Four different formulations of dough prepared with: (i) buckwheat flour, (ii) chestnut flour, (iii) buckwheat flour and chia seeds, and (iv) chestnut flour and chia seeds were used for the production of gluten-free biscuits. For these formulations, buckwheat flour (Ecology Market Food Industry Trade Co., Istanbul, Turkey), chestnut flour (Naturelka CC Tourism Co., Aydın, Turkey), potato flour (Sakarya Agricultural Products, Sakarya Turkey), chia seeds (Güzel Ada Food, Istanbul, Turkey) lactose-free milk (Pınar Dairy, Izmir, Turkey) were provided by the respective companies. In addition, margarine, sugar and whole eggs were purchased from a local grocery store in Turkey.

Ethanol (≥99.8%), acetone (≥99%) and copper were from Delta Agricultural Chemicals Industry and Trade Inc. (Antalya, Turkey), sulfuric acid, sodium hydride, petroleum ether, hydrochloric acid, boric acid, potassium chloride, sodium bicarbonate, aluminium chloride and trifluoroacetic acid were from Merck (Darmstadt, Germany). Gallic acid, 2,2-diphenyl-1-picrylhydrazyl (DPPH) and Trolox ((±)-6-hydroxy-2,5,7,8-tetramethylchromane-2-carboxylic acid) used in this work were from Sigma-Aldrich Co. (Merck, St. Louis, MO, USA).

In this study, four different types of gluten-free biscuit formulations were produced according to a modified method by Öksüz and Karakaş (19). The details are depicted in Table 1. The ingredients were mixed to form the dough and rolled to a thickness of 0.1 cm. Then the biscuits were moulded with a cutter of 50 mm diameter and baked in a conventional oven at 150 ºC for about 25 min. Afterwards, they were left to cool down to room temperature for about 1 h and then kept in airtight containers.

Total moisture, ash, protein and fat contents were determined according to the AOAC methods. The total moisture content was determined by measuring the mass loss after drying for 1 h at 130 ºC according to the AOAC method 925.10 (gravimetric air oven method) (20). The total ash content was determined according to the gravimetric AOAC method 923.03 by ashing the samples in the muffle furnace for 16 h at 600 ºC (21). Total protein content of the samples was measured according to Kjeldahl method (AOAC 920.87) with the factor of 5.7 for converting the total nitrogen into total protein (22). Soxhlet (gravimetric) method (AOAC 945.16) was used to extract the oil from the samples and measure its content (23).

The total amount of dietary fibre was determined using the Total Dietary Fiber Assay Kit (TDF100A; Sigma-Aldrich, Merck), based on AOAC method 985.29 (enzymatic-gravimetric method) (24). The total content of carbohydrates was calculated by subtracting the total amounts of moisture, protein, ash and fat constituents of the sample from 100. The obtained value presents the carbohydrate content of the sample (g/100 g). The total energy (kcal/100 g) was calculated using the conversion factors 9 for each g of fats, and 4 for each g of carbohydrates and proteins.

A slightly modified extraction procedure was applied to prepare the biscuit extracts for measuring the total phenolic content and antioxidant activity (25). In summary, the samples (2 g) were first extracted by adding 15 mL of 80% methanol and then shaken on an orbital shaker (Model SSL1; Cole-Parmer, Staffordshire, UK) for 15 min (4000×g), sonicated (model PMUY-4L-D; Protech Mechanical Equipment Co, Ltd., Zhengzhou, PR China) for 15 min and centrifuged (model Rotofix 32A; Hettich, Tuttlingen, Germany) at 4000×g for 20 min. The extraction was repeated until 30 mL of extract were collected in two replicates for each sample.

The diameter and thickness of the biscuit samples were measured using a digital caliper. Bulk density was measured by dividing the mass of each sample by its volume. The spread ratio was measured by dividing the diameter (D) by the thickness (T). All measurements were made in triplicates. The volume of the biscuits was measured using the rapeseed displacement method by comparing the change in the known volume of rapeseed with each biscuit sample, as described in AACC method 10-05.01 (27) and water absorption test was performed according to the AACC method 56-30.01 (28). In the latter, ground samples (0.2 g) were weighed into pre-weighed centrifuge tubes and 10 mL of distilled water were added. The tubes were kept in a water bath at 30 °C for 30 min, vortexed for 5 s every 5 min and centrifuged at 3000×g (Hettich Rotofix 32A; Merck) for 10 min. The percentage of absorbed water was measured by weighing the sediment.

The colour of the biscuit samples was measured using the colour chromameter (CR-400; Konica Minolta Inc., Tokyo, Japan), using the Hunter scale of L*, a*, b*. Six samples from each biscuit formulation were measured and the average was calculated. Colour parameters for the standard white plane were as follows: L₀=95.9, a₀=0.2 and b₀=2.3.

The difference between the colour parameters of darkness/lightness (L* values), redness (a* values) and yellowness (b* values) of the biscuit samples and the standard white plane parameters given above were calculated to measure the total colour change (ΔE) according to the equation below:

Hardness was measured using TA.XTplus Texture Analyzer (Stable Micro Systems, Surrey, UK), fitted with a 50-kg load cell and coupled with Texture Expert Software, v. 1.22 (29). Measurements were made approx. 24 h after baking to measure the biscuit fracture, and each biscuit was compressed by a three-point bending rig (HDP/3PB) and blade (90 mm long and 3 mm thick) at a crosshead speed of 3 mm/s. The biscuits were 10 mm thick. The maximum force (N) measured in the force-time (distance) curve was taken as the value of hardness (30).

Sensory properties of biscuit samples were evaluated by 10 panellists. The panellists were females, aged between 25 and 35 and they were semi-trained to evaluate and compare the samples for parameters of appearance, colour, smell, taste texture/mouth feel and overall acceptance, on a 9-point hedonic scale ranging from 1 for dislike extremely to 9 for like extremely, according to a modified method of Moretti et al. (31). In the preparatory session, the panel members had a thorough discussion in order to clarify each attribute before their evaluation. For each parameter, the average of the relevant scores given by the panellists was calculated.

All biscuit samples were stored at room temperature (around 22-23 ºC) in locked sterile bags for a period of 45 days. Moisture content of the samples was measured as described above, peroxide value (PV) was determined by iodometric method ISO 3960:2017 (32). In addition, total acidity was determined by measuring the total amount of treatable acids. Their pH values were measured on the 1st, 20th and 45th day. Fatty acid composition of the freshly baked biscuit samples and the samples that were stored for 6 months was evaluated with gas chromatography-flame ionization detector (GC–FID) model CLARUS 580 (Perkin Elmer, Shelton, CT, USA) and a BPx70 column (30 m, 0.25 mm, 0.25 mm film thickness) (Trajan Scientific and Medical, Ringwood, VA, Australia). The fatty acid content of the biscuits was expressed as the percentage of total oil content.

The data were analysed using the SPSS software, v. 18 (33). Analysis of variance (ANOVA) was used for comparison. The Duncan’s test was used as the post-hoc test to determine the differences. Differences between samples were calculated at 95% significance level (34).

According to the findings presented in Table 2, raw materials had significant differences (p<0.05) in terms of proximate components. Buckwheat flour had significantly higher amount of protein (11.4%) than the chestnut and potato flour (6.7 and 5.2% respectively), but lower (p<0.05) than that of the chia seeds (23.1%). Mass fractions of ash (1.88%), fat (1.7%) and carbohydrates (72.4%) in buckwheat flour were not significantly different than in the other flour sources (p>0.05). Similar to our findings, the moisture, protein, ash and fat mass fractions of buckwheat flour that was used as the raw material for cookies with rice flour were 11.3, 12.3, 2.2 and 2.9%, respectively in a study by Torbica et al. (35). In another study, buckwheat flour carbohydrate mass fraction was reported to be 69.7% (36).

Chestnut flour had significantly lower (p<0.05) protein mass fraction (5.2%) than the other ingredients. This finding is commensurate with previous findings reported in the literature, since chestnut flour reported by Šoronja‐Simović et al. (37) and Lopes et al. (38) contained 5.3 and 5.7% protein, respectively. Its fat and carbohydrate mass fractions were statistically different from buckwheat flour (p>0.05). As expected, buckwheat flour and chestnut flour also had higher mass fractions of total carbohydrates than chia seeds (p<0.05). Chestnut flour had a lower moisture (7.7%) and ash (2.1%), but similar protein (5.7%), fat (1.6%), dietary fibre (20.7%) and carbohydrate (62.3%) mass fractions in an earlier study (38).

Total mass fractions of protein (6.7%), ash (2.8%) and carbohydrates (81.1%) in potato flour were slightly different with respect to the other flour sources; although the fat mass fraction was significantly higher than the other ingredients (except for the chia seeds) and the moisture mass fraction was lower (p<0.05). In a recent study (11), potato flour was also used as a raw material for another gluten-free product and reported to contain 9.17% protein, 3.86% ash, 0.26% fat and 20.51% dietary fibre. Therefore, proximate mass fractions were generally higher than in the potato flour used in this study. In contrast, the potato flour used in another study (39) was reported to have 5.47% protein, 2.90% ash, 79.83% total carbohydrate and 5.29% dietary fibre, thus being more similar to the raw material presented in this study.

Chia seeds had significantly higher protein (23.1%), ash (4.58%) and dietary fibre (34.4%), but lower carbohydrate (36.2%) content than the flour sources (p<0.05) used in this study. Comparable to the current findings, chia seeds were reported to have 9.3% moisture, 25.1% protein, 5.5% ash and 26.2% fat (40). Total dietary fibre in chia seeds and chia seed flour was indicated as quite similar to (around 37.4%) (41), or slightly lower (29.6 and 30.1%) (42), respectively than our findings (34.4%).

According to the results of the proximate component analyses of biscuit samples (Table 3), the compositional differences in raw materials affected the proximate compositions of the biscuits, as expected (p<0.05). Protein mass fraction in different biscuit samples was significantly different (p<0.05). The biscuits baked with chestnut flour had the lowest mass fraction of protein on dry mass basis (4.8%), which is in accordance with the relatively lower protein mass fraction of the chestnut flour. Biscuits with buckwheat and chia had the highest mass fraction of proteins (8.2%). Buckwheat flour had significantly higher mass fraction of protein (11.4%; Table 2) than the chestnut and potato flour, although it was lower than that of the chia seeds. Therefore, incorporation of buckwheat flour in gluten-free formulations was found to be an effective method for increasing their protein mass fraction. Similar results have been reported in a study using wholegrain buckwheat flour in gluten-free crackers (43). The increase in total protein mass fraction with chia seed flour was also similar to other studies, as the addition of 5 to 20% chia seed flour in the study by Kaur et al. (36) resulted in increasing mass fractions of measured protein in biscuit samples, ranging from 9.7 to 11.7%.

No significant differences (p>0.05) were detected in the total ash and moisture mass fractions in different biscuit samples (Table 3). On the other hand, the fat mass fraction in the biscuits was mainly affected by the addition of chia seeds, and the biscuits with chia seeds had significantly higher mass fractions of fat (p<0.05). Increase in the total dietary fibre and fat mass fractions in biscuit samples with chia seeds were also evident in previous study (42). Previous findings on the nutritional properties of gluten-free products that combine chia seeds and buckwheat flour highlighted that these products had health-promoting benefits for obese and diabetic individuals by lowering their glycaemic index, increasing satiety, in addition to being gluten-free (44). Among all biscuit samples, the dietary fibre mass fraction was the lowest in biscuits with buckwheat (p<0.05). This finding was in parallel with low mass fraction of dietary fibre in buckwheat flour.

The use of ingredients with relatively high amount of phenolic compounds in gluten-free products has a significant benefit for coeliac patients due to their ability to bind to fibre moieties, thus making them readily absorbable in the colon. In addition to the direct antioxidant activity of phenolic compounds, this binding ability was proposed as a beneficial mechanism for coeliac patients by modifying the composition and immune function of the gut microbiota (45).

The total phenolic content (expressed as gallic acid equivalents (GAE) on dry mass basis) of the biscuit samples changed from 248.2 to 400.2 mg/100 g (Table 3). Biscuits with chestnut had the highest phenolic content among all samples (400.2 mg/100 g). Biscuits with buckwheat also contained relatively high amounts of phenolics (387.0 mg/100 g). A previous review on chestnuts reports high but varying contents of total phenolics (mainly as gallic and ellagic acid) in chestnut fruits (280–910 and 610–2560 mg/100 g) (46).

Chia seed supplementation to the biscuit formulation, on the other hand, significantly (p<0.05) decreased the total phenolics (Table 3). The amount of chia seed incorporation may affect the total phenolic content. In a previous research, total phenolic content was 156.99 µg/g in cooked durum wheat pasta, but lower when 5% chia seeds were added into samples (123.53 µg/g) and higher when 10% chia seeds were added into samples (186.80 µg/g) (8). In addition, the relative decrease in total phenolic content after chia seed supplementation may be related to the richness of chia seeds not only in the nutritional compounds but also in the non-nutritional healthy compounds (44).

Total antioxidant activity (TAA), expressed as Trolox equivalents (TE) on dry mass basis, of the biscuits varied between 78.2 and 155.5 mg/100 g (Table 3). In parallel with total phenolics, biscuits with chestnut had the highest TAA and the addition of chia seeds significantly reduced the measured TAA. In the literature, increases in the measured level of antioxidant activity after baking the biscuits containing buckwheat flour or chestnut flour were also observed and this was explained as the effect of Maillard reaction products formed during heat treatment (2, 47). Relatively low antioxidant activity of chia seeds was also reported and related to the presence of hydrophilic phenolic acids in the seeds (48).

Table 3 shows the quality parameters of gluten-free biscuits. Although no significant differences were detected in the physical parameters of diameter, height and mass of the biscuit samples (p>0.05), chia reduced the volume of the biscuits slightly. This tendency to decrease specific volumes was also evident in bread samples with chia seeds and in a previous study, protein network interruption was presented as the main reason for this change (49). However, the gluten network development in biscuits is quite limited because of its high fat and sugar contents. Therefore, the disruption of gluten network is not as critical in biscuits as it is in bread, and the effect of chia seeds on the volume is rather restricted in the samples presented in this study (9).

Spread ratio (diameter/thickness ratio) is among the most important characteristics of biscuits in determining their quality. Spread ratio values were higher (p>0.05) in the biscuits with buckwheat flour than in its chestnut flour-containing counterparts (Table 3). Moreover, chia seeds seemed to cause a slightly increasing effect on spread ratio. In the previous studies, the spread ratio was strongly linked to the chemical composition of the raw materials in addition to the interference of sugar and fat and relatively low dough viscosity (14).

The results of the sensory evaluation are shown in Fig. 2. According to the sensory evaluation results, only texture was indicated as significantly different among the biscuit samples, as the addition of chia seeds significantly increased the scores for the texture (p<0.05). The overall evaluation of the panellists illustrated a moderately higher acceptance (p>0.05) of biscuits supplemented with chia seeds (both buckwheat and chestnut flour-based biscuits) and the biscuits formulated with buckwheat flour and chia received the highest scores for appearance, flavour and texture.

Biscuits with buckwheat generally had the lowest results (except for the colour and appearance) in terms of the evaluated sensory parameters. Studies indicating that the level of buckwheat flour substitution in biscuits affected their sensorial acceptance could be found (52), where the biscuits with 40% of flour substituted with buckwheat flour had the highest score among three levels (30, 40 and 50%) of substitution, while the overall acceptability was measured in the samples with added xanthan gum. The presence of some phenolic compounds such as rutin, quercetin and protocatechuic acid in buckwheat flour was represented as the main factor for lower scores of buckwheat-incorporated products (52).

In the literature, lower scores in sensory evaluation of buckwheat-incorporated gluten-free biscuits were improved by adding gum acacia and guar gum into the formulations (15). Increase in sensory scores was explained by different interactions occurring between the hydrocolloids and other food ingredients that resulted in improved textural properties. Comparable increasing scores for flavour and overall acceptance after the addition of chia seeds into chestnut biscuits were also evident. This finding is similar to previous findings as combinations of chestnut flour and rice flour with different gums and emulsifiers provided increased acceptance scores in sensory evaluation of the bread samples (15).

Visual properties of appearance and colour of biscuits with chestnut flour were ranked higher than the biscuits with buckwheat flour (Fig. 2). Positive contribution of chestnut flour to the properties such as appearance, shape, crumb structure, etc. was in agreement with the previous studies (37). More specifically, consumer preference for darker colour in gluten-free bakery products was highlighted in previous studies (19).

Although the flavour scores were not significantly different among different formulations (p>0.05), chia seeds seemed to positively affect the flavour of buckwheat biscuits (Fig. 2). Due to the perceived bitterness, the buckwheat flour was reported to affect the product flavour adversely, particularly at high (40%) substitution levels (10).