Do sugar substitutes affect quality characteristics and HMF levels of cakes?

Abstract

BACKGROUND

Sugar substitutes have become a topic of interest with the revelation of the strong link between excessive sugar consumption and many chronic diseases. One of the components known to be harmful to human health as a result of mutagenic, carcinogenic and cytotoxic effects in bakery products is 5‐hydroxymethyl furfural (HMF). Therefore, the present study aimed to investigate the effect of using some sugar substitutes (liquid stevia, agave syrup, date syrup, apple juice concentrate) on the quality and HMF levels of cakes.

RESULTS

The total colour changes of the crumb and crust colour of all cakes with substitutes compared to the control group (sucrose) were visually noticeable (ΔE > 3). Mean values of hardness, baking loss, water activity were significantly higher and mean values of volume and symmetry index were significantly lower (P < 0.05) in all substituted cakes compared to the control group. A 100% substitution of sucrose with conventional sweeteners significantly decreased the energy (between 6.9% and 10.1%) and carbohydrate content (between 18.1% and 47.9%) of the cakes (P < 0.05). The HMF content of the cake sweetened only with liquid stevia was lower than the control group and this decrease was statistically significant (P < 0.05).

CONCLUSION

Stevia sweetened cake seems to be the best alternative with the lowest HMF content and the highest sensory analysis scores compared to other substitutes. Further studies are needed to determine the sweetness ratios of sugar substitutes, as well as to investigate the possibilities of their use in bakery products and develop new formulations to improve quality and sensory properties. © 2025 The Author(s). Journal of the Science of Food and Agriculture published by John Wiley & Sons Ltd on behalf of Society of Chemical Industry.

INTRODUCTION

Refined sugar is a processed product obtained from sugar cane or sugar beet and contains 99.0% sucrose. In modern societies, sugar is an important substance recognised not only for its taste and sweetening properties, but also for its role as a preservative, bulking agent, texturiser and moisturiser. ¹ , ² In addition to its important roles in foods, excessive consumption of simple sugars has been found to be associated with health problems such as obesity, metabolic syndrome, type 2 diabetes and cancer. ² Currently, consumers are more concerned about their health and therefore are more interested in healthy foods and/or beverages. For this purpose, reducing the amount of sugar in many foods has taken its place among the priority targets. ² , ³ Efforts to find more natural and healthier alternatives with sweet taste to replace sucrose have also accelerated. ³

Sugar substitutes are widely used in the food and beverage industry because they are considered safe for human health, are stable under many conditions and have low (or no) energy content. ⁴ Sugars obtained from natural sources (unrefined sugar) also contain various bioactive compounds, minerals, fibres, antioxidants and phytochemicals. As a result of these molecules, it is assumed that substitution of unrefined sugars with refined sugars may have positive effects on health and, accordingly, studies aiming to determine the effects of natural sugar substitutes on human health have accelerated. ⁵ , ⁶

5‐hydroxymethyl furfural (HMF) is a furanic compound formed as an intermediate of Maillard reaction during heat treatment of foods or as a result of sugar dehydration reaction under acidic conditions. ⁷ Furan and furan derivative compounds are mostly formed in foods during heat treatments such as canning, cooking, baking and roasting at 150–200 °C. Therefore, they are found in various foodstuffs such as coffee, baby foods, canned food and bakery products. ⁸ , ⁹ It is used to determine the storage time in processed foods, as well as to determine the compliance of the applied heat treatment with the standards, and therefore it is also accepted as a quality parameter. ⁹

The major concern for HMF is that HMF can be converted to 5‐sulfoxy methyl furfural (SMF), a genotoxic compound, by sulfonation of the allylic hydroxyl functional group catalysed by sulfotransferases in the body. Despite its short half‐life, SMF can react with DNA and other macromolecules to produce toxic and mutagenic effects. For these reasons, it is extremely important to consider maximum concentration limits for certain foods. These concentration limits are determined in foods such as molasses, honey and fruit juice. ¹⁰ , ¹¹ , ¹² , ¹³ For example, the Codex Alimentarius Commission set the maximum limit of HMF in honey at 40 mg kg⁻¹ (80 mg kg⁻¹ for honey from tropical regions) to prevent honey from being exposed to overheating during processing and to ensure consumption safety. ¹⁴ There is no limit value for the amount of HMF in bakery products in the literature. The European Food Safety Authority stated an acceptable daily intake level of 0.5 mg kg⁻¹ for furfural. However, there is currently no information on the tolerable daily intake level of HMF based on available data. ¹⁵ , ¹⁶ Although the results of the studies are in a wide range, HMF exposure is estimated to be 30.0–150.0 mg per person per day. ⁷ , ¹⁷ , ¹⁸ , ¹⁹

Bakery products such as biscuits, cookies and cakes are among the most consumed foods in the world. ²⁰ For this reason, it has been reported that bakery products such as biscuits and cakes come after sugary drinks with 10.0–45.0% sugar content when foods and beverages that increase simple sugar consumption are considered. ²¹ , ²² , ²³ Reducing the amount of sugar in bakery products is among the first options that come to mind when aiming to reduce sugar consumption. However, it is very difficult to reduce sugar in bakery products because sugar has many functions other than providing sweetness and flavour. ²⁴

Factors such as the type of sugar/sugar substitute, pH, baking temperature and time affect HMF formation in cakes. ²⁰ HMF also adversely affects food safety and human health. For this reason, it has become a necessity to determine the amount of HMF formed in foods during production and storage and to carry out studies to reduce it. ²⁵ In the literature, no study has compared the effect of the use of plant extracts (date syrup, apple juice concentrate, agave syrup, liquid stevia) as sugar substitutes in bakery products on HMF, ¹⁰ which is considered as one of the most important chemical pollutants. However, when the prevalence of the use of such products is examined in markets and online shopping sites, it is seen that these types of sugar substitutes are easily accessible today. In addition to their use in industry, it can be concluded that they have started to be widely used in the recipes for domestic use. From this point of view, the main objective of this study was to investigate the effect of the use of various sugar substitutes (date syrup, apple juice concentrate, agave syrup, liquid stevia) on the HMF level of cakes and to examine the effect of these substitutes on the quality and sensory properties of cakes in order to observe the usability of sugar substitutes.

MATERIALS AND METHODS

In the present study, the cake baked with sucrose was considered as a control cake. There were four different cakes baked using sugar substitutes (liquid stevia, agave syrup, date syrup, apple juice concentrate). Each cake was baked under the same conditions and at two different times to enable two replicate analyses. Flour, sugar substitutes, milk, egg, vegetable oil, baking powder, vanillin and table sugar used in the cakes were obtained from local markets in Izmir province. Carrez I [(K₄Fe(CN)₆3H₂O)] and Carrez II [Zn(CH₃CO₂)2H₂O] solutions (#110537; Merck; Darmstadt, Germany), methanol (#106035; Merck) used in chemical analyses were purchased from chemical material suppliers.

Production of cakes

While determining the formulation of the sucrose‐containing control cake, the formulations in the literature and AACC Standard Method No: 10‐90.01 were examined and these formulations were modified by the researchers to make them similar to homemade. ²⁶ , ²⁷ , ²⁸ , ²⁹ , ³⁰ When determining the amount of sugar substitute, the amount recommended by the suppliers of the products used or the usage recommendation on the label was taken into consideration. The cake formulations in the present study are given in Table 1. All of the ingredients used were kept at room temperature for 1 h before making the cake. Sugar substitute or sugar and eggs were whisked with a hand mixer (Ergomixx MFQ36460; Bosch, Gerlingen, Germany) at third speed for 5 min, and then all liquid ingredients (oil, milk) were added and mixing was continued for 2 min. Finally, sifted flour, baking powder and vanillin were added and mixed again at third speed for 5 min. ²⁶ , ³¹

Table 1.

Formulations of the cakes produced in the present study

Ingredients/group	Sucrose (control) SKR	Stevia STV	Agave syrup AGV	Date syrup DAT	Apple juice concentrate APL
Wheat flour (g)	100.0	100.0	100.0	100.0	100.0
Milk (g)	75.0	75.0	75.0	75.0	75.0
Egg (g)	60.0	60.0	60.0	60.0	60.0
Sunflower oil (g)	50.0	50.0	50.0	50.0	50.0
Baking powder (g)	5.0	5.0	5.0	5.0	5.0
Vanilla (g)	2.5	2.5	2.5	2.5	2.5
Sugar or substitute (g)	90.0	8.8	67.5	45.0	67.5
Total product amount (g)	382.5	301.3	360.0	337.5	360.0

The prepared cake batter was poured into paper cake moulds (9 × 5 cm) with 120 g of cake batter in each mould. All cakes were baked in an oven (8314 DG cooker oven; Arçelik, Istanbul, Turkey) at 180 °C for 30 min. Images of the cakes are presented in Fig. 1.

Figure 1.

Internal and external appearance of the cakes produced in replicate. (A) Cake flavoured with sucrose. (B) Cake sweetened with liquid stevia. (C) Cake sweetened with agave syrup. (D) Cake flavoured with apple juice concentrate. (E) Cake flavoured with date syrup.

Physical analyses of cakes

Colour analysis

Colour was measured from the surface and horizontal midpoint sections of the cake samples using a calorimeter (Chroma Meter CR‐5; Konica‐Minolta Sensing Inc., Osaka, Japan). International Commission on Illumination (CIE) L*, a*, b* values were determined separately for the crust and interior of the cake [L*: (0) black – (100) white, a*: (+) red – (−) green and b*: (+) yellow – (−) blue]. ³² Colour measurements were carried out in two replicates and four parallels (n = 8) for all samples. Croma, hue angle and browning index were calculated using: ³³ , ³⁴

$$\mathrm{Hue}\text{angle}\left(h^{*}\right):\arctan \left(b^{*}/a^{*}\right)$$

$$\text{Browning Index}\left(\mathrm{BI}\right):\left[100\left(x–0.31\right)\right]/0.17$$

$$x=\left(a^{*}+1.75L^{*}\right)/\left[\left(5.645L^{*}\right)+a^{*}–\left(0.3012b^{*}\right)\right]$$

Cooking loss and yield

Doughs (DW) before baking and cakes (CW) at room temperature after baking were weighed (two replicates and two parallels). The weight (g) obtained was placed in the relevant places in the formula [(DW – CW) / DW × 100] and the baking weight loss was calculated. Baking efficiency was calculated using: [(CW/DW) × 100].

Structural features

The structural properties of the cakes were determined by modifying the cake measurement template specified in the AACC 10–91 method for this study (two replicates and two parallels). ²⁹ The formula used to determine the structural properties are: ³⁵ , ³⁶ , ³⁷

$$\begin{array}{l}\text{Volume index}\left(\mathrm{mm}\right)={\mathrm{ǀBB}}^{\text{'}}\mathrm{ǀ}+{\mathrm{ǀCC}}^{\text{'}}\mathrm{ǀ}+{\mathrm{ǀDD}}^{\text{'}}\mathrm{ǀ},\\ \text{Symmetry index}\left(\mathrm{mm}\right)={\text{2ǀCC}}^{\text{'}}\mathrm{ǀ}-{\mathrm{ǀBB}}^{\text{'}}\mathrm{ǀ}-{\mathrm{ǀDD}}^{\text{'}}\mathrm{ǀ},\\ \text{Uniformity index}\left(\mathrm{mm}\right)={\mathrm{ǀBB}}^{\text{'}}\mathrm{ǀ}-{\mathrm{ǀDD}}^{\text{'}}\mathrm{ǀ}\end{array}$$

Hardness analysis

Hardness analysis was performed using TA.XTplus Texture Analyzer (Stable Micro Systems Ltd, Godalming, England) according to AACC Method 74‐09. ²⁹ A cylindrical probe with a diameter of 36 mm (P36/R) was used. The test configuration was set as pre‐test speed of 1.0 mm s⁻¹, test speed of 1.7 mm s⁻¹, post‐test speed of 10 mm s⁻¹, stress of 40.0% and trigger force of 5 g, and the hardness value was measured. For each sample, two replicates were carried out in six parallels (n = 12).

Chemical analyses of cakes

A hand blender (TEFAL HB234D; Powelix Activflow Pro; Prodware, Paris, France) was used to grind the cakes for chemical analyses. The cakes were ground first for 10 s at low speed and then for 20 s at high speed.

pH

For pH analysis of ground cakes, the method of Torley et al. ³⁸ was modified. For this, 30 mL of distilled water was added to 3 g of sample taken in tubes and mixed with the help of vortex for 3 min. The samples were kept for 1 h. Then, measurements were performed with a pH meter (HI 2211; Hanna Instruments, Woonsocket, RI, USA) calibrated with pH 4.00 and pH 7.00 buffer solutions.

Water activity (a_w)

The a_w values of the ground cake samples were measured with a water activity meter (HygroPalm; Rotronic, Bassersdorf, Switzerland). ³⁹

Energy and nutrient composition

The moisture content of the samples was determined according to the method of AOAC 925.10. ⁴⁰ In this method, drying containers washed with distilled water were dried in an oven (UFE400; Memmert, Schwabach, Germany) at 105 ± 2 °C until they reached constant weight. After the containers were cooled in a desiccator, 5 g of sample was again dried in an oven at 105 ± 2 °C until constant weight. The percentage moisture content was calculated by proportioning the amount of moisture removed to the weight of the initial sample: (% moisture = [moisture loss (g)/sample weight (g)] × 100).

The carbohydrate content of the cakes was determined by the Atwater method using the formula %Carbohydrate (g) = 100 – [%Moisture (g) + %Protein (g) + %Fat (g) + %Fibre (g) + %Ash (g)]. ⁴¹ The amount of fibre in the samples was determined in accordance with the method of AOAC 991.43. ⁴⁰ Protein content of the samples was determined by using Nitrogen‐Micro Kjeldahl method with a nitrogen conversion factor of 5.80 in accordance with AOAC 960.52. ⁴⁰ Fat analysis was performed using Soxhlet method in accordance with AACC 30‐25 standards. ⁴² The ash content in the samples was determined by burning in a muffle furnace (PLF 120/15; Protherm, Turkey) at 550 ± 10 °C for 4 h in accordance with AOAC 923.03 method (AOAC, 1990). The energy values of the cakes were calculated according to the Atwater method; Energy (kcal/100 g) = ([%Carbohydrate (g) + %Protein (g) × 4]) + [%Fat (g) × 9] + [%Fibre (g) × 2] equation. ⁴¹

HMF

HMF determination was performed according to the method described by Zhang et al. ²⁰ Samples containing 1 g of peel and inner part were taken into a 50‐mL centrifuge tube and 10 mL of pure water was added. Then, 0.5 mL of Carrez I solution [150 mg mL⁻¹ potassium ferrocyanide; K₄Fe(CN)₆3H₂O] and 0.5 mL of Carrez II solution [300 mg mL⁻¹ zinc acetate; Zn(CH₃CO₂)2H₂O)] were added to the samples. The samples were then vortex mixed for 1 min and centrifuged at 3044 x g for 10 min at 4 °C. The supernatant obtained was filtered through a 0.45‐μm membrane filter. The same procedure was repeated for the liquid products used as sugar substitutes (n = 4). The vials were kept in a freezer at −18 °C until the analyses were performed. ²⁰ HMF in the filtrates was analysed by high‐performance liquid chromatography (1260 Series; Agilent, Santa Clara, CA, USA). The chromatographic separations were carried out at 280 nm and 25 °C with ACE 5 C18 column (4.6 × 250 mm, 5 μm). The mobile phase was methanol–water–acetic acid (10:90:1, v:v) with a flow of 1.0 mL min⁻¹ and the injection volume was 20 μL.

Sensory evaluation

Ethical approval for sensory analyses was obtained from Ege University Medical Research Ethics Committee (decision number 23–2.1 T/31). Cakes brought to room temperature for sensory analyses were cut appropriately. Each sample was coded with three‐digit numbers. Fifteen people (three males and 12 females) who had graduated from food engineering or nutrition and dietetics departments participated in the panel. The panellists consisted of healthy individuals aged between 22 and 50 years without any chronic disease. The analysis was carried out in two replicates. The panellists were allowed to participate in the sensory tasting, approximately 2–3 h after the main meal. Participants were advised to drink water before tasting each cake to neutralise the taste in the mouth. Each sample was evaluated on a nine‐point hedonic scale (from ‘1’ indicating ‘extremely poor’ to ‘9’ indicating ‘excellent’) in the categories of appearance, colour (crumb colour, crust colour), flavour, texture and overall liking.

Statistical analysis

Data were analysed using the SPSS, version 25 IBM Corp., Armonk, NY, USA). P < 0.05 was considered statistically significant. The relationship between HMF and crust L* value and BI was evaluated by a Spearman correlation test. Means for all chemical, physical and sensory analyses were evaluated by one‐way analysis of variance (ANOVA). Tukey/Tamhane multiple comparison tests were applied according to the homogeneity of variances to determine the differences between groups. Each group was given a letter to show similarities (P > 0.05) in the comparison results: Sucrose‐sweetened cake was given the letter ‘a’ and denoted SCR^a; stevia‐sweetened cake was given the letter ‘b’ and denoted STV^b; agave syrup‐sweetened cake was given the letter ‘c’ and denoted AGV^c; date syrup‐sweetened cake was given the letter ‘d’ and denoted DAT^d; and apple juice concentrate‐sweetened cake was given the letter ‘e’ and denoted APL^e. Letters above mean values indicate that the value is similar to the group to which it belongs.

RESULTS

Significant differences (P < 0.001 for all) were found between the groups in all values (L*, a* b*, C*, h° and BI) for both crumb and crust colour results (Table 2). The lowest mean ΔE value belongs to the STV group for cake crumb colour and DAT group for crust colour. The mean C* values of the STV and DAT groups for cake crumb and crust colour were similar to the control group (P > 0.05). All of the mean BI values for cake inside colour were different from each other (P < 0.05). The mean crust BI value of the STV group was significantly lower than the other groups (P < 0.05).

Table 2.

Values of crumb and crust color analysis of cakes (n = 8)

	SCRa (x ± SS)	STVb (x ± SS)	AGVc (x ± SS)	DATd (x ± SS)	APLe (x ± SS)	P *
Crumb color analysis of cakes
L *	61.33 ± 0.99	50.23 ± 2.51	41.95 ± 1.17e	47.62 ± 1.79	42.19 ± 1.23 c	< 0.001
a *	2.45 ± 0.27	3.02 ± 0.32	7.03 ± 0.09	5.63 ± 0.20	6.15 ± 0.21	< 0.001
b *	18.51 ± 0.63 b	19.47 ± 1.57 ad	16.14 ± 0.43 e	17.57 ± 0.40b	15.77 ± 0.48 c	< 0.001
ΔE	–	11.28 ± 1.77	20.08 ± 1.05	14.14 ± 2.00	19.71 ± 1.12	–
C *	18.67 ± 0.62 bd	19.70 ± 1.55 acd	17.60 ± 0.39 be	18.45 ± 0.35 ab	16.93 ± 0.49 c	< 0.001
h *	82.46 ± 0.90	81.13 ± 1.19	66.46 ± 0.65	72.23 ± 0.87	68.67 ± 0.61	< 0.001
BI	5.86 ± 0.33	8.21 ± 0.79	15.66 ± 0.33	12.10 ± 0.62	14.05 ± 0.39	< 0.001
Crust color analysis of cakes
L *	37.95 ± 2.15 de	44.62 ± 3.18	31.86 ± 1.21 e	37.07 ± 2.40 ae	34.74 ± 2.15 acd	< 0.001
a *	16.31 ± 0.69	12.17 ± 0.77	13.30 ± 0.78 d	14.26 ± 0.78 ce	14.44 ± 0.66 d	< 0.001
b *	17.60 ± 1.06 de	20.52 ± 1.03	14.04 ± 0.64	17.18 ± 2.01 ae	16.49 ± 1.31 ad	< 0.001
ΔE	–	8.67 ± 3.43	7.90 ± 2.96	4.44 ± 1.83	4.80 ± 2.39	‐
C *	24.01 ± 0.97 bd	23.87 ± 0.88 ad	19.35 ± 0.88	22.36 ± 1.78 abe	21.95 ± 0.78 d	< 0.001
h *	47.14 ± 1.94 cde	59.29 ± 2.26	46.56 ± 1.55 ade	50.19 ± 3.17 ace	48.65 ± 3.29 acd	< 0.001
BI	33.78 ± 2.31 cde	23.61 ± 1.84	32.67 ± 1.35 ade	30.96 ± 2.94 ace	32.99 ± 1.39 acd	< 0.001

SCR, sweetened with sucrose; STV, sweetened with stevia; AGV, sweetened with agave syrup; DAT, sweetened with date syrup; APL, cake sweetened with apple juice concentrate; ΔE, color change; C*, chroma; h*, hue angle; BI, browning index.

^*One‐way ANOVA analysis was used. Differences between groups were determined by Tukey/Tamhane multiple comparison tests according to variance homogeneity. For whichever group a lowercase letter belongs to, the value is similar to that of that group (P > 0.05).

Some physical properties of the cakes are given in Table 3. Accordingly, the baking loss was significantly lower in the SCR group than the other groups (P < 0.05). There is a significant difference between the groups in terms of volume index and symmetry index means (P < 0.001 for both). This significant difference is a result of the mean of the volume index and symmetry index of the SCR group being significantly higher than the other groups according to the multiple comparison test result (P < 0.05). The SCR group has the lowest hardness value and this difference is statistically significant (P < 0.05). The AGV group had the highest mean hardness value and the mean hardness value of the AGV group was similar to the mean hardness value of the APL group (P > 0.05).

Table 3.

Physical properties of the cakes (n = 4)

	SCRa (x ± SS)	STVb (x ± SS)	AGVc (x ± SS)	DATd (x ± SS)	APLe (x ± SS)	P *
Baking loss (%)	10.85 ± 0.33	13.24 ± 0.39 ce	12.34 ± 0.56 bde	12.03 ± 0.67 ce	12.27 ± 0.5 bcd	< 0.001
Volume index	166.50 ± 4.04	135.00 ± 8.12 cde	119.25 ± 0.96 bde	124.75 ± 5.06 bce	115.50 ± 2.08 bcd	< 0.001
Symmetry index	9.00 ± 1.15	3.00 ± 1.41 cde	−1.50 ± 1.29 bde	0.50 ± 4.04 bce	0.75 ± 2.75 bcd	< 0.001
Uniformity index	−0.50 ± 1.91 bcde	6.50 ± 12.45 acde	0.00 ± 2.16 abde	0.50 ± 2.38 abce	0.25 ± 2.87 abcd	0.462
Hardness (g)	537.65 ± 53.73	814.56 ± 137.06	1555.95 ± 202.84 e	1241.42 ± 137.80 e	1342.42 ± 250.15 cd	< 0.001

SCR, sweetened with sucrose; STV, sweetened with stevia; AGV, sweetened with agave syrup; DAT, sweetened with date syrup; APL, cake sweetened with apple juice concentrate.

^*One‐way ANOVA analysis was used. Differences between groups were determined by Tukey/Tamhane multiple comparison tests according to variance homogeneity. For whichever group a lowercase letter belongs to, the value is similar to that of that group (P > 0.05).

The pH values of the SCR group and the STV group were similar (P > 0.05) and significantly higher than the other groups (P < 0.05). There was no significant difference between AGV, DAT and APL groups in terms of pH value (P > 0.05) (Table 4). The STV group had the highest water activity value and the SCR group had the lowest (P < 0.05 for both) (Table 4).

Table 4.

Results of pH value and water activity analysis of cakes (n = 4)

	pH (x ± SS)	P *	aw (x ± SS)	P *
SCRa	7.27 ± 0.06 b	< 0.001	0.8673 ± 0.0083	< 0.001
STVb	7.34 ± 0.02 a		0.9380 ± 0.0033
AGVc	6.87 ± 0.02 de		0.9000 ± 0.0022 e
DATd	6.87 ± 0.02 ce		0.9143 ± 0.0013
APLe	6.92 ± 0.07 cd		0.9005 ± 0.0037 c

SCR, sweetened with sucrose; STV, sweetened with stevia; AGV, sweetened with agave syrup; DAT, sweetened with date syrup; APL, cake sweetened with apple juice concentrate.

^*One‐way ANOVA analysis was used. Differences between groups were determined by Tukey/Tamhane multiple comparison tests according to variance homogeneity. For whichever group a lowercase letter belongs to, the value is similar to that of that group (P > 0.05).

The values of the basic chemical components of the cakes are presented in Table 5. The STV group had the highest mean moisture content and the SCR group had the lowest (P < 0.05 for both). The carbohydrate content was significantly the highest in the SCR group (P < 0.05). The protein content of the SCR group was similar to that of all other groups (P > 0.05). The STV group had the highest fat content and the fat content of the STV group was similar to that of the AGV group (P > 0.05). The mean energy value of the SCR group was significantly higher than the mean energy value of the other groups (P < 0.05).

Table 5.

Chemical composition and energy value of cakes (n = 4) ^a

	SCR a (x ± SS)	STVb (x ± SS)	AGVc (x ± SS)	DATd (x ± SS)	APLe (x ± SS)	P *
Moisture (g)	24.19 ± 0.08	35.80 ± 0.08	32.10 ± 0.38 de	33.04 ± 0.27 ce	33.28 ± 0.05 cd	< 0.001
Ash (g)	1.22 ± 0.03 de	1.64 ± 0.04 de	1.33 ± 0.03 de	1.47 ± 0.12 abce	1.46 ± 0.11 abcd	< 0.001
CHO (g)	46.37 ± 0.89	24.17 ± 2.37 e	37.98 ± 0.57 de	33.51 ± 2.16 e	33.09 ± 3.28 bcd	< 0.001
Fiber (g)	4.08 ± 0.36 bcd	5.53 ± 0.68 acde	3.86 ± 0.12 ab	4.54 ± 0.22 ab	5.31 ± 0.15 b	< 0.001
Protein (g)	6.17 ± 0.52 bcde	7.52 ± 0.24 a	6.27 ± 0.27 ade	6.67 ± 0.15 ac	6.19 ± 0.13 ac	< 0.001
Fat (g)	17.97 ± 0.77 cde	22.85 ± 1.28 c	18.46 ± 0.09 abde	18.53 ± 0.70 ace	18.17 ± 0.23 acd	< 0.001
Energy (kcal)	379.75 ± 4.92	353.52 ± 7.57 cde	350.75 ± 2.06 bd	345.50 ± 4.20 bce	341.25 ± 0.50 bd	< 0.001

SCR: Sweetened with sucrose, STV: Sweetened with stevia, AGV: Sweetened with agave syrup, DAT: Sweetened with date syrup, APL: Cake sweetened with apple juice concentrate. CHO: Carbohydrate.

^a Data are presented on the basis of wet weight.

^*One‐way ANOVA analysis was used. Differences between groups were determined by Tukey/Tamhane multiple comparison tests according to variance homogeneity. For whichever group a lowercase letter belongs to, the value is similar to that of that group (P > 0.05).

The HMF contents of the substances used as sugar substitutes were significantly different from each other (P < 0.001). According to the multiple comparison test result, the mean HMF content of date syrup was significantly higher than the other substitutes (P < 0.05). Although liquid stevia had the lowest HMF content, statistically, the HMF content was similar to agave syrup and apple juice concentrate (P > 0.05). The HMF content of the DAT group was significantly higher than the other groups (P < 0.05). STV group contains significantly lower amount of HMF compared to the control group (P < 0.05). The HMF contents of AGV and APL groups were similar to the control group (P > 0.05) (Table 6). There was a statistically significant positive correlation between HMF and the BI of cake crust (r = 0.420, P = 0.009) and a statistically significant negative correlation between HMF and crust L* value (r = −0.493, P = 0.002) (data not presented).

Table 6.

HMF levels of sugar substitutes and cakes (n = 4)

	HMF Substitutions mg kg−1 (x ± SS)	P *	HMF cake mg kg−1 (x ± SS)	P *
SCRa	–	< 0.001	61.48 ± 11.78 ce	< 0.001
STVb	2.87 ± 0.21 ce		16.70 ± 4.18 ce
AGVc	382.55 ± 9.55 be		152.03 ± 39.38 abe
DATd	3457.00 ± 49.69		405.55 ± 29.70
APLe	324.88 ± 15.80 bc		220.61 ± 42.32 abc

SCR, sweetened with sucrose; STV, sweetened with stevia; AGV, sweetened with agave syrup; DAT, sweetened with date syrup; APL, cake sweetened with apple juice concentrate.

^*One‐way ANOVA analysis was used. Differences between groups were determined by Tukey/Tamhane multiple comparison tests according to variance homogeneity. For whichever group a lowercase letter belongs to, the value is similar to that of that group (P > 0.05).

The mean scores of all parameters (appearance, crumb colour, crust colour, flavour, texture, general taste) questioned in the sensory analysis were significantly different between the groups (P < 0.001 for all). According to the results of the multiple comparison test, the SCR group had the highest score in all parameters except for crust colour (P < 0.05). Although the crust colour score of the SCR group and the STV group were similar (P > 0.05), they were significantly higher than the other groups (P < 0.05 for all) (Table 7).

Table 7.

Results of sensory evaluation of cakes ^a (according to a nine‐point hedonic scale)

	Cake type
	SCR a (x ± SS)	STVb (x ± SS)	AGVc (x ± SS)	DATd (x ± SS)	APLe (x ± SS)	P *
Appearance	8.20 ± 1.24	6.70 ± 1.34	5.30 ± 1.86 de	5.03 ± 1.88 ce	5.03 ± 1.65 cd	< 0.001
Crumb color	8.30 ± 0.99	6.80 ± 1.49	5.10 ± 1.65 de	4.93 ± 1.53 ce	4.83 ± 1.44 cd	< 0.001
Crust color	8.07 ± 1.34 b	7.07 ± 1.34 a	5.80 ± 1.47 de	5.40 ± 1.79 ce	5.67 ± 1.40 cd	< 0.001
Taste	7.90 ± 1.35	6.33 ± 1.18	5.30 ± 1.47 de	4.47 ± 1.55 ce	4.67 ± 1.35 cd	< 0.001
Texture	8.07 ± 1.26	6.20 ± 1.52	4.90 ± 1.75 de	4.70 ± 1.93 ce	4.70 ± 1.76 cd	< 0.001
General rating	8.03 ± 1.16	6.30 ± 1.29	5.00 ± 1.70 de	4.50 ± 1.78 ce	4.43 ± 1.48 cd	< 0.001

SCR, sweetened with sucrose; STV, sweetened with stevia; AGV, sweetened with agave syrup; DAT, sweetened with date syrup; APL, cake sweetened with apple juice concentrate.

^a Sensory evaluation was performed with the same participants (n = 15) in two replicates.

^*One‐way ANOVA analysis was used. Differences between groups were determined by Tukey/Tamhane multiple comparison tests according to variance homogeneity. For whichever group a lowercase letter belongs to, the value is similar to that of that group (P > 0.05).

DISCUSSION

Browning of cake crust is associated with caramelisation of sugars and the Maillard reaction. ⁴³ Both caramelisation and the Maillard reaction occur in the presence of reducing sugars. ⁴⁴ Agave syrup, apple juice concentrate and date syrup contain these sugars. ⁴⁵ , ⁴⁶ , ⁴⁷ Reducing sugars are formed by thermal hydrolysis of sucrose. Sucrose also contributes to crust colour by direct caramelisation. ⁴⁴ Considering that rebaudioside A is very stable under thermal conditions and stevia does not participate in browning and Maillard reactions, it is expected that the crust colour of the cake sweetened with stevia will have the lowest BI value. ⁴⁸ It has been stated that the inclusion of substances with high sweetening properties in the formulation with reducing sugars may lead to negligible changes in cake stability and texture and may improve the crust colour. For this reason, formulations in which reducing sugars are added to the cake sweetened with liquid stevia can be developed to produce the desired brown colour in the crust colour. ⁴⁹ Gao et al. ⁵⁰ showed in their study that cocoa powder reduced the crust colour change caused by stevia. Based on their study, it is suggested that products such as cocoa are added to the standard recipe to make the cake colour similar to the control group in future studies and ensure that the crust colour is similar to the control group.

The internal temperature of the cake does not reach a sufficient level for Maillard and caramelisation during baking. Therefore, the crumb colour of the cake is largely determined by the colour of the ingredients in its formulation. ⁵¹ , ⁵² Considering that the colour of each of the sugar substitutes used is different from each other, it is expected that the L*, a*, b* values of the crumb colour of the cakes produced with these substitutes will be different from the control group. In addition, differences in the total batter weights of the cake formulations may have caused dilution or concentration of the factors (substitute colour, egg colour, etc.) that are effective in the formation of the crumb colour of the cake.

Hardness is one of the most important quality parameters which can be detected by sense of touch and affects the quality of consumption especially in bakery products. ⁵³ , ⁵⁴ When the hardness value of the cakes was examined, the lowest value belonged to the cake flavoured with sucrose. Sucrose has a softening effect on cakes. ⁵⁵ Therefore, it is not a surprising result that the hardness value of sucrose‐free cakes was higher than the control group. Similar to the present study, in a study in which sucrose was substituted with stevia at a rate of 100%, the hardness values of the cakes increased significantly. ⁵⁶ It has also been shown that the increase in the substitution ratio of date syrup to sucrose leads to an increase in the hardness values of cakes. ⁵⁷ , ⁵⁸ However, in another study, the hardness value for all ratios in which sucrose was substituted with agave syrup at a ratio of 25%–50%–75%–100% was found to be significantly lower compared to the control group. ⁵⁹ Similarly, in another study in which the hardness of cakes produced using date and fig syrups were examined, the hardness value of the cakes using syrup was found to be significantly lower compared to the sucrose group. The reason for this situation was stated by the researchers as ‘the organic acids in syrups are effective in hydrolysis of starch and gluten proteins and softening the texture of the cake’. ⁵² From this point of view, the contradictory results in the literature may be related to the purity level of the syrups used or the difference in the chemical components they contain.

Hardness is also closely related to the volume of the cake and therefore the total volume of air contained in the cake. As the cell size of the air bubbles trapped in the cake increases, the product has a lower density and a softer structure. ⁶⁰ When the volume indexes of the cakes in the present study were examined, the volume index value of the sucrose‐sweetened cake, which had the lowest hardness value in parallel with this information, was found to be significantly higher than all cakes using sugar substitute (P < 0.05). The presence of sucrose in the environment is directly related to bakery product volumes. ⁶¹ , ⁶² , ⁶³ Sucrose increases the gelatinisation temperature of starch from 57 °C to 92 °C and thus increases the temperature at which the cake hardens. Thus, the air bubbles are appropriately expanded by carbon dioxide and water vapor before the cake hardens, providing the desired volume in the cake. ⁶⁴ , ⁶⁵ Sucrose also increases the viscosity of the cake batter as a result of its hygroscopic properties. A high viscosity ensures that air bubbles are trapped in the cake and the volume of the final product increases by filling with air. ⁶⁶ For this reason, it is of great importance to develop formulations that include bulking agents (such as isomalt) with high water retention capacity in bakery products to be substituted with sucrose. ⁶⁷ Palamutoğlu et al. ⁶⁸ reported that the substitution of sucrose with stevia at ratios of 25%–50%–75% did not cause a significant difference between the volume indices of cakes. Based on their study, the use of sugar and sugar substitutes together as a result of substituting sucrose at certain ratios instead of 100% substitution of sucrose can eliminate the negativities that will be seen in the volume of the cake.

The symmetry index gives an idea of the volume of air trapped in the cake at the final baking stage. ⁶⁹ A positive symmetry index indicates that the cake is curved, whereas a negative symmetry index indicates that the cake is sunken. ⁷⁰ The only cake sample that was found to have sinking in our study was the one flavored with agave (−1.50 ± 1.29). The cake with the highest symmetry index (most curved surface) was the control group and this difference was statistically significant (P < 0.05).

The water activity value of all cakes with sugar substitutes was found to be significantly higher than that of cakes sweetened with sucrose (P < 0.05). Sucrose has the effect of decreasing water activity as a result of its hygroscopic property. ²³ The fact that the sugar substitutes were made to be 100% and also that the sugar substitutes used were in liquid form may have caused the a_w value of the cakes containing sugar substitutes to be higher compared to the control group. The cake with the highest a_w value was the group sweetened with stevia (P < 0.05). Similarly, it was found that the a_w value of stevia‐containing cakes was higher than sucrose‐containing cakes and the increase in the substitution ratio of sucrose with stevia resulted in an increase in a_w value. ⁶⁸ , ⁷¹ It has been reported that the use of stevia in home‐baked bakery products may cause a decrease in shelf life and additional recommendations should be provided to the consumer regarding the storage conditions of products containing stevia. ⁷² It can be concluded that this food safety information is valid for all cakes in the present study in which sugar substitutes were used (as a result of the high aw value).

The energy values per gram of sugar substitutes are lower than sucrose. In this way, a decrease between 6.86% and 10.03% was observed in the energy value of cakes containing sugar substitutes compared to the control group. However, although the energy value of stevia was equal to zero, unlike the other sugar substitutes, the cake sweetened with stevia had a similar energy value to the cakes sweetened with other sugar substitutes (P > 0.05). This is caused by the different total batter weights of the cakes. In other words, the substitution of sucrose with stevia decreased the amount of carbohydrates in 100 g at the same time as increasing the amount of fat. Considering that 1 g of fat provides 9 kcal and 1 g of carbohydrate provides 4 kcal energy, it becomes understandable that the energy value of the cake sweetened with stevia is similar to other sugar substitutes. In a previous study, ⁷³ as a result of substitution of sucrose with 75% liquid stevia, the fat content increased significantly and the energy content decreased by 5.14% compared to the control group (sucrose), similar to the present study. Unlike the present study, it was observed that the energy value decreased by 24.01% compared to the control group as a result of keeping the fat content constant for 100 g samples of cakes and substituting sucrose with 100% stevia. ⁷⁴ In brief, it is assumed that the decrease in the amount of energy realised as a result of the use of sugar substitutes in cakes will become more significant with formulation changes or fat content corrections considering the dough weights.

Because stevia is a sweetener that is not fermented, does not undergo caramelisation, does not contain reducing sugars, and is stable in a wide pH (3–9) and temperature range, low HMF content is an expected result. ⁷⁵ , ⁷⁶ In parallel with this, the cake with the lowest HMF content in the present study was the cake sweetened with stevia. In addition to this, 100% substitution of stevia to sucrose reduced the HMF content to approximately one‐quarter of that of sucrose‐sweetened cake. This finding is similar to previous studies in the literature. ⁷⁶ , ⁷⁷ , ⁷⁸ The sugar substitute with the highest HMF content is date syrup. Studies have found that date syrups produced at higher temperatures have a higher HMF content. ⁷⁹ , ⁸⁰ Because of its toxic properties, HMF is used as an indicator of heat stress in sugar‐based foods such as honey and syrup. ⁸¹ The high HMF content in date syrup suggests that the product was not exposed to appropriate conditions during the production, storage and transportation stages and that the dates used may comprise an old crop. Factors such as the presence of reducing sugars and amino acids, high water content, low pH, and the type of dates also affect the amount of HMF in date syrup. ⁸² The company supplying the date palm juice in the present study stated that they prepare their product using the traditional method (high temperature). This resulted in an increase in HMF content.

The HMF content of cakes sweetened with all sugar substitutes except stevia was higher compared to the control group. Similarly, the use of honey, corn syrup, date syrup and cane molasses instead of sucrose in biscuits baked at 225 °C caused the HMF content to increase 6.4–42.6 times. ⁸³ It is known that glucose and fructose tend to produce more HMF than sucrose and also that fructose is the most effective monosaccharide in HMF increase. ⁸⁴ In a study conducted aiming to determine the sugar content of date syrup, the amount of fructose was in the range 23.3–32.5%, glucose was in the range 21.8–29.8% and sucrose was in the range 3.28–18.7% with respect to dry weight. ⁴⁵ It was reported that the main carbohydrate of agave syrup was fructose with an mean rate of 84.29%. ⁸⁵ The sugar contents of apple juices, which are frequently used in apple juice concentrate production in our country, vary between 9.30% and 32.2% for glucose, 66.10% and 96.00% for fructose and 8.5% and 55.10% for sucrose. ⁸⁶ As can be seen from the studies, the high fructose and glucose contents of agave syrup, date syrup and apple juice concentrate are the most probable reason for the increase in HMF content in cakes. The already high HMF content of these products, to which consumers turn when aiming to reduce their sucrose consumption, may cause higher HMF formation as a result of baking.

In many studies, the HMF content in bakery products was found to be positively correlated with browning index and negatively correlated with L* value. ⁸⁷ , ⁸⁸ , ⁸⁹ Although there was a statistically significant correlation between HMF and browning index and L* value in the present study, this relationship was moderate. The variability of the ingredients in the formulations and their contribution to the color may explain the lack of correlation between heat treatment and color. In addition, the wide range of HMF in the cake samples in the present study (16.70–405.55 mg kg⁻¹) may have made it difficult to find a stronger correlation.

The present study has some limitations. When determining the ratios of sugar substitutes used in the present study, the manufacturer's recommendations were taken into consideration. As understood from the feedback after the sensory analysis, the sweetness ratios of the cakes were different from each other. The fact that the sweetness ratios were not equal is one of the limitations of our study. Studies in which sweetness ratios are equal can provide more objective results in sensory terms. Another limitation of the present study is that no evaluation of odor during sensory analysis was included. The unique odors of sugar substitutes may have affected the sensory analysis scores.

CONCLUSIONS

The lowest HMF level among sugar substitutes belonged to liquid stevia. Among the cakes, including the control group, the lowest HMF level was found in the cake sweetened with liquid stevia and the highest HMF level was found in the cake sweetened with date syrup. The HMF contents of all cakes sweetened with sugar substitutes except liquid stevia were found to be higher compared to the control group. Therefore, when using such sugar substitutes in bakery products based on the claim that they are healthier, attention should be paid to the increase in HMF content, which is considered one of the most important chemical pollutants. When the studies conducted aiming to determine the HMF content in foods are examined, it is seen that very different results are obtained. Therefore, comprehensive studies are needed to determine and reduce the HMF content in foods. It is assumed that it is necessary to establish legal regulations to ensure that such syrups contain HMF at a level that does not endanger human health.

Considering all chemical, physical and sensory analyses, liquid stevia appears to be the most suitable sugar substitute for cakes among the sugar substitutes used in the present study. The carbohydrate and energy value of the cake decreased with the use of liquid stevia. In addition, the HMF level was also found to be lower compared to the cake using sucrose. Further studies are needed to investigate the possibilities of using stevia in cakes and to develop more suitable formulations.

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available from the corresponding author upon reasonable request.