by EU regulations [21,22] (data are means ± standard deviations).


Different capital letters in the same column indicate significant difference (*p* ≤ 0.001).

#### *3.3. The Quality Parameters of Breads and Their Evolution during Storage*

The *p*-values for all the physical and textural parameters of the bread types with respect to storage time are reported in Table 4.

The specific volumes and weights of the loaves were significant for each of the two factors of variability (type (A), storage time (B), and their interaction (A × B), even with different *p* levels (*p* ≤ 0.001 for storage time, *p* ≤ 0.01 A × B interaction, and *p* ≤ 0.05 per type; see Table 4).

The results of the physical and textural properties of the industrial breads in the MAP conditions during 90 days of storage are shown in Tables 5 and 6.

No significant differences in specific volumes were shown among the bread samples, regardless of the type and level of sea salt (Table 4).

These findings agree with those reported by [23], but they disagree with those reported by [24]. Additionally, no significant differences in specific weight were observed among the controls and other bread samples or during storage time. The addition of different types and quantities of sea salt did not decrease bread yield. After 60 days of storage, the specific weight decreased.

The ratio between the height and diameter of the loaves used in the baking industry to parametrize possible dough failure was significant (*p* ≤ 0.001) for all the factors and their interaction (Table 4). At time 0, control A was found to have the greatest h/d ratio (approximately 4.5) due to the addition of ordinary sea salt (Table 5). The other bread samples, as expected, showed a lower ratio during storage, especially the bread samples containing less traditional sea salt and sea salt with reduced Na+. These findings agree with those reported by [28].

Significant differences were found for loaf porosity among the types (*p* ≤ 0.001) and the A × B interaction (*p* ≤ 0.05), but not for storage time (B) (Table 4). After baking (t0), almost all the types, except for 2B, showed proper development of crumb porosity. Starting from 15 days of storage, the performance of 2A also slightly decreased (Table 5).

Significant differences were found between the types (*p* ≤ 0.001) and storage times (*p* ≤ 0.01 and *p* ≤ 0.001, respectively) but not for A × B interaction as regards internal structure and top crust thickness (Table 4). As for internal structure, only control A had an irregular structure over the whole storage time. Similar results were reported by [28].

As for top crust thickness, for up to 30 days of storage, no remarkable differences were recorded among the types (mean value of 3.8 mm); after 60 days, the values decreased up to 2.67 mm for control B.

No significant difference was highlighted for basis crust thickness between the types, the different storage times, and their interactions (Table 4). Almost all the bread samples exhibited a mean value of basis crust thickness of 4 mm. These findings agree with those reported by [28].

Three of the five parameters of texture profile analysis (hardness, gumminess, and chewiness) were always significant (*p* ≤ 0.001), while resilience and springiness were significant per type and storage time (*p* ≤ 0.001), but not for A × B interaction (Table 4). The two control breads (1A and 1B), as expected, showed lower values for the first three parameters. Starch retrogradation (i.e., the recrystallization of polysaccharide in gelatinized starch) is believed to be the main cause of crumb firmness change during storage [56].

Textural data highlighted high values of hardness, with significant differences among the samples, as reported by [39], and storage time (Table 6).

The hardness values, as expected, increased as the storage period progressed. As regards the bread samples, control A reported the lowest values during the entire storage period. Up to t30, the two controls, albeit with statistically different values, recorded the lowest hardness values. From t60, the control A values remained low, while the control B values increased until reaching about 55 N at the end of storage.


 **4.** Analysis of variance of the physical and textural parameters studied on the loaves (*p*-values).

**Table**

**Factors of**

**Degrees of**

**Specific**

**Specific**

**h**/**d Ratio**

 **Porosity**

**Internal Structure**

**Top Crust Thickness**

**Basis Crust Thickness**

**Hardness**

 **Springiness**

 **Resilience**

 **Gumminess**

 **Chewiness**

286

#### *Foods* **2020** , *9*, 752



No significant differences in springiness or resilience were shown among the bread samples and during the storage times, whatever the type and level of salt (Table 6). Up to 30 days of storage, no remarkable differences were recorded among the breads (mean value of 5.7 mm); after 60 days, the values of springiness increased by up to 7.0 mm. These findings do not agree with those reported by [39].

As for resilience, the average value was around 0.80. During the entire storage period, the two controls showed higher resilience values. From the end of the baking to the end of storage, resilience values decreased slightly. These findings agree with those reported by [39].

With regard to gumminess and chewiness, they increased progressively with increasing storage times and with decreasing salt content, regardless of type, until they reach the maximum at t90 for 2B (58.0 and 426.0). During the entire storage time, the two controls always showed the lowest values, and were similar to each other, except for t90.

Water activity (aw) and moisture content were significant compared to all the factors of variability (Table 7). As for pH and HMF, they were significant compared to all the factors of variability (*p* ≤ 0.001; Table 7).


**Table 7.** Analysis of variance of the chemical and color parameters studied on the loaves (*p*-values).

Crumb lightness and redness were significant compared to all the factors of variability. Crumb yellowness was significant for bread (A) and storage time (B) (*p* ≤ 0.001) but not for their interaction (A <sup>×</sup> B) (Table 7). The effect of the addition of sea salt with reduced Na<sup>+</sup> on the *<sup>L</sup>*\* parameter of crumb during the entire time storage was not significant (Table 7).

Chemical properties of the breads during the storage time are reported in Table 8.

Crumb aw is an important parameter of food processing and conservation technologies that comes into play for food stability and safety. It indicates the amount of free water not linked by bonds with the soluble constituents of the food, i.e., the water that can participate in chemical, physical, biological, and enzymatic reactions.

In general, water activity is a relatively easy parameter to measure, which can be an advantage, especially in the food industry [57].

The aw value ranged from about 0.88 for Control A at t90, to 0.93 for 2A at t0 (Table 7). Similar values have been reported by [55].

After baking, and up to t15, there is no difference among the breads. From t30, water activity decreases for both controls. From t60 to the end of storage, aw decreases slightly for all the types. At t90, only the aw value of Control A is lower than the other types. Moisture content ranged from about 35.5–38.4% at the beginning (Table 8). Bread samples containing natural low Na<sup>+</sup> sea salt show the highest moisture content, and significant differences were found between all the breads. During storage, the breads with NaCl generally show the highest levels of moisture, and at 90 days of storage, the moisture content decreased, ranging from 35.3–32.4%. No significant differences were found between control B and samples 1B (1.22% and 0.25% Saltwell®) and the bread samples with the lowest levels of salt (2A and 2B).

The pH ranges from 5.36 to 5.93 at the beginning; at 90 days of storage, it ranges from 5.73 to 5.82 (Table 8). The variability seems to be more related to the storage time rather than to the different levels and salts used in the recipe. Similar trends were reported both for durum wheat bread with yeast extract and fortified with fiber [28,50].


**Table 8.** Evaluation of the chemical characteristics of the bread samples produced using different types and levels of sea salt during storage (data are means ± standard deviations).

Different letters in the same column indicate significant difference (*p* ≤ 0.01).

HMF is a widely used compound as heat induces the chemical index generally used for monitoring thermal abuse [58–61]. In bread and in other baking products, HMF is used to monitor the heating process, and several factors influence its formation, such as manufacturing conditions and recipe [57–59]. Even if the toxicity risk of HMF is still debated, nowadays, HMF is under evaluation as an emerging ubiquitous processing contaminant since there is evidence to suggest that HMF and its metabolites may have harmful effects on human health [60–63].

Among foods, coffee and bread contribute the most HMF exposure, about 85% of total intake [64].

The HMF parameter was significant compared to all the factors of variability (*p* ≤ 0.001; Table 7). HMF levels at the beginning ranged from about 23 to 39 mg/kg of dry matter (Table 8), and significant differences were found between all samples. These levels were lower than those reported for durum wheat bread with KCl and taste enhancer [28], and it is known that differences in water content in the leavening and/or baking time and the ratio between crumb and crust of the loaf could influence HMF content [58]. Bread samples with the lowest levels of natural low Na<sup>+</sup> sea salt (2 B) had the lowest HMF content. During storage, a decrease in HMF amount was highlighted, though the trend in decrease was not regular. Generally, the bread samples with the lowest levels of salt had the lowest HMF content due to the effects of a high level of NaCl on starch degradation and yeast growth, resulting, in both cases, in higher levels of Maillard indicators [65]. At 90 days of storage, this parameter ranged from about 20.6 to 25.5 mg/kg of dry matter. The HMF trend during storage was similar to those reported by [28,50], suggesting that HMF decrease is more related to storage time rather than recipe.

During storage, crumb redness in the traditional sea salt (control A) test slowly decreased. After t15, the *a*\* value begins to decrease for all breads (Table S1).

#### *3.4. Sensory Evaluation*

The addition of different types and quantities of sea salt had little effect on the sensory characteristics of the bread sample. Table 9 reports the ANOVA results of sensory data and the bread attributes, which significantly differentiated at different *p*-levels (*p* ≤ 0.05; *p* ≤ 0.01; *p* ≤ 0.001), at each sampling. Mean values were reported only for significantly different attributes.


**Table 9.** Influence of type of bread (6) on the attributes and mean scores of the significant sensory attributes (comparison of formulations). Data expressed as means.

Different letters in the same row indicate significant differences at *p* ≤ 0.05 \*, *p* ≤ 0.01 \*\*, *p* ≤ 0.001 \*\*\*.

At t0, the bread samples were evaluated similarly by panellists, with the exception of the "salty" attribute. Obviously, the control breads (Control A and Control B) had the highest value of saltiness.

At 15 and 30 days of storage, the samples were significantly different for the attributes sweet, salty, bread flavor, and overall evaluation. The 0.15 NaCl sample showed the highest intensity of sweet taste, while the control samples, as expected, had the highest score of salt, bread flavor, and overall evaluation.

At 60 and 90 days of storage, the attributes of sweet, salty, and overall significantly differentiated the bread samples. The 0.15 NaCl and 0.15 Saltwell® bread samples had the highest intensity of sweet and the lowest of the attributes salt and overall. The control samples showed the highest intensity of the attribute overall.

The different levels of sea salt did not influence the attributes of texture (i.e., softness), as reported by [28].

Table 10 reports the sensory attributes which significantly differentiated (*p* ≤ 0.05) during the 90 days of storage.


**Table 10.** Mean values of the significantly different sensory attributes (comparison during storage). Three bread loaves were collected at each sampling.

Different letters in the same column indicate significant difference at *p* ≤ 0.05.

Control A showed a significant decrease during storage but only for the attributes of humidity and softness. At 0 and 15 days of storage, Control A had the highest intensity of these two sensory attributes.

Control B showed a significant decrease during storage for the attributes of elasticity, humidity, and softness. These attributes began to decrease after 30 days of storage.

Sample 2A showed a significant decrease only for the attribute humidity, while bread samples 1A, 1B, and 2B did not show any significant differences during the 90 days of storage.

During storage, the bread samples did not develop off-odors or off-flavors in agreement with those reported by [28].

#### *3.5. Multivariate Statistical Analysis*

Principal component analysis (PCA) is a multivariate analysis that allows the reduction and interpretation of large multivariate datasets with some underlying linear structure. In this trial, it was carried out to determine if and which salt (type and concentration) had an influence on the qualitative and sensory traits of the breads. The PCA included the following 24 dependent variables: specific volume, specific weight, h/d ratio, crumb porosity, hardness, gumminess, chewiness, springiness, resilience, water activity, moisture, pH, HMF, acidity, and crust and crumb color parameters (as *L\**, *a\**, *b\**, *h*, *C*).

The two main factors accounting for 56.92% of the total variance were PC1 and PC2 at 37.08% and 19.84% (Figure 1).

There are two types of trends on the first axis: (1) based on salt content, the groups shift from the negative to the positive section, from the breads with minimum salt concentrations (2A and 2B), to those with more (Control A and Control B) (Figure 1); (2) based on days of storage, from the longest (t90) to the shortest (t0) (Figure 1). Convex hulls were used to highlight these trends. They can be defined as the intersection of all convex sets containing a given subset of a Euclidean space. The convex hull of a set of data is the smallest convex set that contains it.

**Figure 1.** Principal component analysis (PCA) scatter diagram defined by the first two principal components (i.e., PC1, PC2) and convex hulls for the measured physico-chemical and textural traits of the breads, grouped by storage time.

The variables that determined these trends were resilience, crust color (as *a\**, *C*), h/d ratio, crumb color (*h*), and moisture, which showed the highest positive loading values (0.272, 0.230, 0.227, 0.228, 0.218, and 0.209 respectively), chewiness, hardness, gumminess, and springiness, with the highest negative loadings (−0.314, −0.311, −0.309, −0.263, respectively).

The groups also showed a gradient with respect to the days of storage, if PC2 is observed: from the positive scores of the longer storage time to the gradually lower scores of the shorter ones (Figure 1).

The variables that positively correlated with PC2 were crust color parameters (*L\**, *h*, *b\**) and moisture (loading values, 0.367, 0.334, 0.271, 0.292, respectively). Moreover, PC2 negatively correlated with aw, specific volume, and crust hardness (−0.294, −0.266, −0.248, respectively).

In summary, sorting the data according to the first two axes distributes the groups in relation to the lowest salt concentration with the maximum storage time, and so on, up to the breads with the highest salt concentrations with the shortest storage times.

PCA loadings did not have the necessary strength to affect the net separation of groups, but this seems to support the hypothesis that the different breads and salt concentrations do not lead to substantial differences in the overall qualitative characteristics and acceptability of the product.

#### **4. Conclusions**

The results of this study showed that replacing traditional sea salt with Saltwell® in durum wheat bread is a possible strategy for reducing sodium intake while maintaining the quality and sensorial characteristics of the bread.

There were no significant differences in the specific volume and bread yield among bread samples and during storage times, regardless of the type and level of sea salt used. The textural data showed high hardness and chewiness values, with significant differences between samples and storage times.

Sensory data showed that the different levels of sea salt did not influence the attributes of softness.

Principal component analysis (PCA) seems to support these findings since, overall, the parameters analyzed were unable to differentiate groups effectively.

Natural low sodium sea salt has made it possible to obtain durum wheat bread with the nutritional claim "low in sodium" (<0.12 g/100 g) and/or "very low in sodium" (<0.04 g/100 g) on food labels, in accordance with EU regulations [20–22]. However, the breads showed good taste and flavor.

These results should encourage the opportunity to produce low-sodium or very low-sodium bread because of consumers' increasing interest in durum wheat bread in accordance with the guidelines for a healthy diet.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2304-8158/9/6/752/s1, Table S1: Colour parameters of the bread samples produced using different types and levels of sea salt during storage (data are means ± standard deviations). Three bread loaves were collected at each sampling..

**Author Contributions:** Conceptualization, E.A., S.M., V.G., and A.S.; data curation, S.M., A.M., and M.A.; formal analysis, S.M., A.M., V.G., and S.B.; funding acquisition, P.R. and B.F.; investigation, E.A. and A.S.; methodology, E.A., S.M., A.M., V.G., S.B., and A.S.; project administration, P.R. and B.F.; resources, E.A., P.R., B.F., and A.S.; software, S.M. and M.A.; supervision, E.A., P.R., B.F., and A.S.; validation, E.A., S.M., A.M., V.G., S.B., M.A., and A.S.; visualization, A.S.; writing—original draft, E.A., S.M., A.M., V.G., and A.S.; writing—review and editing, E.A., M.A. and A.S. All authors have read and agreed to the published version of the manuscript.

**Funding:** Part of this research was funded by the Regional Operational Program of "Regione Siciliana" PO FESR 2007–2013—Asse IV, Obiettivo 4.1.1—Linea di Intervento 4.1.1.2 with the title of the research program "Impiego e valutazione di fibre e sostanze nutraceutiche per l'ottenimento di prodotti da forno salutistici", grant number 5787/3 of the 14/12/2011.

**Acknowledgments:** The authors wish to thank Nick Field for scientific English language editorial assistance. The authors thank Medsalt—Mediterranean Salt Company S.r.l. (Rome, Italy) for kindly donating a sample of Saltwell®.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

MDPI St. Alban-Anlage 66 4052 Basel Switzerland Tel. +41 61 683 77 34 Fax +41 61 302 89 18 www.mdpi.com

*Foods* Editorial Office E-mail: foods@mdpi.com www.mdpi.com/journal/foods

MDPI St. Alban-Anlage 66 4052 Basel Switzerland

Tel: +41 61 683 77 34 Fax: +41 61 302 89 18