TGP was dried at 50°C in a conventional oven until constant weight (24 h). Dried TGP was subjected to size reduction using a domestic coffee grinder to achieve a particle size like that of commercial flour.

Five formulations were prepared following a factorial design with central points to optimize % TGP and % sucralose according to the response variables: total phenol content, ABTS, and ORAC-FL. Two levels of substitution of wheat flour by TGP powder (10 and 20% w/w of total wet mass) and sucralose sweetener (2 and 4% w/w of total wet mass) were selected, as well as a central point with 15% of substitution by TGP powder and 3% of sucralose. In addition, control biscuits with different percentages of sucralose (2, 3, and 4% w/w in the total wet mass) and without substitution of TGP powder were formulated for comparison purposes. The formulations with the amounts of the ingredients are detailed in Table 1. The dough was prepared by mixing all the ingredients and then cut into equal shapes and sizes. They were baked in a conventional oven at 180°C for 12 min. The biscuits were subjected to size reduction obtaining a homogenous powder, packaged, and stored at freezer temperature (−20°C) for further bioactivity analysis.

Multiple regression analysis was performed fitting to the linear model with the following equation:

where β₀ indicates the intersection point, β₁ and β₂ the linear coefficients corresponding to % TGP and % sucralose, and β₁₂ for the interaction between the independent variables (factors %TGP and % sucralose).

The five formulated biscuits were evaluated by 15 semi-trained judges (aged 30–60 years). The biscuits were presented to each judge at the same time and could be selected in the order the judge wished. They had to select the most appropriate attributes to describe them applying a rapid descriptive methodology (Flash Profile). Once the descriptive terms had been established, the five biscuits were evaluated using 10 cm long unstructured scales. The different attributes were anchored with “nil/not at all” and “high.” The evaluations were carried out in a sensory evaluation laboratory designed according to ISO 8589 (25).

Consumer sensory analysis was carried out with the formulations (three biscuits) that showed significant differences in the sensory profile analysis by studying acceptability and perception. The three biscuits were presented to 101 consumers (aged 18–65 years) following a balanced complete block design. Consumers were recruited from the city of Montevideo (Uruguay) based on their interest in participating. For each biscuit, consumers indicate their overall liking using a 9-point hedonic scale. In addition, they had to answer check-all-that-apply (CATA) + just about right (JAR) questions (26) composed of 27 terms. Then, consumers responded to a brief survey of sociodemographic data.

Total phenol content (TPC) was determined by the Folin–Ciocalteu method as described by Fernández-Fernández et al. (11). Measurement was performed after 30 min of incubation in the dark at 750 nm in a microplate reader (Thermo Scientific Multiskan FC model, Massachusetts, USA). The results were expressed as mg of gallic acid equivalents (GAE)/g of sample.

Antioxidant capacity was determined by ABTS and ORAC-FL methods as described by Fernández-Fernández et al. (11). In both cases, the calibration curve was performed with Trolox (0.25–1.5 mM for ABTS and 0.01–0.08 mM for ORAC-FL), and the results were expressed as μmol of Trolox Equivalents (TE)/g of sample.

Inhibition of α-glucosidase and pancreatic lipase was assessed by measuring fluorescence (λexcitation = 360 nm, λemission = 460 nm) of the substrate (4-MUF-α-D-glucopyranoside) hydrolyzed by α-glucosidase and of the substrate (4-methylumbelliferyl-oleate) hydrolyzed by pancreatic lipase, respectively, for 30 min in a Varioskan Lux fluorimeter microplate reader (Thermo Scientific, Massachusetts, USA) (11). The results were expressed as the concentration of biscuit causing 50% inhibition (IC₅₀, mg/mL) of α-glucosidase/pancreatic lipase.

The sensory profiles of the five biscuit formulations (Figure 1) are shown in Table 3. The biscuits with TGP powder increased significantly (p < 0.005) the intensity of all the sensory attributes evaluated, which could have an impact on the consumers' acceptability. The biscuits with 20% of TGP powder were described as harder, with a more intense color, with a greater grape flavor, more acid, with greater flavor intensity and persistence, and with a greater aftertaste than the biscuits with 10% of TGP powder. Some of these attributes have been used to describe the sensory profile of other grape pomace baked goods, showing increased acidity and hardness with the by-product addition (32). The biscuits with 4% of sweetener were perceived as significantly sweeter (p = 0.0066) than the samples with the same amount of TGP powder but with 2% sweetener. The biscuits with 15% of TGP powder showed an intermediate sensory profile.

PCA was performed on the sensory data of the five evaluated biscuit formulations. The first two components (PC1 and PC2) accounted for 96.3% of the total variance (88.1 and 8.2% contributed by PC1 and PC2, respectively; data not shown). The color intensity, hardness, crispness, dryness, sandiness, sourness, grape flavor, flavor intensity, aftertaste, and flavor persistence contributed positively to PC1. These sensory descriptors were strongly correlated with each other. For PC2, sweetness contributed positively.

Three formulations of biscuits (20/4, 15/3, and 10/2%) showed significant differences in the sensory profile analysis by semi-trained judges, so they were selected for consumer analysis.

The overall liking was 5.0, 5.3, and 5.2 for 20/4, 15/3, and 10/2% biscuits. These results were in agreement with those reported by Fernández-Fernández et al. (13) for Tannat grape skin biscuits and with data reported by Kuchtová et al. (33) attributing the decrease in overall liking to the astringent and bitter taste due to the increase in the amount of by-product. The identification of consumer groups was achieved by the hierarchical cluster analysis according to the overall linking toward the products. They were identified as clusters 1, 2, and 3, composed of 37, 27, and 37 consumers, respectively (Table 4), considering a minimum score of 6.0 on a 9-point hedonic scale as the lowest acceptable score for a product to be considered commercially viable (34). It is noteworthy that due to the low number of consumers in each cluster, the results may have limitations.

Consumers of Cluster 1 showed significant differences (p = 0.0091) between the samples for the biscuit with the highest amount of TGP powder, the preferred one over the other two samples. The 20/4% biscuit, preferred by this group, was described as having adequate hardness, smooth texture, adequate sweetness, intense flavor, acid flavor, grape flavor, and being delicious. On the other hand, consumers of Cluster 2 preferred the biscuit with the least amount of TGP powder over the other two samples (p < 0.0001). The general liking decreased significantly with the addition of higher amounts of TGP powder showing quite low values of acceptability (below the commercial limit). The perception of these samples regarding negative attributes, such as not very sweet, dry, with aftertaste and strange taste, explain the low acceptability assigned by this group of consumers to all biscuits, indicating that Cluster 2 cannot be considered as potential consumers of TGP biscuits. Among the consumers of Cluster 3, significant differences (p < 0.0001) were also found between the evaluated samples, with the preference of the two biscuits with the highest amount of TGP powder over the sample 10/2%. This group assigned low acceptability scores to the biscuits evaluated, mainly to the ones that contained the least amount of TGP powder; therefore, they could be considered potential consumers of the biscuits with the highest % of TGP powder. The terms most used by consumers in Cluster 3 were adequate color, adequate hardness, slightly sweet, mild flavor, acid flavor, grape flavor, and strange flavor.

The attributes that were mentioned by the consumers and their frequency of mention for TGP biscuits were in agreement with Tannat grape skin biscuits (13), which were described as follows: from adequate color to too dark, of adequate crunch and being dry, adequate sweetness, with intense and persistent flavor, acid flavor, with adequate grape flavor.

TPC and antioxidant capacity increased in the biscuits with the substitution of wheat flour for TGP powder compared with their control biscuits (Table 5). The biscuit with a higher percentage of TGP and sucralose (20/4%) showed the highest values of TPC and antioxidant capacity (p < 0.05). These results indicated that the phenolic compounds present in the TGP powder remained with antioxidant capacity in the biscuits after baking. These results are in agreement with those reported by Theagarajan et al. (35) where biscuits added with 6% grape pomace powder showed significantly higher total polyphenols and antioxidant capacity compared with biscuits with 4% of grape pomace powder and a control biscuit without the incorporation of the by-product. Similarly, the fortification of cakes with 4, 6, 8, and 10% of Grape Pomace Powder (GPP) resulted in a significant increase in the antioxidant capacity measured by DPPH and FRAP assays, compared to the control cakes with no addition of GPP. The same trend was observed by the authors, where the cake with 10% of GPP showed higher antioxidant capacity (36). Accordingly, the Tannat grape skin biscuit (20% w/w Tannat grape skin of total dough wet mass) showed increased antioxidant capacity compared to the control biscuit (p < 0.05) (13).

Figure 2 surface plots show the relationship between the TGP percentage and that of sucralose incorporated into the biscuits, presenting a linear association with the response variables. Multiple regression analysis fitted to Equation 1 resulted in the following three equations (Table 6): TPC = 1.38+0.24^*TGP; ABTS = 35.02+9.43^*TGP; ORAC-FL = 34.76+8.19^*TGP. The non-significant coefficients (p-value > 0.05) were taken out from the analysis, showing that the response variables were only dependent on factor %TGP and positively influenced by this factor. The factor %sucralose was non-significant showing no significant influence on the response variables. R² values indicated that the variability in TPC, ABTS, and ORAC-FL is poorly explained by the fitting of the model because of the low values (< 80%). The highest TPC and antioxidant capacity corresponded to the biscuit with the highest substitution of wheat flour by TGP powder (20%).

The use of enzyme inhibitors in the prevention and treatment of diseases is applied in diabetes to reduce postprandial glucose levels by the inhibition of carbohydrate-hydrolyzing enzymes such as α-glucosidase. In the same way, inhibition of lipases related to triacylglycerol absorption is used to reduce fat levels and thus treat obesity (37). Our results suggested greater inhibition of α-glucosidase and pancreatic lipase with the highest substitution of wheat flour by TGP powder (20%) in biscuits (Table 7), showing the potential of TGP powder as a functional ingredient for regulating hyperlipidemia and hyperglycemia. These inhibitory capacities may be attributed to grape phenolic compounds (37).

When analyzing the TGP phenolic composition in flavan-3-ols, flavonols, and anthocyanins, as well as taking into account similar food matrices to TGP, an association of α-glucosidase and pancreatic lipase inhibition capacities can be made. Among anthocyanins, cyanidin and its glycosides have demonstrated α-glucosidase inhibition capacity (38). Proanthocyanidins have been reported for non-competitively inhibiting pancreatic lipase inducing its aggregation and subsequent stabilization of the aggregates (39). Regarding similar food matrices' phenolic composition to TGP, extracts of anthocyanin sources, which present flavonol glycosides (berry extracts, from black currant and rowanberry) such as TGP, have shown α-glucosidase inhibition capacity (40, 41). In addition, red fruit extracts from aronia (Aronia melanocarpa, A.), pomegranate (Punica granatum L., P.), and red grape (Vitis vinifera, RG) have shown high α-glucosidase inhibition capacity, being the anthocyanins, flavonols, and ellagitannins responsible for the inhibition (42). Particularly, the fractionation of the red grape extract showed higher inhibition capacity for the polymeric fraction (tannins such as condensed flavan-3-ols) when compared to anthocyanin (malvidin-3-glucoside, petunidin-3-glucoside, peonidin-3-glucoside, malvidin-3-acetylglucoside, coumaroyl derivatives of anthocyanins) and copigment fractions (hexoside of coumaric acid and isorhamnetin) (42). Other anthocyanin sources, blackthorn, açaí, and maqui blends, have been reported for inhibiting pancreatic lipase activity being negatively correlated with their anthocyanin content, as well as showing α-glucosidase inhibition capacity (43). Tannin-rich extracts from persimmon (Diospyros kaki Thunb.) have shown α-glucosidase inhibition capacity as well as glucose uptake and transport reduction in the Caco-2 cells model (44). A phenolic-concentrated extract of native Chilean red strawberry (Fragaria chiloensis ssp. chiloensis f. patagonica) has shown α-glucosidase (IC₅₀ = 0.80 μg/mL) and pancreatic lipase (41.36% inhibition at 50 μg/mL) inhibition capacities (45). Phenolic extracts from Canadian lentil cultivars (Lens culinaris), mainly composed of flavonoids (kaempeferol glycosides, catechin/epicatechin glucosides and procyanidins) such as TGP, presented good α-glucosidase and pancreatic lipase inhibition capacities (46). The main responsible for these exerted capacities may be the flavonols (46).

On the other hand, berry seeds such as maqui seeds, which are composed of polyunsaturated fatty acids, tocols, tocotrienols, tocopherols, and sterols, have shown α-glucosidase (IC₅₀ > 265.9 μg/mL) and pancreatic lipase (IC₅₀ > 71.33 μg/mL) inhibition capacities (47). Fruit fractions of saskatoon berry showed similar IC₅₀ values to the current work for α-glucosidase (IC₅₀ > 23.60 mg/mL) and pancreatic lipase (IC₅₀ > 81.90 mg/mL) inhibition capacities (48). In contrast, Merlot grape pomace extract presented no inhibition of α-glucosidase probably because of the low concentrations that were tested (< 250 μg/mL) (49). Particularly, Tannat grapes have shown α-glucosidase inhibition capacity (50), as well as Tannat grape skin (IC₅₀ = 11.67 mg/mL) (13) and their extracts (lowest IC₅₀ = 0.8886 mg/mL) (11). Moreover, a Tannat grape skin extract has shown pancreatic lipase inhibition capacity (IC₅₀ = 2.431 mg/mL) (11). Extracts from red grape pomace, skin, and seeds presented α-glucosidase inhibition capacity (IC₅₀ of 0.55, 0.30, and 0.36 mg/mL, respectively), as well as yogurt fortified with those extracts (51). In sum, TGP inhibition capacities can be attributed to grape phenolic compounds.

Regarding the inhibition capacities of foods with the addition of these phenolic compounds, the results of α-glucosidase inhibition are in agreement with the ones of a flatbread enriched with a source of oligomeric proanthocyanidins (white grape skin), showing increased activity compared to non-enriched food (52). The authors found that the inhibition increase was lower than expected (based on the grape skin added amount), probably because of proanthocyanin/protein complexes' formation with food matrix components which lowered solubility, hence reducing their accessibility for bioactivity determinations. Moreover, the baking temperature might have caused changes in those complexes as well as partial proanthocyanidins degradation (52). These statements are in agreement with the current results. Phenolic compounds can interact with food matrix macro-components such as lipids, proteins, and carbohydrates, changing their structure with subsequent changes in their bioactivity (53) by impairing their interaction with these enzymes.

In another work, blackcurrant pomace (by-product from fruit juice production) showed inhibition of glucose release during in vitro digestion when added to muffin formulations, by inhibiting starch digestive enzymes, counteracting pre-gelatinized starch (extruded wheat flour) hyperglycemic effect (54). Moreover, rye bread fortification with powders from Saskatoon berry fruits led to an increased inhibitory activity against α-glucosidase and pancreatic lipase (55). Kefir fortified with UAE-derived Sangiovese skin extract (from red grape pomace) presented α-glucosidase and pancreatic lipase inhibition (56). Grape seed powder inhibited cream (containing milk fat) lipolysis, whereas no inhibition was observed for beef fat, indicating that lipolysis may be determined by the dietary fat source of a meal (57). As to biscuits added with other food matrixes, a coffee fiber-containing biscuit showed a reduction of α-glucosidase activity (58). Biscuits with orange pomace have shown lower IC₅₀ values (< 25 mg/mL, higher α-glucosidase inhibition capacity) (59) than TGP biscuits. In contrast, particularly biscuits with 20% of Tannat grape skin have shown higher IC₅₀ values (>55 mg/mL, lower α-glucosidase inhibition capacity) (13) than 20% TGP biscuits.

Although red grape by-products have shown these capacities (11, 13, 60), to our knowledge, it is the first time that they have been determined in biscuits with grape pomace.

The biscuit 20/4% was the one that presented the highest antioxidant, antidiabetic, and antiobesity properties; therefore, LC-MS analysis was performed on this formulation.