1. Introduction
Sponge cake, which starts as a fluid batter, is classified by its solid porous structure after baking. It is prepared using four main ingredients, namely, eggs, sugar, wheat flour and fat, which are responsible for multiple functions in the batter and during baking. The presence of fat and sugar in food products yields positive hedonic responses which may override metabolic responses such as satiety [
1].
According to WHO (World Health Organization) guidelines (2011), adults and children should consume only 10% of calories taken from sugars in their daily diet, which is about 10–14 teaspoons of sugar. In Ireland, this number achieved 14.6%, which is 31.5% higher than the recommended dosage [
2]. This contributes to dietary imbalances associated with overweight and obesity and consequently the prevalence of non-communicable diseases [
3]. As cakes contribute greatly to the dietary intake of sugar and fat in developed countries [
4,
5], it is necessary to find ways to reduce/replace sugar and fat in these products.
Before we consider sucrose and fat replacement/reduction strategies, we must first understand the functions performed by these ingredients in cakes. Sucrose is the most common sugar used in baking [
6]. Sucrose is responsible for the tenderisation of cakes by competing with gluten for available water during batter mixing, thus impeding on the formation of a strong gluten network [
7]. During baking, sucrose increases the temperature of starch gelatinisation and protein denaturation by binding water and therefore limiting water availability to starch granules [
8]. In doing so, sucrose increases the volume and bulk of the final product. The most commonly known function of sucrose is that it provides sweetness to cakes and also acts a flavour enhancer [
6]. In relation to the function of fats, fat entraps air in the batter during the creaming step which provides a structure for leavening gases [
7]. Thus, fat also affects the volume and bulk of the final product. Fat tenderises and adds lubricity to the texture of cakes by coating the protein and starch particles, thereby interfering with the protein matrix [
9]. Fat also emulsifies liquid in cake production which adds further moisture and softness to the product [
10]. Fat sources such as butter contribute to flavour and can act as a flavour enhancer [
10].
Thus, as sugar and fat perform many key functions in cakes, it is difficult to reduce and replace these ingredients without affecting the physical quality and important sensory attributes associated with liking and overall acceptability.
Several authors have studied the impact of sucrose replacement (SR) using a variety of different replacers such as polyols, fibres and artificial intense sweeteners on the overall quality of cakes [
11]. Fibres such as inulin are of particular interest for use as sugar replacers as they possess prebiotic properties [
12]. Inulin (
Figure 1) is a plant-derived storage polysaccharide which is predominantly produced by chicory roots on a commercial basis.
Roberfroid [
13] reported that fructans such as inulin may have positive effects regarding colon cancer prevention. Furthermore, the incorporation of dietary fibre into foods is now considered as a principal prevention strategy against the risk of non-communicable disease [
14]. Fibres such as inulin provide the bulk properties (texture and volume) by binding water [
15] and therefore competing with gluten proteins for available water and delaying starch gelatinisation. Inulin contains 25–35% of the energy of digestible carbohydrates and its sweetness level is 10% that of sucrose [
12]. Cakes containing 30% sugar replacement (SR) with inulin were shown to be similar to control sugar samples with regard to bubble size distribution and physical and sensory properties in a study conducted by [
16]. As mentioned, inulin has a low sweetness value relative to that of sucrose, and therefore it is necessary to combine inulin with an intense sweetener. In a study conducted by Struck et al. (2016) [
11,
17], a 30% SR in muffins with inulin and Rebaudioside A (Reb A) produced cakes with a similar descriptive sensory profile to reference muffins. Reb A is a steviol glycoside (
Figure 2) whose sweetness is 300 times greater in comparison with a sucrose [
18,
19]. Reb A represents white odorless crystals. Steviol glycosides such as Reb A were approved by the European Union for use in foods and are considered natural intense sweeteners [
20]. These sweeteners are a special class of intense sweeteners as they are considered natural unlike most other intense sweeteners [
20] because they are produced from Stevia leaves [
21].
Therefore, the present study investigated the impact of a sequential SR (0, 25, 50 and 75) using a combination of inulin and Reb A on the sensory acceptance of sponge cake to determine the minimum levels of sucrose needed in the formulation to maintain sensory acceptance.
In relation to the reduction in/replacement of fat in cake products, three main classes of fat substitutes exist, namely, carbohydrate-based, protein-based and fat-based [
22]. Carbohydrates are used as fat substitutes as they are similar to triglycerides in relation to physical and chemical structure [
23] and they bind water and therefore contribute to texture and mouthfeel. In a study conducted using three commercial carbohydrate-based fat replacers derived from pectin, gums and oat bran, moisture content increased significantly in biscuits with a fat substitution [
23]. Legumes such as beans are also used as fat replacers as beans are a source of complex carbohydrates roughly containing up to 60% total carbohydrates. Legumes are low glycaemic and high in protein and fibre and their consumption has been associated with a decreased risk of coronary heart disease [
24]. In a study conducted on brownies, pureed black beans successfully replaced fat in relation to sensory acceptance at a level of 30% replacement [
25]. Furthermore, this study reported that a 90% substitution of fat with black beans still produced an acceptable product. In a study conducted on oatmeal cookies, white beans were successful in replacing fat up to a level of 25% without significantly affecting sensory properties [
26]. Cannellini beans have been used to successfully replace fat in brownies by up to 50% without affecting OA, texture and flavour properties [
27].
For the above reasons, the present study utilised pureed butter beans (Pbb) for the sequential replacement of fat (0, 25, 50 and 75%) in reduced sugar sponge cake. Determining minimum levels of both sugar and fat needed to maintain sensory acceptance is imperative and could be a significant development in reducing the dietary intake of both fat and sugar.
Thus, the main aim of this work was, firstly, to determine the minimum levels of sucrose needed in sponge cake to maintain sensory acceptance similar to that of the control sucrose formulation. This was determined through sensory acceptance testing (SAT). Secondly, and following experimental results, the aim was to determine the minimum levels of fat needed to maintain sensory properties associated with liking and OA of reduced sugar sponge cake. This was determined by SAT and optimised descriptive profiling (ODP). Other characteristics such as texture, colour and compositional properties were studied for cakes during sequential FR of sucrose-replaced sponge cake.
2. Materials and Methods
Food ingredients used in this trial included free-range eggs (Upton brand, Ireland); caster sugar (99.9% sucrose, 0.3% moisture and 0.01% sodium, Tate & Lyle brand, UK); cream plain flour (82.7% carbohydrate, 2% of which were sugars, 11.7% protein, 3.4% fibre, 1.4% fat and 0.81% salt, Odlums brand, Ireland); Irish creamery butter (81% total fat, 65.4% of which were saturated and 15.1% moisture, Dunnes stores brand, Ireland); inulin (89% fibre and 8% sugar, Bioglan brand, Australia); Reb A, (Bulk Powders brand, Ireland); butter beans (16.8% carbohydrate, 1.4% of which were sugars, 0.6% fats 0.1% of which were saturated, 5.2% fibre, 0.01% sodium and 0.03% salt, Suma brand, UK); baking powder (3% sodium, Royal brand, US); and whole milk (4.7% carbohydrate, 4.7% of which were sugars, 3.5% fat, 2.3% of which were saturated, 3.4% protein and 0.1% salt, Dunnes stores brand, UK). All food products were obtained from a local supermarket and stored under refrigerated or cool, dry conditions prior to sample preparation.
2.1. Reb A Concentration Adjustment
Optimised descriptive profiling (ODP) was used to determine the concentration of Reb A needed to replace the sweetness concentration of sucrose, ensuring iso-sweetness. These intensity tests were carried out twice using 21 assessors. Concentration adjustments for Reb A were carried out according to the method of [
28,
29,
30,
31] using the same concentrations of stevia (0.06–0.16 g/L) and 24 g/L sucrose. In order to compare the means of the obtained date, one-way ANOVA was used. Tukey’s post hoc test was used to adjust for multiple comparisons between treatment means using SPSS statistics 20 software (IBM, Armonk, NY, USA).
2.2. Sponge Cake Treatments
The formulation used for the preparation of the control sponge cake treatment was based on conversations had with local bakeries, cookbook recipes and associated websites. Three separate batches of sponge cake for all experimental treatments (8) were formulated and manufactured using recipes outlined in
Table 1.
During the first phase of this trial, a control treatment with 0% SR was prepared. Sucrose was replaced by increments of 25% using a combination of inulin and Reb A (0, 25, 50 and 75%). Thus, 4 treatments were formulated during the first phase of this study and samples were identified as follows; SC0/0, SC25/0, SC50/0 and SC75/0. The most accepted sample was chosen from sensory acceptance testing and used for further analysis in determining the minimum levels of fat needed to maintain sensory properties associated with liking and OA of sugar-replaced sponge cake. Thus, four more formulations were prepared where fat was replaced sequentially by increments of 25% using pureed butter beans. The samples were identified as follows; SC30/0, SC30/25, SC30/50 and SC30/75, where the first digit denotes the SR replacement level and the second digit represents the FR level.
2.3. Sponge Cake Preparation
Butter was creamed in an electronic mixer (Kitchen Aid Professional mixer) and sugar was added gradually until soft and light in colour, at speed 2–3 for 4 min. Eggs (average and similar in size, without weighing) were added one at a time and beaten well between each addition at speed 2 for approximately 1 min, for each egg. Flour and baking powder were sifted in to the mixture and were mixed together at minimum speed for 2 min, before milk was added. For SR, inulin and stevia were added to the mixture during the creaming stage in partial replacement of sucrose. For FR, butter beans were firstly drained and then pureed in a Stephan mixer (UMC-5 Stephan u. Sohner & Co, Hameln, Germany) at 21 RPM for 5 min before being used in partial replacement of butter. Batter was poured into circular baking tins (9 inch) and cooked in a Zanussi convection oven (C. Batassi, Conegliano, Italy) for 30 min at 180 °C. After cooking, sponge cakes were left for 20 min in the tin. Then, cakes were placed on a rack to cool down to room temperature and were placed inside plastic storage containers before testing.
2.4. Sensory Analysis
2.4.1. Sensory Acceptance Testing (SAT)
SAT was carried out in the sensory science laboratory at University College Cork according to ISO 11136:2014 [
32]. Using 25 untrained assessors (
n = 25) who were all familiar with the products being tested, SAT took place over six separate sessions as three independent trials were carried out for both phases of this study (FR and SR). A three-digit random code was used to assign the samples (20 × 20 × 20 mm). Sensory sessions were carried out at room temperature under white light. Participants were instructed to use the water provided to cleanse their palates between tastings and used the following hedonic descriptors to rate their degree of liking of sponge cake samples: appearance, flavour, texture, colour and aroma liking. Assessors were asked to indicate their degree of liking for samples on a 9-point, numbers-only hedonic scale. Words were only used to anchor the scale at both ends with the term “extremely dislike” on the far-left end of the scale and the term “extremely like” on the far-right-hand side of the scale. Overall acceptability (OA) of samples was also determined using this scale. The 150 responses were collected for each sample (25 + 25+ 25 × 2) in three independent trials for each phase, using 25 untrained assessors (samples were presented in duplicate). The statistical analysis method is shown in
Section 2.7.
2.4.2. Optimised Descriptive Profiling (ODP)
Optimised descriptive profiling [
30,
31,
32] was also carried out in the special sensory science laboratory of University College Cork. A separate panel of 21 assessors (
n = 21) who were all regular consumers of sponge cake and who had all received prior training in descriptive analysis were trained and participated in this separate ODP test. In order to prevent the carry-over effect and first order, these assessors were served all samples randomly [
33]. ODP only took place on samples during the second phase of this study after an accepted minimum level of sucrose content was determined by SAT. Thus, ODP sessions ran concurrently with SAT sessions during fat optimisation and therefore took place over three weeks. Sensory descriptors were selected from the panel discussion as the most appropriate and reflected the main variation in the samples profiled. The consensus list of intensity descriptors (
Table 2) was measured on a 10 cm continuous line scale. The term “none” was used as the anchor point for the 0 cm end of the scale and the term “extreme” was used as the anchor point for the 10 cm end of the scale, unless stated otherwise in
Table 2. The samples (20 × 20 × 20 mm) were served coded and presented simultaneously to assessors [
34]. All samples were measured on the same scale for each intensity descriptor and presented in replicate. The statistical analysis method is shown in
Section 2.7.
2.5. Physicochemical Analysis
Physicochemical properties were determined for the following samples: SC0/0, SC30/0, SC30/25, SC30/50 and SC30/75.
2.5.1. Texture Profile Analysis (TPA)
Two sponge cake samples (30 × 30 mm) were obtained from the centre of each cake and used for texture analysis. Thus, results obtained for TPA on the Texture Analyser 16 TA-XT2I (Stable Micro Systems, Surrey, UK) represent a mean of 6 values (3 × 2 = 6). Sponge cake samples were sliced horizontally to a height of 30 mm. A double compression test for 50% was used for TPA with use of a 35 mm (in diameter) flat-ended cylindrical probe (P/35), at a speed of 1 mm/s with a 5 s time between the two cycles. This was carried out in accordance with the method of [
35]. The test was carried out in triplicate for all treatments.
2.5.2. Colour
Two sponge cake samples (30 × 30 mm) from the centre of each cake were used for colour analysis. The CIE L*a*b* method was used for crust and crumb colour measurement. A Minolta CR-200B Chroma Meter (Minolta Camera Co. Ltd., Osaka, Japan) was used for L* (lightness), a* (green-red) and b* (blue-yellow) colour components measurement. Colour parameters were measured at two separate points directly from the top of each individual sponge cake sample. For this, the sponge cake samples were horizontally cut to remove the crust, and crumb colour was measured directly at two separate points. As two measurements for crust and crumb colour were taken for each individual sample and two samples were tested for each individual trial, of which there were three, crust and crumb colour values represent a mean of twelve measurements (2 × 2 × 3).
2.5.3. Moisture and Fat
Moisture and fat were detected according to Bostian et al. [
36]. Two samples (30 × 30 mm) from each cake treatment were used for moisture and fat determination. As three independent trials were carried out, results obtained for moisture and fat represent a mean of 6 values (3 × 2 = 6). The Büchi Mixer B-400 (Büchi Labortechnik AG, Switzerland) was used in order to homogenise for compositional analysis. Moisture content was detected on the CEM SMART system and fat content was detected on the SMART Trac system (CEM GmbH, kamp-Lintfort, Germany). Two fibreglass pads were placed in the drying chamber of the CEM SMART system and their weight was tared. The pads were removed and the homogenised samples (2–4 g) were weighed accurately on the pads using separate weighing scales. Following this, one pad was placed over the sample, pressed together and placed back into the drying chamber to begin drying. The moisture (%) was displayed within a few minutes. A sheet of Smart Trac film was used to determine a percent of fats by wrapping the fibreglass pads with the sample. After wrapping, the samples were placed in a tube of the Smart Trac system and positioned in the Smart Trac NMR unit. After about 5 min, the percentage of fat was determined.
2.5.4. Protein
Results obtained from protein analysis represent a mean of 6 values (3 × 2 = 6). Protein content (%) was determined using the Kjeldahl method which was carried out according to [
37]. Before testing commenced, the digestion block (Foss Tecator Digestor, Hillerød, Denmark) was pre-heated to 410 °C. Samples (0.5–2.0 g) were weighed accurately into digestion tubes. Two “kjeltabs” were added to each sample in the fume hood followed by 15 mL of sulphuric acid and 10 mL hydrogen peroxide. Two controls containing no sample were prepared in the same way. Tubes were placed in the heated digestion block and left there for roughly 30–40 min until they became colourless. At this point, the distillation unit was turned on and rinsed out by connecting a blank tube and receiver flask to the unit and pressing the steam button. Distilled water (50 mL) was added to each digested sample in the fume hood after the samples had cooled thoroughly. Samples were placed one by one into the distillation unit (Foss Kjeltec 2100, Hillerød, Denmark) along with a receiver flask containing 50 mL of 4% boric acid with an indicator. After the distillation process was complete, the contents of the receiver flask were titrated with 0.1 N hydrochloric acid until the green colour reverted back to the original red colour. Nitrogen content was converted to protein content using the factor 6.25.
2.5.5. Ash
Results obtained represent a mean of 6 values (3 × 2 = 6). A muffle furnace was used for ash content (%) detection. Homogenised samples (5 g) were weighed into small silica dishes and placed into the muffle furnace. The samples were heated up to 600 °C until only the inorganic material was left which was indicated by the light-white colour of the samples. The silica dishes containing the samples were then placed in a desiccator to cool and the dishes were weighed carefully. Ash (%) was calculated using the following equation:
2.5.6. Carbohydrate
Carbohydrate (%) content was determined using the following calculation:
2.5.7. Dietary Fibre
Total dietary fibre was established according to the AOAC 985.29 method [
38]. Sponge cake samples went through sequential enzymatic digestion using heat-stable α-amylase (95–100 °C, pH 6.0, 15 min), Subtilisin A (60 °C, pH 7.5, 30 min) and amyloglucosidase (60 °C, pH 4.5, 30 min) to remove digestible carbohydrates and protein. To establish total dietary fibre content, the enzyme digests were precipitated with four volumes of 95% (
v/
v) ethanol, filtered and then washed with 78% (
v/v) ethanol and acetone before being dried in an oven at 110 °C.
2.5.8. Total Sugars and Sucrose Content
Total sugars and sucrose content of samples was determined by ion chromatography, using a method developed internally by an independent accredited laboratory based in Ireland.
2.6. Sponge Cake Images
Photographs of the following sponge cake samples: SC0/0, SC30/0, SC30/25, SC30/50 and SC30/75, were taken one day after baking in the photograph room of the Food Science Building, University College Cork. Photographs were taken using a digital camera, Nikon D5300, equipped with a lens, Nikon DX AF-P NIKKON (18–55 mm, 1:3.5−5.6 G), Japan. Images were taken without flash with the following modes: macro, F6.3, 1/125, ISO 400. After images were taken, sharpness was increased to 100% and the saturation was adjusted for all images.
2.7. Statistical Analysis
Raw data obtained from sensory and physicochemical analysis were coded in Microsoft Excel. For sucrose optimisation, the significance of hedonic sensory properties in discriminating between the samples was analysed using ANOVA and Tukey’s post hoc test (SPSS statistics 20 software (IBM, Armonk, NY, USA). The relationship between the set of samples (4) and the set of hedonic sensory variables was determined by partial least squares (PLS) regression using Unscrambler software (Unscrambler 10.3 CAMO software ASA, Trondheim, Norway). In the PLS regression, only hedonic sensory properties that discriminated significantly between samples were used. The X-matrix was defined as the different sample treatments. The Y-matrix contained the significant sensory variables of the design. For fat optimisation, the relationship between the set of sample treatments (X) and the set of sensory (hedonic and intensity) and physicochemical variables (Y) was examined by PLS regression. Again, only sensory and physicochemical properties that discriminated significantly between samples (4) were used. Both the sensory and physicochemical data were normalised during pre-processing by taking the logarithm to achieve uniform precision over the whole range of variation. Data were also standardised by dividing each variable (sensory and physicochemical) by its standard deviation. This process was necessary as the units of the studied variables were different. To achieve significant results for the relationships determined in quantitative PLS regression analysis, coefficients were analysed by jack-knifing which was based on custom cross-validation and stability plots [
39]. Statistical significance for the relationships analysed by PLS was defined as
p < 0.05–0.01 (significant),
p < 0.01–0.001 (highly significant) and
p < 0.001 (extremely significant).
TPA and proximate composition data were presented as a mean of six values ± standard deviation. Estimated fibre and sucrose contents were presented as a mean of three values ± standard deviation. Colour (crust and crumb) data were presented as a mean of twelve values ± standard deviation. One-way ANOVA was used to compare the means of the data obtained from physicochemical analysis. Tukey’s post hoc test was used to adjust for multiple comparisons between treatment means using SPSS statistics 20 software (IBM, Armonk, NY, USA).
4. Conclusions
Texture liking and, subsequently, OA were negatively affected by a 50% SR using a combination of inulin and Reb A. However, other important hedonic attributes such as flavour liking and aroma liking were not negatively affected by this level of replacement. This demonstrates the value of these replacers in relation to maintaining flavour properties associated with liking. Further work to improve texture, such as the addition of emulsifiers, for example, polysorbates, or the addition of hydrocolloids, is necessary to permit higher levels of sucrose replacement using this combination of replacers.
Pureed butter beans were very successful in the replacement of fat by up to a level of 50% replacement. Flavour and aroma intensity attributes and, subsequently, liking and OA were negatively affected by a 75% FR. Perhaps the addition of flavourings to compensate for the reduction in flavour lost with butter would permit higher levels of fat replacement using pureed butter beans.
The present study achieved a simultaneous 50% FR in 30% sucrose-replaced sponge cake, using natural substitutes and without negatively affecting OA. Sponge cake samples contained 42% less fat and 28% less sucrose than the original sample. Thus, the criteria to meet the standard for a “reduced fat” claim were achieved. More work is necessary to permit further sucrose and fat replacement using these ingredients and without negatively affecting important quality parameters, in order to meet the criteria for reduced/low-sugar and low-fat health claims. Optimised samples contained 115% more dietary fibre than the original sample.