Production of Vegan Ice Cream: Enrichment with Fermented Hazelnut Cake

Abstract

The growing demand for sustainable plant-based dairy alternatives has spurred interest in valorizing agro-industrial byproducts like hazelnut cake, a protein-rich byproduct of oil extraction. This study developed formulations for vegan ice cream using unfermented (HIC) and Aspergillus oryzae-fermented hazelnut cake (FHIC), comparing their physicochemical, functional, and sensory properties to conventional dairy ice cream (DIC). Solid-state fermentation (72 h, 30 °C) enhanced the cake’s bioactive properties, and ice creams were characterized for composition, texture, rheology, melting behavior, antioxidant activity, and enzyme inhibition pre- and post-in vitro digestion. The results indicate that FHIC had higher protein content (64.64% vs. 58.02% in HIC) and unique volatiles (e.g., benzaldehyde and 3-methyl-1-butanol). While DIC exhibited superior overrun (15.39% vs. 4.01–7.00% in vegan samples) and slower melting, FHIC demonstrated significantly higher post-digestion antioxidant activity (4.73 μmol TE/g DPPH vs. 1.44 in DIC) and angiotensin-converting enzyme (ACE) inhibition (4.85–7.42%). Sensory evaluation ranked DIC highest for overall acceptability, with FHIC perceived as polarizing due to pronounced flavors. Despite textural challenges, HIC and FHIC offered nutritional advantages, including 18–30% lower calories and enhanced bioactive compounds. This study highlights fermentation as a viable strategy to upcycle hazelnut byproducts into functional vegan ice creams, although the optimization of texture and flavor is needed for broader consumer acceptance.

1. Introduction

Ice cream is a frozen dairy dessert traditionally made from milk, cream, sugar, and stabilizers, with possible inclusions of eggs, fruits, nuts, or flavorings [1]. Its origins trace back to ancient China, where snow and ice were mixed with fruit and honey, while Persian empires developed sharbat, a chilled dessert that later influenced Italian sorbetto [2]. The modern iteration emerged in 17th-century Europe, with industrial production beginning in the 19th century following advances in refrigeration and homogenization. Global consumption exceeds 18 billion liters annually, led by the U.S., Australia, and New Zealand, while artisanal varieties—such as Italian gelato (denser and lower in fat), Turkish dondurma—Maraş (chewy, with salep— a plant-based emulsifier), and Japanese mochi ice cream (rice-wrapped)—highlight cultural adaptations. Recent trends focus on functional ice creams fortified with probiotics, plant proteins, or reduced sugar to align with a health-conscious demand [3].

However, ice cream is a globally consumed dairy-based dessert, and growing consumer demand for plant-based alternatives has driven innovation in vegan ice cream formulations [4]. Vegan ice cream is a plant-based frozen dessert that replicates the sensory properties of conventional dairy ice cream by utilizing non-animal-derived ingredients such as nut, legume, grain, or seed milks, combined with plant-based fats, sweeteners, and stabilizers [4]. While early iterations date back to the 19th century with sorbets and coconut-based formulations, the modern vegan ice cream industry emerged in the late 20th century alongside rising lactose intolerance awareness and ethical consumerism [5,6]. Nutritionally, vegan ice cream typically contains comparable energy density (180–350 kcal/100 g) to dairy versions but often provides higher fiber, unsaturated fats, and bioactive compounds (e.g., polyphenols and tocopherols) from plant sources, although the protein content varies significantly based on the base ingredients [7]. Common plant-based milks used in production include almond, soy, coconut, oat, and cashew, with newer innovations leveraging pea, hemp, and upcycled byproducts (e.g., fermented hazelnut cake) to enhance functionality [8,9]. Global consumption reached $1.2 billion in 2023, with North America and Europe dominating the market, while Asia-Pacific shows the fastest growth due to rising flexitarian diets [10]. Future projections estimate a 12.8% CAGR (2024–2032), driven by advances in fermentation, protein texturization, and sustainable ingredient sourcing (e.g., microalgae fats and upcycled food waste) to improve sensory and nutritional profiles [11]. Regulatory frameworks and clean-label demands are expected to further shape product development, positioning vegan ice cream as a mainstream choice in the frozen dessert sector. Traditional dairy substitutes, such as almond, soy, and coconut milk, are widely used in vegan ice cream production, but there is increasing interest in upcycling agro-industrial byproducts to enhance nutritional and functional properties [9,12]. Hazelnut cake, a byproduct of hazelnut oil extraction, is rich in proteins, fibers, and bioactive compounds but remains underutilized despite its potential [13].

Solid-state fermentation (SSF) of oilseed or nut cakes—a byproduct of oil extraction—utilizes microbial cultures (e.g., Aspergillus oryzae and Rhizopus oligosporus) to enhance their nutritional and functional properties. SSF occurs under low-moisture conditions, enabling fungi or bacteria to hydrolyze anti-nutritional factors (e.g., phytates and tannins) while increasing protein digestibility, bioactive peptides, and phenolic compounds [14]. Fermented cakes exhibit elevated antioxidant capacity (via free radical scavenging assays), angiotensin-converting enzyme (ACE)-inhibitory activity (antihypertensive potential), and prebiotic oligosaccharides due to microbial enzymatic action [15]. SSF offers several advantages, including simplified product recovery, lower overall production costs, reduced fermenter volume, minimal downstream processing requirements, and decreased energy demands for both agitation and sterilization. Challenges also include process standardization and scalability, but SSF aligns with circular economy goals by valorizing agro-industrial waste [16].

Despite these benefits, limited research has explored the application of fermented hazelnut cake in food products, particularly in vegan ice cream. Previous studies have investigated the use of fermented plant proteins in dairy alternatives, but none have specifically optimized SSF-treated hazelnut cake for this purpose. Furthermore, the impact of fermentation on textural, sensory, and functional properties of vegan ice cream remains underexplored. Therefore, this study aimed to incorporate fermented hazelnut cake into vegan ice cream and evaluate its physicochemical, textural, and sensory properties compared to conventional dairy ice cream formulation. This study also assessed the potential health benefits, including antioxidant and antihypertensive effects, through in vitro digestion assays. By valorizing hazelnut cake through fermentation, this research contributes to sustainable food production while offering a functional plant-based ice cream alternative with enhanced nutritional benefits.

2. Materials and Methods

2.1. Materials

The hazelnut cake from the cold-pressed oil process was obtained from Cansızzade Doğal Bitkisel Yağlar brand, İstanbul, Türkiye. It was then ground (ZM200, Retsch GmbH, Haan, Germany) into a powder with an average particle size of 250 μm and fermented according to the method described in Ozdemir et al. (2024) [9]. For ice cream production, the full-fat cow milk (Carrefour brand, Adana, Türkiye), salep (Buldan, Türkiye), powdered sugar (Dr. Oetker, İzmir, Türkiye), dairy cream (35% fat content, İçim Şef, Sakarya, Türkiye), sweetener of sodium saccharin (Takita, İzmir, Türkiye), and coconut oil (WeFood, Doğavera Gıda A.Ş, İstanbul, Türkiye) were purchased from a local supermarket in Adana, Türkiye. Potassium-sodium tartrate, aluminum chloride, Folin–Ciocalteu phenol reagent, Ferrozine^®, copper (II) sulfate, iron (II) chloride tetrahydrate, angiotensin-converting enzyme, captopril, N-[3-(2-Furyl)acryloyl]-Phe-Gly-Gly (FAPGG), α-amylase from porcine pancreas, α-glucosidase from Saccharomyces cerevisiae, 3,5-Dinitrosalicylic acid (DNS), starch, 4-nitrophenyl α-D-glucopyranoside, 2,2-diphenyl-1-picrylhydrazyl (DPPH), 2,2′-azinobis- (3-ethylbenzothiazoline-6-sulfonate) (ABTS), 2,4,6-tris(2-pyridyl)-Striazine (TPTZ), gallic acid, trolox, 4-(2-aminoethyl) benzenesulfonyl fuoride hydrochloride (AEBSF), pepsin, pancreatin, amylase o-phthaldialdehyde, di-Na-tetraborate decahydrate, Na-dodecyl-sulfate, dithiothreitol, and serin were purchased from Sigma-Aldrich Ltd. (St. Louis, Missouri, the USA). The bile salt, sodium acetate, sodium nitrite, potassium chloride, potassium dihydrogen phosphate, calcium chloride, and hydrogen peroxide were bought from Fisher Scientific UK Ltd. (Leicestershire, UK). All solvents used were of analytical or HPLC grade.

2.2. Production of Fermented Hazelnut Oil Cake

Fermented hazelnut oil cake (FHOC) was produced under the optimized process conditions described by Ozdemir et al. (2024) [9]. Briefly, 15 g of hazelnut oil cake (500 μm average particle size) with 69.8% initial moisture content was inoculated using 10⁷ spore/g DM substrate of Aspergillus oryzae 200828 and incubated at 24.6 °C for 4.6 days in a static incubator (Memmert IN 110, Schwabach, Germany). The FHOC was kept at −18 °C for further use.

2.3. Ice Cream Production

The method given by Atalar et al. (2021) was modified and tested to produce ice cream from fermented hazelnut cake [17]. First, the powdered fermented hazelnut cake was homogenized in distilled water (10 g/100 mL) at 10,000 rpm for 10 min (IKA, T18 Ultra-Turrax). The components of the produced ice creams are listed in

Table 1. Formulations of ice cream products.

Dairy Ice Cream (DIC) (100 g)	Hazelnut Cake Ice Cream (HIC) (100 g)	Fermented Hazelnut Cake Ice Cream (FHIC) (100 g)
18 g powdered sugar	13.45 g sweetener	13.45 g sweetener
0.75 g salep	0.75 g salep	0.75 g salep
16.4 g dairy cream (35% oil content)	8 g coconut oil	8 g coconut oil
73.25 g milk	77.8 g hazelnut cake beverage (10% w/v)	77.8 g fermented hazelnut cake beverage (5% w/v)

. Each prepared ice cream mixture was homogenized again (at 16,000 rpm for 10 min at 70 °C), and when the mixture temperature reached 80 °C, it was subjected to pasteurization in a water bath for 10 min. After pasteurization, the mixture was cooled to +4 °C and kept at the same temperature for 24 h. The mixture was then transferred to an ice cream machine for ice cream production (Cuisinart ICE30BCE, Stamford, CT, USA). The ice cream produced was stored at −18 °C for 24 h and kept at the same temperature until analysis. Prior to storage, the ice cream was portioned and stored in appropriate containers with sealed lids. The production was repeated twice.

2.4. Characterization of Ice Cream Products

Component analyses: Total moisture content was determined gravimetrically according to the method reported in Lamsal et al. (2007) [18], total protein amount was determined using the Kjeldahl method (Tamayo Tenorio et al., 2016) [19] (Vodest, 45 G, C. Gerhardt GmbH & Co. KG, Königswinter, Germany) with a protein conversion factor of 6.25, soluble protein content was determined spectrophotometrically (Lowry et al., 1951) [20] (Cary 60 UV-Vis, Agilent Technologies, Inc., Santa Clara, CA, USA), total fat content was determined according to the method number AACC 30-25 [21] (Soxtherm SE414, C. Gerhardt GmbH & Co. KG, Königswinter, Germany), and total ash amount was determined according to the method number AACC 08-01 [22] (AACC, 2009a, b). The total carbohydrate content was obtained by calculating the total protein, fat, moisture, and ash content in 100 g of sample and subtracting it from 100. Color analyses of the resulting products were performed using a color measurement device (CM-5, Konica Minolta, Tokyo, Japan), and the results were reported as L*, a*, and b*. Identification and quantification of volatile compounds were carried out using a Gas Chromatography-Flame Ionization Detector (GC-FID) (6890N, Agilent Technologies, Inc., Santa Clara, CA, USA) and associated Mass Spectrometry (MS) (5975B VL MSD, Agilent Technologies, Inc., Santa Clara, CA, USA). The Solid-Phase Micro-Extraction (SPME) technique was used in the isolation of volatile compounds. Equilibration, extraction, and injection processes were performed using an automatic injection module (GC Injector 80,Agilent Technologies, Inc., Santa Clara, CA, USA). Mixing on/off time was 5/2 s, vial needle penetration was 11 mm, and vial fiber open area was 22 mm. Divinylbenzene/Carboxen/Polydimethylsiloxane (DVB/CAR/PDMS, 50/30 μm, SF 23GA Auto, 57299-U, Sigma, St. Louis, MO, USA) coated fiber was used in the isolation of volatile compounds. The extraction conditions considered to be applied for the samples to be analyzed were those determined by Salum et al. (2017) [23] (54.8 °C extraction temperature, 86 min incubation time, and 250 rpm stirring speed). Desorption of the fiber took place at 250 °C for 180 s in the GC injection port (7890B, GC System, Agilent Technologies, Inc., Santa Clara, CA, USA). Separation was carried out on a DB-Wax (30 m × 250 μm × 0.25 μm; 122–7032, Agilent Technologies, Inc., Santa Clara, CA, USA) polar column, and helium gas was used as the carrier gas at a flow rate of 2 mL/min. The oven temperature was programmed as follows: increased to 70 °C by 5 °C per minute, held there for 1 min, then increased to 240 °C by 10 °C per minute, and maintained at this temperature for 4 min. Following this, the column flow was transferred to the FID and MS detectors. The detector temperature was set to 260 °C. MS ionization energy was 70 eV, and scanning was performed between 30 and 400 mass/charge. The internal standard method was used to determine the number of compounds. Deconvolution of the chromatograms was performed with the AMDIS (Automated Mass Spectral Deconvolution and Identification AMDIS32 V2.1) program. MS libraries (NIST 11, Wiley 7.0) and standard compounds were used to identify the peaks. In addition, alkane series (C8–C20) were injected, and the linear retention indices (LRIs) of the peaks were calculated (Van Den Dool and Kratz, 1963) [24].

Flow behaviors: A viscometer (DV-II+ Pro, Brookfield Engineering, USA) was used to determine the viscosity of each ice cream sample at different temperatures, and the viscosity values of the samples were measured at temperatures ranging from 4 °C to 35 °C. However, at temperatures below 20 °C, the fat (milk fat or coconut oil) in the produced ice cream remained in its solid form, and since the produced ice creams were not fluid enough to be measured with the device used, measurements were performed at 25 and 35 °C in this study. A circulating water bath (ICC Basic Eco 8, IKA^®-Werke GmbH & Co., KG, Staufen, Germany) and a Small Sample Adaptor (Brookfield Engineering, USA) accessory were used to adjust the temperatures of the samples. In the measurements conducted in this section of this study, SC4-18 or SC4-25 spindles were used depending on the viscosity of the samples (Brookfield Engineering, Middleboro, MA, USA). To determine the flow properties of the measured ice creams, the shear stress (τ, Pa) values obtained from measurements at different shear rates (γ, s⁻¹) were modeled using a “Power Law” type equation, and the flow behavior index (n) and viscosity coefficient (K, Pa.s) values were determined [1]:

τ = K × γⁿ

(1)

Overrun and melting properties: The weights of the ice cream mixture and the ice cream produced were determined for the same volume. The volume increase (overrun) was calculated using the following formula:

Volume increase = [(weight of mixture − weight of ice cream)/weight of ice cream] × 100

(2)

A sample of ice cream was placed on a wire mesh on a beaker. The amount of ice cream melted at 25 °C for 120 min was weighed at different times (30, 60, 90, and 120 min) to determine the melting rate. The melting rate was calculated using the following formula:

Melting rate (%) = (Melted ice cream weight/ice cream weight) × 100

(3)

The time at which the ice cream samples first dripped was taken as the first drop time, and the results were expressed in minutes [25].

Textural properties: The textural properties of the ice cream samples were determined by applying a compression test using a Brookfield CT3 Texture Analyzer (Brookfield Engineering Laboratories, INC. Middleboro, MA, USA) device with a 5 kg load cell, modified from Acu et al. (2021) [26]. For this purpose, measurements were taken at temperatures between −6 °C and −9 °C for each 50 mL sample. Temperature monitoring was performed using a K-type thermocouple connected to a digital thermometer (Fluke 568). The ice cream samples were kept at the desired temperature range by placing them in an ice mold. During measurements, the room temperature was recorded at a maximum of 25 °C. Necessary precautions were taken to prevent the container filled with samples from moving during pressing. Care was taken to ensure that the samples were at least 2 cm high inside the container. A conical probe (TA15/1000, 30 mm diameter, 45°) was used, with probe speeds of 2.0 mm/s (pre-test), 0.5 mm/s (test), and 0.5 mm/s (post-test). The target distance and trigger load were 10 mm and 4.5 g, respectively. Measurements were taken from five points spaced apart in each sample container. The samples were tested in two parallel runs. The test results recorded the values for hardness (g), adhesiveness (mJ), cohesiveness, and gumminess (g).

Bioactive properties:

Determination of ABTS⁺ and hydroxyl free radical binding capacities: The free radical binding capacities of the samples were determined spectrophotometrically (Re et al., 1999; Smirnoff and Cumbes, 1989) [27,28]. The reaction was determined by measuring the change in absorbance value at 734 nm over 6 min between 0.1 mL of sample solution and 1.9 mL of 7 mmol/L ABTS radical solution (prepared in 2.5 mmol/L potassium persulfate). For the hydroxyl radical binding capacity, 100 μL of sample and 3 mL of Smirnoff reagent were incubated at 37 °C for 30 min. Absorbance measurements were then performed at 510 nm. The results were expressed as trolox equivalents.

Determination of iron chelating activity: The iron ion chelating capacities of the samples were determined spectrophotometrically (Arcan and Yemenicioglu, 2007) [29]. Briefly, 2 mL of appropriately diluted sample solution was mixed with 0.1 mL of 1 mmol/L FeCl2.4H₂O solution and incubated for 30 min. Then, 0.1 mL of 5 mmol/L ferrozine was added, mixed, and incubated for another 10 min. The resulting color was measured spectrophotometrically at 562 nm, and the results were expressed as EDTA equivalent.

ACE inhibition: Briefly, 10 μL of 0.25 units/mL ACE prepared in 0.01 mol/L saline phosphate buffer (pH: 7.0, NaCl concentration: 0.5 mol/L) was mixed with 10 μL of sample solution. The mixture was incubated at 37 °C for 15 min, and then, the reaction was started by adding 150 μL of 1.75 mmol/L FAPGG substrate solution (37 °C) prepared in saline phosphate buffer. The absorbance of the reaction was measured kinetically at 340 nm for 30 min at 37 °C. ACE activity was determined using the slope of the initial linear region of the absorbance–time curve, and captopril was used as a control (Aydemir et al., 2014) [30].

Determination of acetylcholine esterase (AChE) inhibition activity: The inhibition activity of the samples was carried out as described in Vinutha et al. (2007) [31]. The principle of inhibition activity is the measurement of the yellow anion TNB (5-thio-2-nitrobenzoate), which was released as a result of the reaction between thiocholine and DTNB (5,5′-dithio-2-nitrobenzoic acid) hydrolyzed by acetylcholinesterase, and at 412 nm absorbance, acetylthiocholine iodide (10.85 mg/5 mL phosphate buffer) was used as a substrate for activity determination. The solution containing various concentrations of sample or standard to be prepared for analysis was prepared with 250 μL of 200 mM phosphate buffer at pH 7.7, 80 μL DTNB, and 10 μL enzyme (2 U/mL). The solution was incubated at 25 °C for 5 min for pre-incubation. Then, 15 μL of substrate was added to the solution and incubated again for 5 min, and the yellow-colored compound was formed.

Total phenolic content: Total phenolic content was determined using the Folin–Ciocalteu reagent (Singleton and Rossi, 1965), and the results were expressed as gallic acid equivalents [32].

α-amylase/α-glucosidase inhibition: α-Glucosidase and α-amylase inhibition experiments were carried out according to the methods specified in Aydemir et al. (2022) [33]. Briefly, 100 μL of α-glucosidase enzyme (1 unit/mL) to be prepared in 0.1 mol/L sodium phosphate buffer (pH 6.8) was pre-incubated with 50 μL of sample for 10 min at 37 °C. Then, 50 μL of 10 μM (4-Nitrophenyl β-D-Galactopyranoside) was added as a substrate to initiate the enzymatic reaction and the reaction, which continued at 37 °C for 30 min and was then terminated by adding 1 mL of 0.1 mol/L Na₂CO₃. Enzyme activity was determined by measuring the formed p-nitrophenyl at 400 nm. For the other analysis, 100 μL of α-amylase enzyme (1 unit/mL) prepared in 0.1 mol/L sodium phosphate buffer (pH 6.8) was incubated with 100 μL of sample for 5 min at 37 °C. Then, 250 μL of 1% (mg/mL) starch prepared in sodium phosphate buffer (pH 6.8) was added to start the reaction. The reaction was carried out at 37 °C for 5 min, and 200 μL of DNS reagent (1% 3,5-dinitrosalicylic acid and 12% sodium potassium tartrate in 0.4 mol/L NaOH) was added. The reaction was carried out by heating at 95 °C for 10 min. Then, it was diluted with 1 mL of distilled water in an ice bath. α-Amylase activity was determined by measuring absorbance at 540 nm. A solution without sample was used as the control, a solution without substrate was used as a blank (for each sample), and a solution without enzyme was used as an enzyme blank (for each sample).

Lipase-inhibitory activity: The enzymatic spectrophotometric method used to determine the lipase-inhibitory activity of pure eritadenine, extracts, dried samples, and samples taken at different digestion phases following in vitro digestion (McDougall et al., 2009) [34]. The interaction of the lipase enzyme with the p-nitrophenyl laurate reagent resulted in the formation of the p-nitrophenyl reagent. The degree of lipase inhibition by the samples added to the reaction medium was determined.

In vitro digestion: In vitro digestion of the samples was performed according to the method described by Brodkorb et al. (2019) [35]. Appropriate environmental conditions for in vitro oral, gastric, and intestinal digestion of the samples were prepared as described in the publication. The oral phase involved diluting the food 1:1 (by weight) with simulated salivary fluid (SSF) containing salivary amylase. The sample was incubated at pH 7.0 for 2 min. After completion of the oral phase, the resulting mixture was diluted 1:1 (volume/volume) with simulated gastric fluid (SGF) and gastric enzymes (pepsin and gastric lipase). The gastric phase was performed for 2 h at pH 3.0 in a shaking incubator. The resulting gastric phase mixture was then diluted 1:1 (v/v) with simulated intestinal fluid (SIF), bile salts, and pancreatic enzymes (pancreatin based on trypsin activity) and incubated at pH 7.0 for 2 h with shaking. The digestion process was stopped by using 5 mM Pefabloc SC (4-(2-aminoethyl) benzenesulfonyl fluoride hydrochloride) as a protease inhibitor and 5 mM 4-bromophenylboronic acid as a pancreatic lipase inhibitor. Before digestion, the enzyme activities were determined according to the methods in the referenced article; samples were prepared as described in the study, and the entire digestion process was carried out according to the INFOGEST protocol, which is described in detail in the relevant article.

Sensorial properties: The panelists (n = 13, university students and professors in the Department of Food Engineering, aged between 22 and 42) were selected and trained in accordance with ISO 11136:2014, using six basic tastes, taste thresholds, and odor tests [36]. The selected panelists were trained prior to the sensory analysis of the samples using ice cream and different plant-based milks. In the sensory analyses, descriptive sensory tests were used to examine various characteristics of the ice cream samples, including color, odor, chewy texture, icy feeling, smooth texture, melt-in-the-mouth sensation, presence of foreign compound/flavor, and overall acceptability. A unipolar graphic scale was used to evaluate these characteristics.

The experiments performed in this study were analyzed using variance analysis, which was statistically conducted using Design Expert (11.0.0 trial version, Stat-Ease, Minneapolis, MN, USA) software. The significance of the model was evaluated according to ANOVA, where the variables having a p-value lower than 0.05 were accepted to have a significant effect on the results.

3. Results and Discussions

3.1. General Characterization of Ice Cream Products

The general composition and color properties of hazelnut oil cake (HOC) and fermented hazelnut oil cake (FHOC) that were used in vegan ice cream production are given in

Table 2. General characteristics of HOC and FHOC [9].

	HOC	FHOC
Dry matter (%)	91.08 ± 0.02 a,†	85.43 ± 0.23 b
Water activity (25 °C)	0.733 ± 0.00 a	0.704± 0.00 b
Protein (%)	58.02 ± 0.77 b	64.64 ± 0.28 a
Lipid (%)	8.41 ± 0.12 b	9.19± 0.35 a
Ash (%)	6.05 ± 0.00 b	7.11 ± 0.0 a
Carbohydrate (%)	27.52	19.06
Color values
L*	77.34 ± 0.03 a	54.83 ± 0.38 b
a*	2.39 ± 0.010 b	6.39 ± 0.07 a
b*	13.18 ± 0.05 b	16.47 ± 0.08 a

^† Values are given as mean ± standard deviation. If the letters (a–b) used as superscripts are the same, there is no difference between samples produced with different formulations (p > 0.05).

.

In ice cream production, we aimed to achieve equal adjustment of the dry matter of each product studied in the product formulations (30%) (

Table 3. General characteristics of ice cream products.

	DIC	HIC	FHIC
Dry matter (%)	30.95 ± 0.21 c,†	29.10 ± 0.14 b	24.35 ± 0.35 a
Protein (%)	2.33 ± 0.03 c	3.67 ± 0.04 b	2.08 ± 0.07 a
Lipid (%)	7.88 ± 0.18 a	8.00 ± 0.35 a	8.25 ± 0.71 a
Ash (%)	0.51 ± 0.01 c	0.44 ± 0.00 b	0.24 ± 0.03 a
Carbohydrate (%)	20.07	16.99	14.02
Dietary fiber (%)	3.8	3.22	3.16
Calorie (kcal)/100 g	162	133	115
pH	6.80 ± 0.02 b	6.60 ± 0.02 c	8.17 ± 0.12 a
Color values
L*	86.52 ± 0.40 c	76.12 ± 0.06 b	63.76 ± 0.87 a
a*	0.21 ± 0.04 c	1.63 ± 0.01 b	3.33 ± 0.13 a
b*	9.64 ± 0.13 b	9.35 ± 0.08 b	13.49 ± 0.42 a

^† Values are given as mean ± standard deviation. If the letters (a–c) used as superscripts are the same, there is no difference between samples produced with different formulations (p > 0.05).

). Although the dry matter values were quite similar in DIC and HIC, FHIC had approximately 15% lower dry matter than others since FHIC had swollen and become a mud-like structure when the same amount of the liquid was used in ice cream formulation with DIC and HIC. The processing was very challenging; hence, a reduced quantity of FHOC was utilized in the ice cream manufacturing. Due to the reduced quantity of fermented HOC, the protein, ash, and carbohydrate contents of FHIC were lower than HIC. It should be noted that HIC had 60% higher protein content and 15% lower carbohydrate content compared to DIC. In addition, HIC and FHIC had 18% and 30% lower calorie values than DIC, respectively. This general characterization revealed that the vegan ice cream products had increased nutritional content and reduced energy intake than dairy ice cream. Vegan ice cream is darker than DIC due to phenolic compounds in cake components. The fermentation process also increased the free phenolics in the medium. The higher pH also triggered the oxidation or darkening of free phenolics in the fermented cake. Therefore, FHIC had the darkest color, followed by HIC and DIC. FHIC had slightly more redness and yellowness than other ice creams. These results are generally consistent with the results found in the study by Atalar et al. (2021) [17]. During a literature search (Web of Science), no study on ice cream made with hazelnut drink (milk) was found, except for Atalar et al. (2021) [17].

3.2. Volatile Compounds in Ice Cream Products

The volatile profiles of DIC, HIC, and FHIC were analyzed (

Table 4. Volatile compound composition of ice cream products (μg/kg sample).

LRI *	Compound	DIC	HIC	FHIC
1210	Limonene	-	46.0 ± 2.2	-
1215	3-Methyl-1-butanol	-	-	13.2 ± 0.2
1445	Acetic acid	15.3 ± 0.2	13.7 ± 0.6	14.5 ± 0.4
1506	Pyrol	-	-	39.2 ± 1.4
1525	Benzaldehyde	-	-	182.3 ± 2.7
1571	2-Methyl Propanoic acid		-	32.2 ± 0.1
1843	Hexanoic acid	15.5 ± 1.1	6.0 ± 0.1	-
1958	δ-Octalactone	-	108.3 ± 0.9	172.2 ± 2.6
2072	Octanoic acid	116.0 ± 4.9	43.0 ± 3.4	50.2 ± 3.0
2079	p-Cresol	-	-	9.1 ± 0.2
2180	δ-Decalactone	-	49.2 ± 1.6	94.8 ± 2.1
2413	Benzoic acid	-	-	7.5 ± 0.1

Values are given as mean ± standard deviation. * Linear retention index (LRI) values calculated for the DB-Wax column.

), with concentrations reported as μg/kg sample and linear retention indices (LRIs) provided. FHIC exhibited the highest diversity and concentration of volatiles (615.2 ± 5.3 μg/kg; 10 compounds), followed by HIC (266.3 ± 0.7 μg/kg; 5 compounds) and DIC (146.9 ± 6.2 μg/kg; 3 compounds). DIC volatiles were only acetic acid, octanoic acid, and hexanoic acid, consistent with prior reports of dominant milk/cream headspace compounds [37,38]. In HIC, limonene, acetic acid, hexanoic acid, and octanoic acid were identified, aligning with hazelnut cake profiles [9]. Notably, δ-octalactone (40.7% of headspace) and δ-decalactone (18.5%) were detected compounds absent in hazelnut cake but likely derived from coconut oil [39]. Among FHIC volatiles, benzaldehyde (29.6% of headspace) was the most abundant, alongside pyrrole, 2-methylpropanoic acid, p-cresol, and benzoic acid. δ-Octalactone (28.0%) and δ-decalactone (15.4%) were also present, reinforcing their potential coconut oil origin. Unique to FHIC was 3-methyl butanol, undetected in fermented hazelnut cake but plausibly formed via alcohol dehydrogenase-mediated reduction of its precursor, 3-methyl butanal [40], which was identified in the cake [9]. According to the study, fermentation-driven mechanisms (such as the conversion of aldehyde to alcohol) may differentiate the volatile profile of FHIC, while lactones in HIC/FHIC may indicate lipid oxidation or thermal degradation of coconut oil.

3.3. Flow Behavior of Ice Cream Products

The rheological properties of ice cream mixes were analyzed at 25 °C and 35 °C using shear rate versus shear stress measurements (

Table 5. Flow properties of DIC, HIC, and FHIC at different temperatures.

Ice Cream	25 °C		35 °C
	n (-)	k (Pa.sn)	n (-)	k (Pa.sn)
DIC	0.48 ± 0.01 a	40.57 ± 1.49 c	0.29 ± 0.04 a	80.87 ± 14.08 b
HIC	0.61 ± 0.02 c	26.94 ± 2.14 b	0.42 ± 0.11 a	55.94 ± 21.41 b
FHIC	0.51 ± 0.00 b	2.63 ± 0.08 a	0.63 ± 0.00 b	1.85 ± 0.07 a

Values are given as mean ± standard deviation. If the letters (a–c) used as superscripts are the same, there is no difference between samples produced with different formulations (p > 0.05). k: consistency constant; n: flow behavior index.

). The fluids exhibited non-Newtonian behavior, necessitating logarithmic transformation (ln) of shear rate and shear stress values for graphical representation. The flow behavior index (n) and consistency coefficient (k) were derived from these plots (Supplementary Table S1).

All ice cream formulations demonstrated shear-thinning (pseudoplastic) behavior, as evidenced by “n” values < 1.0, consistent with prior studies on both dairy and vegan ice creams. Shear-thinning is typical in ice cream due to the disruption of colloidal structures (e.g., fat globules and protein networks) under shear stress [41]. The lower “n” values (higher pseudoplasticity) in DIC align with Goff et al. (2025) [1], who attributed this behavior to casein micelles and fat globule interactions in dairy systems. Temperature dependence varied by formulation in DIC and HIC, and pseudoplasticity increased with temperature (reflected by a decrease in “n”), whereas FHIC exhibited the opposite trend. At 25 °C, significant differences (p < 0.05) in flow behavior were observed among formulations, with DIC showing the lowest n values and HIC the highest. At 35 °C, DIC retained the lowest “n” values, but the difference from HIC became statistically insignificant (p > 0.05). The temperature-dependent augmentation of pseudoplasticity (reduced “n” at 35 °C) aligns with the observations of Adapa et al. (2000), who reported improved fat coalescence at elevated temperatures [42]. The shear-thinning behavior of HIC parallels the report by Soukoulis et al. (2009) on fiber-enriched ice creams, where coconut oil and hazelnut cake proteins might form shear-sensitive networks [43]. The anomalous trend in FHIC (increase in “n” with temperature) may reflect fermentation-induced protein hydrolysis, reducing structural integrity, a phenomenon observed in fermented plant protein matrices by McClements et al. (2019) [4].

The analysis of “k” values revealed pronounced differences in consistency. Fermented hazelnut ice cream exhibited markedly lower k values (1.9–2.6) compared to plain (26.9–80.9) and hazelnut pulp ice creams. At 25 °C, the highest k values were observed in plain ice cream, followed by hazelnut pulp ice cream; both increased with temperature. In contrast, fermented hazelnut ice cream displayed negligible temperature-dependent changes in consistency. The significantly lower “k” values in FHIC (1.9–2.6 vs. 26.9–80.9 in DIC/HIC) suggest reduced viscosity, likely due to fermentation effects. The fermentation process may deteriorate polysaccharides (e.g., hemicellulose in hazelnut cake) and proteins, reducing their thickening ability. The higher “k” in DIC and HIC reflects the stabilizing role of dairy proteins (caseins), hazelnut proteins, and complex carbohydrates. Similar results were reported by Aboulfazli et al. (2015) [44].

The flow behaviors of DIC and HIC were similar, with both showing increased pseudoplasticity and consistency at higher temperatures, converging statistically at 35 °C. Fermented hazelnut ice cream diverged from this trend, likely due to enzymatic and structural modifications during fermentation. Specifically, fermentation may degrade hydrocolloid polysaccharides and/or proteins, reduce viscosity-enhancing interactions, and enhance solubility.

3.4. Overrun and Melting Properties of Ice Cream Products

The melting behavior, structural stability, and overrun (volume increase) of three ice cream types are given in

Table 6. The initial melting time, melting time, and overrun of DIC, HIC, and FHIC.

Ice Cream	Overrun (%)	First Melting Time (min)	Melting Time (min)	Melting Ratio (%)
				15th min	30th min	45th min	60th min	75th min	90th min
DIC	15.39 ± 0.41 c	17.5 ± 0.7 c	99.2 ± 0.0 b	-	5.8 ± 0.6	19.0 ± 0.6	36.0 ± 7.8	59.8 ± 2.0	86.5 ± 4.7
HIC	4.01 ± 0.82 b	6.0 ± 0.0 b	68.0 ± 0.7 a	8.40 ± 0.1	28.8 ± 0.9	63.0 ± 2.6	86.8 ± 2.1	-	-
FHIC	7.00 ± 0.50 a	3.0 ± 0.3 a	66.5 ± 1.1 a	19.7 ± 0.3	49.1 ± 0.0	74.8 ± 0.3	94.3 ± 0.8	-	-

Values are given as mean ± standard deviation. If the letters (a–c) used as superscripts are the same, there is no difference between samples produced with different formulations (p > 0.05).

. The melting rate values are also given in detail graphically in Figure 1. The initial melting time of DIC was higher compared to the plant-based ice cream products, i.e., HIC and FHIC. Upon examining the graph depicting melting rates, it was observed that DIC exhibited superior durability, followed by HIC and FHIC, respectively. It was determined that no melting was observed in DIC in the first 15 min when they were kept at room temperature, 50% melting occurred after 60 min, and approximately 14% of the ice cream did not melt completely even after 90 min. In contrast, 8% of HIC and 20% of FHIC melted in the first 15 min. When they waited for 60 min, 13% of HIC did not melt, while only 6% of FHIC remained unmelted. DIC melted completely in 99 min, HIC in 68 min, and FHIC in 67 min. The highest volume increase values were also determined in DIC. This means that DIC trapped more air, also contributing to slower melting. The diminished structural stability and decreased volume expansion of HIC and FHIC may result from the inferior emulsion and foam formation activity and stability characteristics of partially hydrolyzed hazelnut proteins in comparison to cow’s milk proteins. This also leads to faster melting. In particular, the fact that fermented hazelnut proteins contain higher soluble protein and more small molecular weight peptides may also lead to weaker functional properties of fermented hazelnut cake. In addition, the lower carbohydrate and dietary fiber content of plant-based ice creams may also influence their rapid melting. It is also thought that the highest total solid content in DIC may have an effect on these results. The milk proteins, such as casein and whey, might form a stronger network in DIC. Dairy fats might also lead to better emulsion stability than coconut oil. The coconut may crystallize differently from milk fat, and this may affect the texture of HIC and FHIC. Moreover, DIC had higher total solid content, which means less free water, thus supporting slower melting than vegan ice cream.

The initial melting time, melting rate, and volume increase values (overrun) of ice creams are directly related to the quality of the ice cream. A high initial melting time and a low melting rate indicate that the ice cream maintains its shape for a longer time, which is a desired feature [45]. The volume increase and the distribution of air bubbles within the structure that occur during ice cream production are parameters that directly affect the heat flow into the ice cream [25]. It has been observed in many studies that ice cream with low volume increase values melts faster [45,46]. The greater the amount of air bubbles in the ice cream, the less heat is transferred into the ice cream and the longer the melting time of the ice cream [47]. Low volume increase may also be caused by the ice cream mixture not being stable or consistent during mixing [48]. In addition, the fat content of ice cream, the type of fat, the type and amount of emulsifier used, and the dimensions of the emulsions formed are also related to the volume increase and melting rate [46,47,48,49,50]. The total concentration of soluble matter in ice cream significantly influences its structure, as a lower soluble matter content results in a higher water content, leading to increased freezing and the formation of more ice crystals inside the ice cream [45,47]. Plant stabilizers, such as gums, fibers, etc., may be added to reduce free water, the protein sources may be optimized by combining pea protein/soy protein and hazelnut for better emulsification, and fat composition may be adjusted by blending coconut oil with cocoa butter for better crystallization in vegan ice creams.

3.5. Textural Properties of Ice Cream Products

The texture of ice cream is one of the most important quality attributes that affect consumer preference [51]. Texture is considered the sensory reflection of the structure; therefore, the creation and maintenance of the optimum ice cream structure is critical for maximum textural quality. The colloidal structure of ice cream starts with a simple emulsion mixture with a separate phase consisting of partially crystallized fat globules surrounded by an interfacial layer of proteins and surfactants [47]. The structural properties of DIC, HIC, and FHIC were examined by texture analysis (Figure 2). Hardness is the resistance of ice cream to deformation when an external force is applied [25]. The hardness levels of HIC and FHIC are quite close to each other and significantly higher than DIC (p < 0.05). Leahu et al. (2022) attributed the differences in hardness between ice creams obtained from almond milk and hemp milk to different types and amounts of fat and protein in the products [52]. They reported that this situation affected the formation of ice cream crystals, thus increasing the hardness and consistency of ice cream. The hardness values of soy ice cream and sesame ice cream samples were reported as 1442 g and 1820 g, respectively [53]. Like our study, the hardness value of traditional ice cream (836 g) was significantly lower (p < 0.05). These values are in accordance with the hardness values of our DIC, HIC, and FHIC. Hardness can be considered as a reflection of the mixture of components (fat, protein, sugars, and hydrocolloids) and the process conditions (homogenization, ripening, and freezing) of the final product. The hardness of ice cream is influenced by various elements, including the types and quantities of protein and fat, the production and dispersion of ice crystals, fat instability, and the volume of the ice phase [54]. In vegan ice cream, the hazelnut proteins might form a more rigid network than dairy proteins. In addition, coconut oil might lead to a firmer texture at serving temperatures due to a higher melting point than milk fat. Vegan ice creams may have larger or more ice crystals due to differences in freezing behavior. It should also be noted that HIC and FHIC had a lower overrun, which could contribute to a denser and harder texture. Fermentation may also affect the hardness since it breaks down the proteins into smaller peptides, which might reduce elasticity and increase brittleness.

Adhesiveness is the force required to separate food from the surface of other materials [55]. In the case of ice cream, the adhesiveness value represents the additional energy required to overcome the attractive forces between the ice cream and the mouth [45]. The adhesiveness values of DIC ranged from 0.86 to 2.75 mJ, which is significantly higher than HIC and FHIC. Ghaderi et al. (2021) produced soy (4.2 mJ), sesame (8.7 mJ), and traditional dairy (6.9 mJ) ice creams that had considerably higher adhesiveness values compared to our study [53]. Poursani et al. (2020) stated that adhesiveness values may increase when the protein–protein network structure develops in ice creams [55]. Barros et al. (2021) stated that adhesiveness values increased in ice cream samples to which whey powder was added and that this was due to the increase in the total solid content [56]. In addition, Karaca et al. (2009) attributed this increase to the powder added to the ice cream interacting with water, other proteins, and flavor components and affecting the texture [57]. Their findings indicated that the rise in viscosity enhances the adhesiveness of ice cream, aligning with our conclusions about its flow behavior. DIC and HIC exhibit markedly greater viscosity compared to FHIC, correlating with elevated adhesiveness values relative to FHIC. Due to protein breakdown from fermentation and weakening of the matrix, FHIC had the lowest adhesiveness. Milk proteins (casein/whey) may form stronger protein–protein networks, thus increasing adhesiveness, whereas hazelnut proteins may lack cohesive interactions such that they can reduce adhesiveness. Dairy fat destabilization also creates a smoother and creamier melt, thus enhancing adhesiveness, but coconut oil may not destabilize as effectively, leading to less sticky textures.

The higher hardness values in vegan ice cream may be perceived as “less creamy” by consumers. In order to overcome this problem, fat blending to soften texture, hydrocolloid addition to improve smoothness, and air incorporation to increase overrun for reducing density might be implicated. The lower adhesiveness in vegan ice cream may reduce richness perception. The incorporation of plant-derived emulsifiers (e.g., lecithin) to enhance protein–oil interactions, the application of high-viscosity stabilizers (e.g., xanthan gum) to improve adhesion, and the mixture of hazelnut protein with pea or soy protein for superior network formation may be utilized to augment the perception of richness.

3.6. Antioxidant Activity Profiles of Ice Cream Products Before and After In Vitro Digestion

The antioxidant capacities of DIC, HIC, and FHIC were evaluated using ABTS⁺, DPPH, and hydroxyl radical scavenging assays (

Table 7. Antioxidant activities of ice cream products.

Ice Cream	ABTS+ Radical Scavenging Activity (μmol Trolox/g)	DPPH Radical Scavenging Activity (μmol Trolox/g)	Hydroxyl Radical Scavenging Activity (mg Ascorbic Acid/g)	Metal (Fe+2) Chelating Activity (%)
DIC	2.25 ± 0.04 a	N.D.	183.7 ± 2.9 c	N.D.
HIC	2.20 ± 0.05 a	0.68 ± 0.28 a	193.7 ± 3.4 a	31.26 ± 1.10
FHIC	2.91 ± 0.00 a	0.73 ± 0.23 a	188.4 ± 1.8 b	2.55 ± 0.09
In vitro digestion
DIC	11.87 ± 0.39 c	1.44 ± 0.04 b	N.D.	N.D.
HIC	4.01 ± 0.09 b	1.71 ± 0.06 c	N.D.	N.D.
FHIC	10.03 ± 0.13 c	4.73 ± 0.07 d	N.D.	N.D.

Values are given as mean ± standard deviation. If the letters (a–d) used as superscripts are the same, there is no difference between the samples produced with different formulations (p > 0.05). N.D.: Not detected.

). There were no significant differences (p > 0.05) in ABTS⁺ radical scavenging activity among samples, with values ranging from 2.20 ± 0.05 to 2.91 ± 0.00 μmol TE/g. DIC exhibited no detectable DPPH activity, while HIC and FHIC showed comparable but low activity. Vegan ice creams (HIC and FHIC) demonstrated significantly higher activity (183.74 ± 2.93–193.71 ± 3.43 mg AAE/g) than DIC (p ≤ 0.05), suggesting greater phenolic or peptide-based antioxidant contributions. DIC showed no metal chelation capacity, likely due to the absence of chelating phytochemicals. On the other hand, HIC exhibited ~12-fold higher chelating activity than FHIC, possibly due to structural modifications in fermented hazelnut proteins that reduced metal-binding efficiency. In post-digestion, no chelating activity was detected in any sample, likely because bioactive chelators might be hydrolyzed into inactive fragments. It should also be noted that complete metal ion sequestration occurred in the digestion medium, which caused measurable activity. Digestion significantly enhanced (1.5–6.5-fold) the antioxidant potential of all samples (p ≤ 0.05), attributable to the enzymatic release of bioactive peptides and phenolic compounds. ABTS⁺ radical scavenging activity increased 5.3-fold, while DPPH activity (initially undetectable) rose to 1.44 ± 0.04 μmol TE/g in DIC, suggesting milk protein-derived peptides contribute to antioxidant effects. FHIC had 2.5 times higher ABTS⁺ and 2.8 times higher DPPH radical scavenging activity than HIC in post-digestion, indicating fermentation may enhance bioactive peptide release [9]. Digested FHIC was comparable with DIC in ABTS⁺ radical scavenging activity (p ≤ 0.05) and surpassed it in DPPH radical scavenging activity, highlighting its potential as a functional alternative. Here, digestive enzymes (e.g., pepsin and pancreatin) might enhance the radical quenching capacity of the samples by breaking down polymeric proteins into low-MW peptides. Plant-based matrices (hazelnut cake) may also release bound phenolics during digestion, amplifying their antioxidant effects. Fermentation may alter protein structures, resulting in more digestible bioactive peptides; hence, FHIC exhibited reduced native chelating activity alongside an increase in post-digestion antioxidant capacity.

3.7. Enzyme Inhibition Profiles of Ice Cream Products Before and After In Vitro Digestion

The inhibitory effects of DIC, HIC, and FHIC on α-amylase, α-glucosidase, angiotensin-converting enzyme (ACE), acetylcholinesterase (AChE), and lipase activities are presented in

Table 8. Enzyme inhibition activities of ice cream products.

Ice Cream	α-Amylase/α-Glucosidase Inhibition (%)	ACE Inhibition (mg captopril/g)	AChE Inhibition (%)	Lipase Inhibition (%)
DIC	N.D.	N.D.	N.D.	45.12 ± 0.34 a
HIC	N.D.	0.76 ± 0.01 d	N.D.	39.40 ± 2.98 b
FHIC	N.D.	N.D.	N.D.	39.94 ± 2.23 b
In vitro digestion
DIC	N.D.	7.10 ± 0.47 b	51.56 ± 0.23 c	N.D.
HIC	N.D.	7.42 ± 0.08 a	74.37 ± 1.56 a	N.D.
FHIC	N.D.	4.85 ± 0.03 c	57.69 ± 0.84 b	N.D.

Values are given as mean ± standard deviation. If the letters (a-d) used as superscripts are the same, there is no difference between the samples produced with different formulations (p > 0.05). N.D.: Not detected.

. None of the ice cream samples exhibited α-amylase or α-glucosidase inhibition before or after in vitro digestion. Similarly, no significant inhibition was detected in the digestion medium, whereas hazelnut cake and fermented hazelnut cake demonstrated notable enzyme inhibition [9]. The absence of inhibitory activity in ice cream samples may be attributed to the dilution effect, thermal and processing degradation, or structural disruption during digestion. Bioactive enzyme inhibitors from hazelnuts may have been significantly diluted within the ice cream matrix, reducing their concentration below detectable levels. The enzyme-inhibitory compounds (e.g., polyphenols and peptides) naturally present in hazelnut may have been substantially diluted in the ice cream matrix, reducing their concentration below the detection threshold. Jakobek (2015) reported that milk proteins (particularly caseins) can bind and precipitate polyphenolic compounds through hydrophobic interactions and hydrogen bonding, thus reducing measurable activity [58]. Moreover, the high fat content of dairy matrices can partition lipophilic hazelnut bioactives into the lipid phase, making them inaccessible for aqueous extraction [59]. The complex ice cream matrix (containing proteins, fats, and emulsifiers) can physically encapsulate bioactive compounds, as shown in studies on nutraceutical delivery in frozen desserts [60]. In addition, inhibitory compounds may have been inactivated during ice cream production due to processing conditions, such as pasteurization, homogenization, aging, and freezing, or bioactive peptides responsible for inhibition may have been hydrolyzed or structurally altered in the gastrointestinal environment [1,61].

Before digestion, negligible ACE inhibition (0.76 ± 0.01%) was noted exclusively in HIC, whilst other samples exhibited no measurable action, perhaps due to quantities falling below the assay’s detection threshold. Post-digestion, ACE inhibition increased significantly, with values ranging from 4.85 ± 0.03% to 7.42 ± 0.08%. The observed increase in ACE-inhibitory activity following in vitro digestion is primarily attributed to the enzymatic hydrolysis of milk and hazelnut proteins, which liberates bioactive peptide sequences capable of ACE inhibition [62,63,64]. The highest inhibition was observed in HIC, whereas FHIC exhibited the lowest inhibition. Notably, DIC, which lacked ACE inhibition before digestion, displayed substantial inhibitory activity post-digestion, likely due to the release of bioactive peptides from hydrolyzed milk proteins—a well-documented source of ACE-inhibitory peptides [65,66]. Despite expectations of enhanced bioactivity in FHIC due to fermentation, its ACE inhibition was only 65% of HIC’s activity, suggesting that inhibitory potency depends more on peptide sequence and amino acid composition than on peptide abundance.

No AChE inhibition was detected in any ice cream sample before digestion. However, post-digestion, inhibition levels ranged from 51.56 ± 0.23% to 74.37 ± 1.56% at a concentration of 1 mg/mL. The lowest inhibition was observed in digested DIC, with a slight increase in digested HIC. Consistent with ACE inhibition trends, fermentation-derived peptides in FHIC may contribute to higher AChE inhibition. In contrast to the observed increase in AChE inhibition following the digestion of ice cream-derived peptides, gastrointestinal digestion has been shown to significantly reduce (p < 0.05) the AChE-inhibitory activity of plant phenolic-rich extracts (e.g., carob, grape pomace, and chestnut shell). This is likely due to the structural degradation of labile phenolic compounds (e.g., gallotannins and anthocyanins) under digestive pH conditions, altered bioavailability from interactions with digestive enzymes and bile salts, and depolymerization of high-molecular-weight polyphenols into inactive monomers [67,68].

Unlike other enzymes, lipase inhibition was observable in undigested ice cream samples (39.40 ± 2.98% to 45.12 ± 0.34% at 0.0025 g/mL) but was not present after digestion, indicating the degradation of lipase-inhibitory components by digestive enzymes or interactions with other gastrointestinal elements. Notably, the digestion medium itself exhibited 32.12% lipase inhibition, indicating potential interference from digestive components. This pattern diverges from trends observed with phenolic-rich plant extracts. The lipase inhibition increased post-digestion due to the release of bioactive phenolics (e.g., gallic acid and epigallocatechin gallate in green tea) because digestive enzymes hydrolyze polyphenol complexes, thus enhancing their activity [69,70]. A similar loss of lipase inhibition post-digestion was reported for milk phospholipids due to bile salt displacement of inhibitors at lipid–water interfaces [71].

3.8. Sensorial Evaluation of Ice Cream Products

A trained panel (n = 13) evaluated three ice cream formulations in two sessions: (1) Plant-based ice cream: Hazelnut cake (HIC) and fermented hazelnut cake (FHIC) variants; (2) Dairy-based control ice cream (DIC): Traditional formulation (cow’s milk cream, sugar, and salep). The panelists frequently associated DIC with “traditional childhood ice cream” due to its familiar flavor and texture. It received the highest scores for color (9.64), taste (9.26), aroma (8.99), and chewy texture (7.75) (Figure 3). Its melt-in-mouth and smoothness were comparable to plant-based variants. No negative comments were reported for HIC, and the presence of foreign compound/flavor detection was minimal (1.59). Conversely, FHIC was characterized as “very interesting”, yet polarizing, with certain panelists expressing that they “would find it challenging to consume regularly.” Among all samples, the presence of foreign compound/flavor (8.66) detection was highest for FHIC, while it was zero for DIC. The overall acceptability of the ice cream products is ranked as follows: DIC > HIC > FHIC (based on panelists’ rankings).

4. Conclusions

The vegan ice cream formula, derived from hazelnut oil cake and fermented hazelnut oil cake, was successfully manufactured. Nonetheless, DIC surpassed vegan alternatives in melting resistance and overrun, owing to enhanced protein functioning and elevated solids. HIC and FHIC possess the potential for commercial production for people adhering to a vegan diet due to their augmented health benefits.

This study underlines the difficulties in recreating the structural properties of dairy ice cream using plant-based alternatives while offering insights for future formulation enhancements. DIC had a softer and stickier texture, which is preferred for creaminess. Future vegan ice cream formulations should prioritize protein optimization (blends and functional modifications), improvements to the fat system (softer fats and enhanced emulsification), and the incorporation of stabilizers to replicate a dairy-like texture. Vegan ice creams (HIC/FHIC) demonstrated enhanced hydroxyl radical scavenging compared to DIC, presumably attributable to plant phenolics. Digestion significantly improved antioxidant activity, with FHIC exhibiting the highest post-digestion DPPH activity, highlighting fermentation’s capacity to enhance bioaccessibility. Metal chelation was diminished after digestion, requiring additional investigation into stable chelating compounds (e.g., encapsulation). Dairy and fermented vegan ice creams exhibited similar ABTS⁺ activity post-digestion, indicating that FHIC serves as a functionally competitive alternative. The fermentation procedure diminished the overall acceptability of vegan ice cream, with HIC being comparable to DIC for the presence of foreign compounds/flavors, which were nearly absent.