Effects of fermentation time and sugar type on the processing characteristics and quality of special low-protein steamed sponge cakes

Abstract

The development of special low-protein foods has gained significant attention in medical and food science research. There is a notable lack of research on these foods with protein contents under 1 g/100 g. This study aimed to develop special low-protein steamed sponge cakes (LPSSCs) and analyze the effects of fermentation time and sugar type on their properties. The LPSSCs exhibited low protein contents of 0.55–0.68 g/100 g. Among the tested fermentation times (0–120 min), 45 min was optimal for achieving desirable sensory properties, including chewing toughness. Both fermentation time and sugar addition induced significant differences in volatile organic compound profiles. Fermentation times of 45 and 90 min induced higher starch hydrolysis compared to 120 min. High sugar-to-low-protein-powder mass ratios (25% and 35%) led to greater starch hydrolysis than the 15% ratio than corresponding trehalose-to-low-protein-powder mass ratios. This work provides useful insights for understanding the component-processing-quality relationships of special low-protein foods.

Highlights

• •Low-protein steamed sponge cakes (LPSSCs) had protein contents of 0.55–0.68 g/100 g.
• •Fermentation time of 45 min induced the best sensory properties.
• •Both fermentation times and sugar type affected volatile organic chemicals.
• •LPSSCs fermented for 45/90 min maximized starch hydrolysis (36.4%–37.9% at 3 h).
• •Sucrose induced higher starch hydrolysis of LPSSCs than trehalose did.

1. Introduction

Special low-protein foods serve as essential, lifelong therapeutic nutritional interventions for patients requiring strict dietary restriction of proteins and specific amino acids. For individuals with chronic kidney disease (CKD), these foods help reduce the production of urea and other nitrogenous waste products, thereby alleviating renal burden (Mafra, Brum, Borges, Leal, & Fouque, 2025). For patients with amino acid metabolism disorders (Handoom et al., 2018) or organic acidemias (Garcia-Arenas et al., 2023), they limit the intake of problematic precursors, mitigating associated risks such as neurodevelopmental injury. Phenylketonuria (PKU), a classic amino acid metabolism disorder requiring strict phenylalanine restriction, is a prime example (Trepp et al., 2024). Consequently, the development and clinical application of special low-protein foods have garnered significant attention in the field of medical nutrition.

Many researchers have explored the preparation and properties of special low-protein foods. Low-protein pastas were developed with lower protein contents (2.94–5.76 g/100 g dry pasta) than that of regular pasta (Yaseen & Shouk, 2011). Low-protein breads were developed with relatively low protein contents (1.40–1.43 g/100 g bread) (Scortegagna et al., 2020). Low-protein biscuit was developed with a relatively low protein content of 1.27 g/100 g biscuit (Azaripour & Abbasi, 2020). Recently, our group developed low-protein steamed buns with low protein contents of 0.50–0.51 g/100 g steamed buns (Shi, Zi, Liu, Cai, & Zhong, 2025). These efforts collectively demonstrate the feasibility of achieving low-protein foods through tailored processing. Nonetheless, achieving protein contents below 1.0 g/100 g in palatable, structured cereal products continues to be a key objective in the field.

Several limitations hinder the development of ideal low-protein foods. First, proteins play a triple role as nutrients, texture agents, and flavor agents (Foegeding, 2015), making it challenging to maintain low-protein foods comparable to that of their conventional counterparts. Furthermore, studies on cereal-based low-protein foods remain scarce. Finally, existing research has predominantly focused on the effects of raw materials—such as starch types (Azaripour & Abbasi, 2020; Shi et al., 2025; Yaseen & Shouk, 2011) and potato varieties (Scortegagna et al., 2020)—rather than on elucidating fundamental processing mechanisms. Consequently, there is a clear need to develop foods with protein levels below 1.0 g/100 g and to rigorously analyze how critical processing and component parameters govern their properties.

The steamed sponge cake is a traditional delicacy widely popular in South Asia, prized for its distinctive honeycomb structure, soft texture, and notable elasticity (Cai et al., 2024). It can be prepared using various flour blends, including potato/rice (Cai et al., 2024) and wheat/banana combinations (Sinaga, 2023). Despite its culinary significance, a systematic investigation into the preparation methods and physicochemical properties of steamed sponge cakes remains lacking. Furthermore, conventional formulations typically exhibit relatively high protein contents (e.g., 6.9–8.7 g/100 g) (Fitrasyah et al., 2023), which limits their suitability for protein-restricted diets. Therefore, it is crucial to develop special low-protein steamed sponge cakes (LPSSCs) and to systematically analyze the effects of key processing parameters on their production and final properties.

This study aimed to develop low-protein steamed sponge cakes (LPSSCs) with a protein content of <1.0 g/100 g and to investigate the effects of fermentation time and sugar type (Sucrose, Suc; trehalose, Tre) on their preparation process and final properties. The preparation was optimized by examining processing factors and sugar-to-low-protein powder (LPP) mass ratio. The nutritional composition was analyzed to verify the low-protein status of the products. Additionally, the effects of freezing duration and post-steaming cooling time on LPSSC quality were evaluated to explore practical storage and consumption scenarios. The developed LPSSCs were characterized through (1) descriptive sensory analysis for sensory properties, (2) headspace–gas chromatography–ion mobility spectrometry (HS-GC-IMS) for volatile organic compound profiling, and (3) an in vitro simulated intestinal model for starch digestibility assessment.

2. Materials and methods

2.1. Materials

This study utilized three commercially available low-protein powders (LPPs) and one wheat flour. Aishushu™ LPP for Steamed Bun (ALPPSB; 0.47 g protein/100 g powder) and Aishushu™ LPP for Noodle/Dumpling (ALPPND; 0.22 g protein/100 g powder) were procured from Shanghai Zhunshen Food Dongtai Co., Ltd. (Jiangsu, China). Zhongen™ LPP for Steamed Bun (ZLPPSB; 0.32 g protein/100 g powder) was obtained from Cangzhou Enji Biological Products Co., Ltd. (Hebei, China). The ingredient lists were as follows: ALPPSB—wheat starch, potato starch, corn starch, acetylated starch, pea starch, edible glucose, glucono-delta-lactone, calcium carbonate, and sodium bicarbonate; ALPPND—wheat starch, potato starch, acetylated starch, corn starch, pea starch, acetylated mono- and diglycerides, sodium alginate, guar gum, and ascorbic acid; ZLPPSB—wheat starch, pregelatinized starch, fructooligosaccharides, oat fiber, pumpkin powder, and glucoamylase. Jinlongyu™ Low-Gluten Wheat Flour for Pastry (JLGWFP; 8.0 g protein/100 g flour) was sourced from Yihai Kerry Food Marketing Co., Ltd. (Shanghai, China).

Sucrose (First-grade Yutang™ white granulated sugar) was supplied by Dafeng YinMore Sugar Industry Co., Ltd. (Jiangsu, China). Trehalose (food-grade Treha™ crystalline trehalose dihydrate) was provided by Hayashibara Co., Ltd. (Okayama, Japan). Instant dry yeast (Saccharomyces cerevisiae) was purchased from Angel Yeast Co., Ltd. (Hubei, China).

2.2. Preparation of LPSSCs

The preparation of low-protein steamed sponge cakes (LPSSCs) involved three main steps: batter preparation, fermentation, and baking (Fig. 1). Briefly, 12.5 g of sucrose was dissolved in 40 mL of water. Then, 50 g of ALPPSB was gradually incorporated with stirring to form a homogeneous batter. Separately, 0.75 g of instant dry yeast was dispersed in 5 mL of water. This yeast suspension was then mixed into the batter (resulting in a total water addition of 45 mL, equivalent to 90% of the LPP weight). The batter was fermented at 40 °C in a preheated electric oven (PT3520W-G, Media) for 45 min. After fermentation, the batter was gently stirred to remove large air bubbles. Subsequently, 40 g of the fermented batter was transferred into custom-designed, donut-shaped carbon-steel molds (inner diameter: 6.8 cm, depth: 2.2 cm). The molds were covered with carbon-steel lids and baked at 150 °C in the preheated oven for 45 min. After baking, the LPSSCs were demolded, cooled to room temperature, sealed in seven-layer PE/PE/PE/TIE/PA/TIE/PA composite grid-embossed bags (Zhejiang Mingke Plastic Industry, Taizhou, China), and stored at −18 °C.

Fig. 1.

Schematic of the preparation process for low-protein steamed sponge cakes (LPSSCs). Detailed procedures are provided in the subsection “Preparation of LPSSCs.” The base low-protein powder (LPP) used was Aishushu™ Low-Protein Powder for Steamed Bun (ALPPSB). Key process parameters were set as follows: Water/ALPPSB mass ratio: 90%; Sucrose/ALPPSB mass ratio: 25%; Fermentation time: 45 min; Baking temperature: 150 °C.

Three LPPs and one JLGWFP were used in the above procedures to analyze their effects on the steamed sponge cakes. Different water amounts (60%, 90%, and 120% of ALPPSB) were used in the above procedures to analyze their effects on the steamed sponge cakes. Different sugar amounts (15%, 25%, and 35% of ALPPSB) were used in the above procedures to analyze their effects on the steamed sponge cakes. Different fermentation times (0, 45, 90, and 120 min) were used in the above procedures to analyze their effects on the steamed sponge cakes. Different baking temperatures (120, 150, and 180 °C) were used in the above procedures to analyze their effects on the steamed sponge cakes. Two sugars (Suc and Tre) were used in the above procedures to analyze their effects on the steamed sponge cakes.

2.3. Texture analysis

After cooling to room temperature for approximately 1 h, the LPSSCs were sectioned from the top into four equal parts. Textural properties were analyzed on one of the quarters (placed face down) using a TA.GEL texture analyzer (Shanghai BosinTech, China) equipped with a texture profile analysis (TPA) module (Chen et al., 2024). All measurements were completed within 30 min after sectioning. A cylindrical TA/36R probe was used with pre-test, test, and post-test speeds set at 3.00, 1.00, and 1.00 mm/s, respectively. The samples were compressed to 50% of their original height with a trigger force of 0.0981 N. Data were processed using the instrument's proprietary software.

2.4. Nutritional composition

The protein, moisture, and ash contents were analyzed by Shanghai Weipu Testing Certification Co., Ltd. (Shanghai, China) following Chinese National Standards: GB 5009.5–2016 (Kjeldahl method for protein), GB 5009.3–2016 (direct drying method for moisture), and GB 5009.4–2016 (total ash method for ash). The laboratory procedures are accredited by both the China Metrology Accreditation (CMA) and the China National Accreditation Service for Conformity Assessment (CNAS). Fat content was considered negligible (assigned as 0 g/100 g) as no fats were present in the raw materials. Finally, energy and carbohydrate contents were calculated in accordance with the Chinese national guidance standard GB/Z 21922–2008 (Basic terminology and definitions of nutritional components in foods).

2.5. Frozen storage

Freshly prepared LPSSCs were individually sealed in seven-layer PE/PE/PE/TIE/PA/TIE/PA composite grid-embossed bags and stored at −18 °C for designated periods (7, 21, 49, and 120 days). Following frozen storage, the samples were thawed by steaming over boiling water in an electric steamer (model MZ-ZGC322301, Midea Group Co., Ltd., China) for 15 min. Their textural properties were then assessed according to the method described in Section 2.3. Freshly prepared LPSSCs (day 0) served as the control group for comparison.

2.6. Cooling time

The freshly prepared LPSSCs were individually sealed in seven-layer PE/PE/PE/TIE/PA/TIE/PA composite grid-embossed bags and stored at −18 °C for 3 days. After frozen storage, the samples were steamed over boiling water in an electric steamer (model MZ-ZGC322301, Midea Group Co., Ltd., China) for 15 min. They were then allowed to cool at room temperature for various durations (0.25, 0.5, 1, 2, 4, 8, 24, and 48 h). Immediately after each cooling period, the textural properties of the LPSSCs were measured following the procedure described in Section 2.3.

2.7. Descriptive sensory analysis

The frozen LPSSCs were steamed over boiling water in the MZ-ZGC322301 electric steamer for 15 min. After steaming, the samples were placed in disposable bowls labeled with random three-digit codes. Sensory evaluation was conducted within 2 h via descriptive analysis, following the Chinese standard GB/T 35991–2018 “Inspection of grain and oils—Steamed buns of wheat flour processing quality evaluation” (Guo, Yin, Cheng, Chen, & Ye, 2022). No formal ethical approval was required because it did not involve medical research according to the Hospital Ethics Committee (Streule et al., 2024). All the participants were ask to sign informed consent forms to give their consent to take part in the sensory study and use their information. The authors confirmed an appropriate protocol was applied to protect the he rights and privacy of all participants were utilized during the execution of the research.

Twelve trained panelists (aged 20–35 years), who were fully informed about the study and provided written consent, participated in the evaluation. The panel size of 12 meets the minimum requirement (≥10) specified in GB/T 35991–2018. Panelists were trained to assess the LPSSCs based on eight attributes: specific volume (15 points), finger-press resilience (10 points), color (10 points), appearance (10 points), porous structure (20 points), chewing toughness (15 points), teeth stickiness (10 points), and taste (10 points). The overall sensory score for each sample was calculated as the sum of the scores for all eight attributes.

The eight sensory attributes were evaluated according to distinct criteria. The scoring guidelines for each attribute are detailed below: i) Large volume (12–15 score), ii) medium volume (8–11 score), and iii) small volume (5–7 score). As for “finger press resilience”, the evaluation criteria were: i) Good (8–10 scores), ii) weak (6–7 scores), and iii) not rebound or press difficult (4–5 scores). As for “color”, the evaluation criteria were: i) good gloss (8–10 scores), ii) slightly dark (6–7 scores), and iii) grey (4–5 scores). As for “appearance”, the evaluation criteria were: i) Smooth surface (8–10 scores); ii) shrinkage, collapse, bubble, hard or hotspots (4–7 scores). As for “Porous structure”, the evaluation criteria were: i) Fine and uniform pores (18–20 scores); ii) fine and uniform pores with large individual pores (13–17 scores); iii) Basically uniform pores (10–12 scores); iv) uneven pores (5–9 scores). As for “chewing toughness”, the evaluation criteria were: i) Strong toughness (10–15 scores); ii) general toughness (5–9 scores); iii) poor, dross, or dry toughness (1–4 scores). As for “teeth stickiness”, the evaluation criteria were: i) Refreshing and non-sticky feeling (8–10 scores); ii) slightly sticky feeling (6–7 scores); iii) sticky feeling (4–5 scores). As for “Taste”, the evaluation criteria were: i) good taste (8–10 score), ii) tasteless (5–7 score), and iii) uncomfortable taste (1–4 score).

2.8. Headspace-gas chromatography-ion mobility spectrometry

The LPSSCs were diced into approximately 2 × 2 mm pieces. A 2.0 g aliquot of the diced sample was placed into a 20 mL headspace vial, which was then immediately sealed. Following incubation at 80 °C for 30 min, volatile compounds were analyzed using a FlavourSpec® headspace gas chromatography–ion mobility spectrometry (HS-GC-IMS) system (G.A.S. Gesellschaft für analytische Sensorsysteme mbH, Dortmund, Germany) as described in our previous work (Chen et al., 2021). The injection volume was 500 μL, and the syringe temperature was maintained at 85 °C.

The following gas chromatography (GC) conditions were applied. Separation was performed using an MXT-WAX capillary column (30 m × 0.53 mm, 1.0 μm film thickness; Restek, Bellefonte, PA, USA) maintained at 60 °C. High-purity nitrogen (≥99.999%) served as the carrier gas, with a flow program set as follows: held at 2.0 mL/min for 2 min, increased to 10.0 mL/min over 8 min, further increased to 100.0 mL/min over 10 min, and then maintained at 100.0 mL/min for 30 min. The ion mobility spectrometry (IMS) conditions were as follows: High-purity nitrogen (≥99.999%) was used as the drift gas at a constant flow rate of 75.0 mL/min. The IMS drift tube temperature was set to 45 °C.

VOCs were analyzed using the commercial software VOCal (version 0.4.03) with the following four plug-ins: (i) Reporter was used for generating topographic plots and conducting differential analysis; (ii) Gallery Plot was used for fingerprint spectrum comparison and VOC differentiation; (iii) Dynamic PCA was used for principal component analysis; (iv) Fingerprint Similarity Analysis (FSA) was used for evaluating fingerprint similarity. Qualitative identification of VOCs was performed within VOCal by matching against the built-in NIST 2020 and IMS databases.

2.9. In vitro starch digestibility

The in vitro starch digestibility of the LPSSCs was evaluated using a simulated intestinal digestion model, as described in previous studies (Dhital, Lin, Hamaker, Gidley, & Muniandy, 2013; Guo et al., 2022; Shaha, Torbati, & Frick, 2021; Wang, Tian, Zhang, Tian, & Zhang, 2024). Briefly, 10 mL of 0.2 mol/L sodium acetate-acetic acid buffer (pH 5.2; Shanghai Macklin) was added to 129 mg of porcine pancreatic α-amylase (9 U/mg; Shanghai Yuanye, China). The mixture was magnetically stirred at 300 rpm for 30 min and then centrifuged at 2039 ×g for 10 min (centrifuge model 5810 R, Eppendorf, Germany). The supernatant was collected and mixed with 1 mL of Wanvi® glucoamylase solution (2000–3300 U/mL; Shanghai Linghan Scientific Instrument, China) to obtain the final α-amylase/glucoamylase enzyme cocktail.

The frozen LPSSCs were first steamed over boiling water in an MZ-ZGC322301 electric steamer for 15 min and cooled to room temperature. The cooled cakes were diced into pieces of approximately 2 × 2 mm. Then, 200 mg (recorded as $W_1$) of the diced sample was transferred into a tube containing 5 mL of deionized water. After adding seven glass beads (4.00 mm diameter), the mixture was boiled for 30 min, followed by the addition of 15 mL of 0.2 M sodium acetate-acetic acid buffer, resulting in a total liquid volume ($V$) of 20.0 mL. The mixture was equilibrated at 37 °C with shaking at 100 rpm in a water bath shaker (DKZ-1, Shanghai Yiheng, China) for 10 min. Subsequently, 0.1 mL of the pre-heated (37 °C, 10 min) α-amylase/glucoamylase enzyme cocktail was added, and digestion was allowed to proceed at 37 °C with continuous shaking for 180 min. At predetermined time points, 0.2 mL aliquots were withdrawn and immediately mixed with 0.8 mL of absolute ethanol (dilution factor $D$ = 5). The diluted samples were centrifuged at 3889 ×g for 5 min in a MiniSpin centrifuge (Eppendorf, Hamburg, Germany), and the supernatants were collected for glucose quantification. Glucose concentration ($C$, mg/mL) in the digestion mixture was determined using a d-glucose (GOPOD) assay kit (Biosharp, Labgic Technology, Shanghai, China) according to the following equation:

$$C\left(g/L\right)=C_{\mathit{Standard}}\times \frac{A_{\mathit{Supernatant}}-A_{\mathit{Water}}}{A_{\mathit{Standard}}-A_{\mathit{Water}}}\times D$$ (1)

where $C_{\mathit{Standard}}$ is the concentration (1 g/L) of the standard glucose solution, $A_{\mathit{Sample}}$ is the absorbance of the supernatants, $A_{\mathit{Water}}$ is the absorbance of the ultrapure water, and $A_{\mathit{Standard}}$ is the absorbance of the standard glucose solution.

The moisture content of the LPSSC pieces was determined by oven-drying at 103 °C for 18 h, in accordance with the Chinese National Standard GB 5009.3–2016 “Determination of moisture in foods” (Nie et al., 2022). The starch content (SC) was then calculated using the following equation:

$$\mathit{SC}\left(\%\right)=\frac{W_4-W_2}{W_3}\times 100$$ (2)

where $W_2$ is the mass of the container, $W_3$ is the mass of the initial LPSSC pieces, and $W_4$ is the mass of the container with dried LPSSC pieces after 18 h at 103 °C.

Finally, the starch hydrolysis (SH) percentages were calculated according to the below equation:

$$\mathit{SH}\left(\%\right)=\frac{\mathit{Hydrolyzed\; starch}}{\mathit{Starch\; in\; LPSSC}}=\frac{0.9\times C\times V}{W_1\times \mathrm{SC}}\times 100$$ (3)

In this eq. 0.9 is the conversion factor from glucose molecular mass to starch monomer unit (Jiang, Wang, Ou, & Zheng, 2021).

2.10. Statistical analysis

All experiments were conducted in triplicate, and results are presented as mean ± standard deviation. Statistical differences among groups were analyzed by one-way analysis of variance (ANOVA) followed by Duncan's multiple range test, with a significance level set at p < 0.05.

3. Results and discussion

3.1. Effects of preparation factors

The preparation of low-protein steamed sponge cakes (LPSSCs) consisted of three main stages: batter preparation, fermentation, and baking, as illustrated in Fig. 1 and detailed in Section 2.2. The effects of four key preparation parameters on the physical and textural properties of the LPSSCs were investigated. These included: LPP type (Figs. 2A–E), baking temperature (Figs. 2F–J), fermentation time (Figs. 2K–O), and water-to-ALPPSB mass ratio (Figs. 2P–T).

Fig. 2.

Effects of preparation parameters on the appearance and textural properties of LPSSCs. (A–E) Influence of flour type on digital images (A) and textural properties (B–E). Abbreviations: JLGWFP, Jinlongyu™ low-gluten wheat flour for pastry; ALPPND, Aishushu™ low-protein powder for noodle/dumpling; ZLPPSB, Zhongen™ low-protein powder for steamed bun. (F–J) Influence of baking temperature on digital images (F) and textural properties (G–J). (K—O) Influence of fermentation time on digital images (K) and textural properties (L–O). (P–T) Influence of water/ALPPSB mass ratio on digital images (P) and textural properties (Q–T).

Steamed sponge cakes were prepared using four different powder bases: one conventional low-gluten wheat flour (JLGWFP) and three commercial low-protein powders (LPPs: ALPPSB, ALPPND, ZLPPSB). As shown in Fig. 2A, cakes made with JLGWFP and ALPPSB exhibited a well-formed, uniform external appearance, whereas those prepared with ALPPND or ZLPPSB showed visible surface irregularities. Cross-sectional images, however, revealed that all samples developed a fine, homogeneous porous structure, suggesting that each powder possesses the basic potential to produce aerated cakes under optimized processing conditions.

Texture-profile analysis (Figs. 2B–E) indicated that the textural parameters (e.g., hardness, chewiness) varied significantly with the LPP type. Cakes containing ALPPND and ALPPSB displayed consistently higher texture values than those made with ZLPPSB or JLGWFP. The superior performance of ALPPSB can be attributed to its specific formulation. While proteins normally provide both nutritional and structural functions in baked goods (Foegeding, 2015), the very low protein content in LPPs creates a structural deficit. ALPPSB contains calcium carbonate and sodium bicarbonate—common chemical leavening agents that promote gas formation and improve product volume and texture (Liu, Zhang, Mujumdar, & Yu, 2021; Gensberger-Reigl, Rodrigues Guimarães Abreu, & Pischetsrieder, 2022). The presence of these agents likely compensated for the lack of protein-derived structure, enabling the ALPPSB-based cake to achieve an appearance and internal porosity comparable to the conventional wheat-flour control. Based on these results, ALPPSB was selected as the base LPP for all subsequent experiments.

Three baking temperatures (120, 150, and 180 °C) were evaluated for the preparation of LPSSCs. As shown in Fig. 2F, the surface and cross-sectional color of the cakes shifted from light yellow to dark yellow with increasing temperature. Cross-sectional images indicated that all samples retained a well-developed porous structure, suggesting that baking temperature within this range did not substantially hinder pore formation. However, the sample baked at 180 °C exhibited pronounced yellowing at the edges, which may indicate lipid oxidation—a potential quality and safety concern (Goh et al., 2019). Consequently, 180 °C was deemed unsuitable for LPSSC production. Textural properties of the LPSSCs varied significantly with baking temperature (Figs. 2G–J). Among the three temperatures tested, 150 °C yielded the most favorable texture profile, exhibiting optimal hardness, springiness, and cohesiveness. Based on these results, 150 °C was selected as the standard baking temperature for all subsequent experiments.

LPSSCs were prepared using four fermentation times (0, 45, 90, and 120 min). As shown in Fig. 2K, both the surface and cross-sectional color of the cakes deepened from light to dark yellow with extended fermentation. Prolonged fermentation typically promotes greater batter expansion due to increased gas production (Balasubramanian & Viswanathan, 2007), resulting in greater contact between the batter and the mold lid during baking. This enhanced contact likely contributed to the more pronounced surface coloration observed at longer fermentation times. Cross-sectional images revealed that samples fermented for 45–120 min developed a more uniform and well-defined porous structure compared to the non-fermented (0 min) control, underscoring the importance of adequate fermentation for achieving desirable cake texture. As shown in Figs. 2L–O, the textural properties of the LPSSCs varied significantly with fermentation time. Compared to the unfermented control (0 min), LPSSCs fermented for 45–120 min exhibited lower chewiness (Fig. 2L) and hardness (Fig. 2N), alongside higher resilience (Fig. 2M) and cohesiveness (Fig. 2O). Given that shorter fermentation times are generally preferred in the food industry for safety considerations (Surya, Nugroho, Kamal, & Tedjakusuma, 2023), a fermentation time of 45 min was selected for all subsequent experiments in this study.

LPSSCs were prepared using three water-to-ALPPSB mass ratios (60%, 90%, and 120%). As shown in Fig. 2P, the surface color of the cakes first intensified and then lightened as the water ratio increased. In contrast, the internal color of the cross-sections gradually faded with higher water addition. All samples exhibited well-formed porous structures regardless of the water content. Textural properties were strongly influenced by the water ratio (Figs. 2Q–T). Both chewiness and hardness increased progressively with higher water levels. However, the 90% water ratio produced the highest resilience and cohesiveness, representing an optimal balance between firmness and structural integrity. Based on these findings, a water-to-ALPPSB mass ratio of 90% was selected for all subsequent experiments.

3.2. Effects of sugar/ALPPSB mass ratios

The influence of sugar-to-ALPPSB mass ratios on the preparation and properties of LPSSCs is presented in Fig. 3. All prepared cakes exhibited an appealing appearance with a characteristic bakery yellow color (Fig. 3A). Cross-sectional images, however, revealed that the average pore diameter within the crumb decreased slightly as the sugar proportion increased (for both sucrose and trehalose in the 15–35% range). Previous studies have reported that higher sugar levels can increase porosity in conventional cakes by inhibiting gluten network formation and raising the denaturation temperature of egg proteins (Farzi, Saffari, & Emam-Djomeh, 2015). In the present low-protein system (LPSSCs), sugar may similarly interfere with the limited protein-starch matrix, but here it appears to restrict rather than promote pore expansion, leading to a finer, more compact pore structure.

Fig. 3.

Effects of sugar-to-ALPPSB mass ratio on the appearance and textural properties of LPSSCs. (A) Representative digital images of LPSSCs prepared with different sugar types (sucrose, Suc; trehalose, Tre) and mass ratios (15%, 25%, 35%). (B–E) Textural parameters: (B) chewiness, (C) resilience, (D) hardness, and (E) cohesiveness.

Textural properties were significantly influenced by the sugar-to-ALPPSB mass ratio (Figs. 3B–E). Neither resilience (Fig. 3C) nor cohesiveness (Fig. 3E) showed clear variation across sugar levels (0%, 15–35% sucrose, or 15–35% trehalose). However, chewiness (Fig. 3B) was consistently higher in samples containing 15–35% sugar compared to the sugar-free (0%) control. A notable difference was observed in hardness (Fig. 3D). LPSSCs with 15–35% sucrose/ALPPSB mass ratios exhibited significantly greater hardness than the sugar-free sample, whereas those with 15–35% trehalose/ALPPSB mass ratios did not differ statistically from the control. This result contrasts with reports that sugar generally improves the texture of conventional cakes (Gökçe, Bozkurt, & Maskan, 2023). The discrepancy may be attributed to the very low protein content of ALPPSB, which limits gluten-like network formation, and to the distinct structural and functional properties of sucrose (a glucose-fructose disaccharide) and trehalose (a non-reducing glucose-glucose disaccharide linked by an α,α-1,1-glycosidic bond) (Li, Wang, Wang, Yu, & Zhang, 2024).

3.3. Nutritional composition

The nutritional composition of three LPSSC formulations—with different sugar/ALPPSB mass ratios (0% sugar, 25% sucrose, and 25% trehalose)—was analyzed according to Chinese national standards. The formulation with 0% sugar, 0% yeast, and 0 min fermentation served as the control. As summarized in Table 1, the nutritional profile varied significantly with sugar addition. Compared to the control, all LPSSCs containing yeast exhibited higher energy, protein, ash, and carbohydrate contents, attributable to the nutritional contribution of yeast as an ingredient (Yamada & Sgarbieri, 2005). Furthermore, both the 25% sucrose and 25% trehalose samples showed lower protein and moisture contents, but higher carbohydrate levels than the 0% sugar sample, reflecting the dilutive and hygroscopic effects of added sugars. Minor differences in nutritional composition between the sucrose and trehalose samples were also observed, likely resulting from their distinct molecular structures: sucrose is a glucose–fructose disaccharide, whereas trehalose is a non-reducing glucose–glucose disaccharide linked by an α,α-1,1-glycosidic bond (Li et al., 2024).

Table 1.

Nutritional composition of the LPSSCs with different sugar/ALPPSB mass ratios.

Item	Control	Sugar 0%	Suc 25%	Tre 25%
Energy (kJ/100 g)	972	1017	1142	1003
Protein (g/100 g)	0.2	0.68	0.57	0.55
Fat (g/100 g)	0	0	0	0
Water (g/100 g)	42.5	39.8	32.4	33.6
Ash (g/100 g)	0.32	0.37	0.42	0.45
Carbohydrate (g/100 g)	56.98	59.15	66.61	65.4

Note: The control sample was an LPSSC prepared with 0% sugar, 0% yeast, and 0 min fermentation. Protein, moisture, and ash contents were determined by an accredited third-party laboratory using methods compliant with both the China Metrology Accreditation (CMA) and the China National Accreditation Service for Conformity Assessment (CNAS).

All LPSSC formulations (with 0% sugar, 25% sucrose, and 25% trehalose) exhibited low protein contents, ranging from 0.55 to 0.68 g/100 g. These values are substantially lower than those reported for other developed low-protein foods, including low-protein pasta (2.94–5.76 g/100 g dry pasta)(Yaseen & Shouk, 2011), low-protein bread (1.40–1.43 g/100 g)(Scortegagna et al., 2020), and low-protein biscuits (1.27 g/100 g)(Azaripour & Abbasi, 2020). Consequently, the LPSSCs developed in this study achieved the targeted low-protein content of <1.0 g/100 g, positioning them as promising medical food candidates for the nutritional management of conditions requiring protein restriction, such as chronic kidney disease (Mafra et al., 2025), amino acid metabolism disorders (Handoom et al., 2018), and organic acidemias(Garcia-Arenas et al., 2023).

3.4. Effects of frozen times

Freezing is a widely used and effective method for food preservation (Hu, Zhang, Liu, Mujumdar, & Bai, 2022). This study evaluated the impact of frozen storage (0–120 days) on the textural properties of LPSSCs prepared with different formulations: varying fermentation times (Figs. 4A–D), different sucrose/ALPPSB mass ratios (Figs. 4E–H), and different trehalose/ALPPSB mass ratios (Figs. 4I–L). The results indicated that frozen storage for up to 120 days had no significant effect on the texture of LPSSCs prepared with different fermentation times (45, 90, and 120 min) (Figs. 4A–D). Similarly, texture remained stable in samples with sugar/ALPPSB ratios of 15% and 25%, regardless of sugar type (Figs. 4E–L). However, for LPSSCs containing a higher sugar ratio (35%), the effects of frozen storage were selective. While resilience and cohesiveness remained unchanged (Figs. 4F, J, H, L), chewiness and hardness showed significant variation over time (Figs. 4E, I, G, K). This pattern can be attributed to sugar recrystallization during frozen storage, a known cause of quality deterioration in sponge cakes (Díaz-Ramírez et al., 2016). In summary, our findings suggest that sugar/ALPPSB mass ratios ≤25% impart good frozen-storage stability to LPSSC texture, whereas ratios ≥35% can lead to undesirable changes in chewiness and hardness during prolonged frozen storage.

Fig. 4.

Effect of frozen storage duration on the textural properties of LPSSCs. (A–D) LPSSCs prepared with different fermentation times (45, 90, 120 min). (E–H) LPSSCs prepared with different sucrose/ALPPSB mass ratios (15%, 25%, 35%). (I–L) LPSSCs prepared with different trehalose/ALPPSB mass ratios (15%, 25%, 35%).

3.5. Effects of cooling times

Proper cooling is an important factor in food consumption (Schaffner et al., 2015). The influence of cooling time (0.25–48 h) on the textural properties of LPSSCs was investigated, as shown in Fig. 5. As cooling time increased, chewiness (Fig. 5A) and hardness (Fig. 5C) exhibited upward trends, while resilience (Fig. 5B) and cohesiveness (Fig. 5D) gradually decreased. These results confirm that cooling duration should be taken into account when consuming LPSSCs, which aligns with general practical understanding. Given that all four textural parameters (chewiness, hardness, resilience, and cohesiveness) underwent only minor changes within the first 2 h of cooling, it is recommended that LPSSCs be consumed within 2 h after steaming to maintain their optimal sensory quality.

Fig. 5.

Effect of the cooling time on the textural properties of LPSSCs. (A) chewiness, (B) resilience, (C) hardness, and (D) cohesiveness.

3.6. Descriptive sensory analysis

The influence of fermentation time on the sensory properties of LPSSCs is presented in Table 2. Fermented LPSSCs (45–120 min) exhibited significantly higher scores for all sensory attributes, such as chewing toughness, compared to the unfermented control (0 min). This aligns with findings that fermentation generally exerts positive effects on food quality, as demonstrated in products like tea (Cosme, Patarata, & Nunes, 2025). However, among the fermented samples, almost all sensory scores decreased progressively as fermentation time increased from 45 to 120 min, indicating that excessive fermentation can diminish product acceptability. This trend is consistent with reports that an intermediate fermentation duration is optimal for food quality, as seen in Yunnan congou black tea (Wang et al., 2022). Therefore, 45 min was determined to be the optimal fermentation time among the tested fermentation times (45–120 min) for preparing LPSSCs, yielding ideal sensory properties, including textural properties such as chewing toughness. It also implied that excessive fermentation was not ideal for the development of other special low-protein foods.

Table 2.

Sensory score results of Low-protein steamed sponge cakes with different fermentation times and sugar/ALPPSB mass ratios.

Samples in Figs. 2K and 3	Specific volume (15)	Finger press resilience (10)	Color (10)	Appearance (10)	Porous structure (20)	Chewing toughness (15)	Teeth stickiness (10)	Taste (10)	Sum
Fermentation times
0 min	10.72 ± 2.63c	5.25 ± 1.53c	6.00 ± 1.83d	5.42 ± 1.55c	11.50 ± 5.75c	9.33 ± 4.07b	6.67 ± 1.49d	7.25 ± 1.88a	62.13 ± 14.89c
45 min	14.08 ± 0.86ab	8.92 ± 0.95a	8.25 ± 0.83ab	8.00 ± 0.91ab	17.25 ± 1.64ab	12.17 ± 1.95ab	8.33 ± 1.03abc	8.25 ± 0.92a	85.25 ± 5.28a
90 min	13.33 ± 1.55ab	8.58 ± 0.95a	7.75 ± 1.23abc	8.50 ± 0.87a	17.08 ± 1.89ab	11.15 ± 3.03ab	7.75 ± 1.16bcd	8.5 ± 1.38a	82.65 ± 7.62ab
120 min	13.42 ± 1.38ab	8.17 ± 1.07ab	6.42 ± 1.38cd	7.08 ± 1.19ab	16.67 ± 1.80ab	11.50 ± 2.96ab	7.92 ± 1.26abc	8.75 ± 1.09a	79.92 ± 7.26ab
Sugar/ALPPSB mass ratios
Suc 15%	13.50 ± 1.12ab	8.58 ± 1.11a	7.75 ± 1.01abc	7.42 ± 1.44ab	16.25 ± 2.92ab	11.75 ± 2.49ab	8.0 ± 1.15abc	7.58 ± 1.11a	80.83 ± 7.39ab
Suc 25%	14.08 ± 0.86ab	8.92 ± 0.95a	8.25 ± 0.83ab	8.00 ± 0.91ab	17.25 ± 1.64ab	12.17 ± 1.95ab	8.33 ± 1.03abc	8.25 ± 0.92a	85.25 ± 5.28a
Suc 35%	12.58 ± 1.71b	7.17 ± 2.15b	7.42 ± 1.38bc	7.50 ± 1.89ab	13.83 ± 5.73bc	10.48 ± 3.50ab	7.33 ± 1.49cd	8.00 ± 1.47a	74.32 ± 15.56b
Tre 15%	14.25 ± 1.01a	9.08 ± 1.11a	8.33 ± 1.37ab	6.92 ± 1.55b	15.50 ± 3.10ab	11.67 ± 2.32ab	8.92 ± 0.95ab	7.92 ± 1.38a	82.58 ± 7.64ab
Tre 25%	14.08 ± 0.76ab	9.25 ± 0.83a	8.83 ± 1.28a	8.08 ± 1.26ab	17.50 ± 1.98a	12.75 ± 2.24a	9.00 ± 0.82a	8.17 ± 1.21a	87.67 ± 6.86a
Tre 35%	14.42 ± 0.76a	9.33 ± 1.03a	8.25 ± 1.30ab	7.75 ± 1.09ab	15.92 ± 2.84ab	12.50 ± 1.98ab	9.00 ± 0.91a	8.08 ± 1.19a	85.25 ± 6.91a

Note: Values are expressed by means ± standard deviation and the different letters in the same column indicate significant differences (n = 12, p < 0.05). The “45 min” sample was the same to the “Suc 25%” sample.

The influence of sugar/ALPPSB mass ratio on the sensory properties of LPSSCs is summarized in Table 2. Among the tested ratios (15–35%), a 25% sugar level yielded the highest overall sensory score. Trehalose-based samples achieved slightly higher total scores than their sucrose counterparts, suggesting that trehalose may offer a marginal advantage in enhancing the sensory quality of LPSSCs. The finding that moderate sugar addition (25%) is crucial for optimal sensory quality aligns with established principles in conventional food formulation. Notably, although trends were observed, differences across all eight individual sensory attributes were relatively small. These results indicate that both sucrose and trehalose are suitable sweeteners for developing special low-protein foods, with a moderate addition level around 25% representing an optimal range for balancing sweetness and texture.

3.7. VOCs of the LPSSCs with different fermentation times

VOCs in LPSSCs prepared with different fermentation times (Control, 0, 45, 90, and 120 min) were analyzed using HS-GC-IMS. Based on two-dimensional separation in the topographic plots (Fig. S1A) and topographic subtraction plots (Fig. S1B) (Zhang et al., 2020), a total of 61 VOCs were identified (Table S1). These comprised: 10 alcohols, 1 acid, 6 ketones, 12 aldehydes, 16 esters and lactones, 11 others, and 5 unidentified compounds.

The VOC fingerprint profiles of the samples were compared using the Gallery Plot plugin of the commercial HS-GC-IMS software VOCal, as shown in Fig. 6A. All samples with different fermentation times exhibited distinct fingerprint patterns, reflecting variations in the intensity of individual VOCs. Several characteristic VOCs, marked by red squares in Fig. 6A, were identified for specific fermentation treatments: three (1-pentanol (D), 2-furanmethanol acetate, and acetic acid propyl ester) for the control; two (area 58 and N-nitroso-N-ethylaniline) for the 0 min sample; five (isovaleric acid methyl ester, area 57, hexyl acetate, 2-methylbutanal, pyrrolidine) for the 90 min sample; and four (2-butanone-3-hydroxy, cyclopentanone, α-terpinolene, and β-myrcene) for the 120 min sample. Interestingly, no characteristic VOCs were detected in the 45 min sample, which coincided with this fermentation time yielding the highest overall sensory score (Table 2). This suggests that the fourteen characteristic VOCs identified in other samples (Fig. 6) may not be essential for achieving the optimal sensory quality of LPSSCs.

Fig. 6.

Analysis of volatile organic compounds (VOCs) in LPSSCs prepared with different fermentation times using headspace-gas chromatography–ion mobility spectrometry (HS-GC-IMS). The control sample was an LPSSC prepared with 0% sugar, 0% yeast, and 0 min fermentation. (A) Fingerprint comparison. Color intensity corresponds to VOC concentration (redder = higher). Each row displays the complete VOC profile of one sample; each column represents the same VOC across samples. Red squares highlight characteristic VOCs for each formulation. (B) Fingerprint similarity analysis (FSA). (C) Principal component analysis (PCA) score plot. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

N-nitroso compounds are recognized food-borne chemical carcinogens (Seyyedsalehi et al., 2023); therefore, the presence of N-nitroso-N-ethylaniline in the 0 min sample raises safety concerns and warrants further quantitative assessment to determine whether its level complies with relevant food safety standards. In contrast, the other characteristic VOCs identified have generally acceptable or desirable odor qualities. Isovaleric acid methyl ester is a characteristic aroma compound in coffee (Vezzulli, Lambri, & Bertuzzi, 2023). Hexyl acetate imparts a fruity note (Shieh & Chang, 2001). 2-Methylbutanal contributes a pleasant nutty aroma (Chen, Zhou, Yu, Yuan, & Tian, 2021). 2-Butanone-3-hydroxy is associated with buttery and cheesy odors (Yoon, Chang, & Lee, 2013). α-Terpinolene presents plastic-like and cucumber-like notes (Seo & Baek, 2005), while β-myrcene typically exhibits a woody aroma (Song, Li, Jiang, & Lin, 2024). Thus, aside from the safety-related N-nitroso-N-ethylaniline, the majority of the characteristic VOCs identified are compatible with—and may even contribute positively to—the flavor profile of LPSSCs prepared under different fermentation conditions.

Both fingerprint similarity analysis (FSA) and principal component analysis (PCA) were conducted to characterize the differences in volatile profiles among LPSSCs prepared with varying fermentation times (Fig. 6B and C). FSA, a dedicated plugin for GC-IMS data, evaluates VOC pattern similarity by calculating and comparing Euclidean distances (Sun, Wan, Han, Liu, & Wei, 2023). The FSA results (Fig. 6B) indicated distinct VOC profiles across all samples. PCA, a multivariate statistical tool widely used for dimensionality reduction of quantitative datasets (Chen, Tao, et al., 2021), was also applied. The first two principal components (PC1 and PC2) explained 51% and 23% of the total variance, respectively (Fig. 6C), and further confirmed significant separation among the samples based on their VOC composition. Together, both FSA and PCA demonstrate that fermentation time significantly influences the volatile compound profile of LPSSCs.

3.8. VOCs of the LPSSCs with different sugars

VOCs in LPSSCs formulated with different sugar types (Control, 0% sugar, 25% sucrose, and 25% trehalose) were analyzed by headspace-gas chromatography-ion mobility spectrometry (HS-GC-IMS). Based on the two-dimensional separations visualized in the topographic plots (Fig. S2A) and topographic subtraction plots (Fig. S2B) (Zhang et al., 2020), a total of 61 VOCs were identified (Table S2), categorized as follows: 10 alcohols, 1 acid, 6 ketones, 12 aldehydes, 16 esters and lactones, 11 miscellaneous compounds, and 5 unidentified compounds. Notably, the VOC composition identified here was identical to that observed in LPSSCs prepared with different fermentation times (Table S1), indicating that sugar type (sucrose vs. trehalose) did not alter the overall VOC profile under the experimental conditions tested.

The VOC fingerprint profiles of samples prepared with different sugars were compared using the Gallery Plot plugin of the commercial HS-GC-IMS software VOCal, as shown in Fig. 7A. Each sugar formulation exhibited a distinct fingerprint pattern, reflecting variations in the signal intensity of individual VOCs. Characteristic VOCs, marked by red squares in Fig. 7A, were identified for specific formulations: seven ((E)-2-octenal, (E)-2-heptenal, area 16, methylpentanoate, 1-pentanol (D), acetic acid propyl ester, and 2-furanmethanol acetate) in the control; three (allyl sulfide, 2-methyl-2-hepten-6-one, and (Z)-3-nonen-1-ol) in the 0% sugar sample; and three (area 60, 2-formyl-1-methylpyrrole, and butyl-2-propenoate) in the 25% sucrose sample. Notably, no characteristic VOCs were detected in the 25% trehalose sample, which coincided with this formulation achieving the highest overall sensory score among the sugar-containing samples (Table 2). This observation suggests that the ten characteristic VOCs identified in the control and 0% sugar samples may not be essential for optimal sensory quality in LPSSCs. In contrast, the three characteristic VOCs associated with the 25% sucrose sample could potentially contribute to the sensory profile of sucrose-sweetened LPSSCs, although their role appears less critical than the overall absence of distinguishing volatiles in the top-performing trehalose formulation.

Fig. 7.

Analysis of VOCs in LPSSCs prepared with different fermentation sugars using HS-GC-IMS. The control sample was an LPSSC prepared with 0% sugar, 0% yeast, and 0 min fermentation. (A) Fingerprint comparison. Color intensity corresponds to VOC concentration (redder = higher). Each row displays the complete VOC profile of one sample; each column represents the same VOC across samples. Red squares highlight characteristic VOCs for each formulation. (B) FSA. (C) PCA score plot. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Allyl sulfide is the predominant volatile in garlic oil (Wang et al., 2023). 2-Methyl-2-hepten-6-one imparts a citrus-fruity note (Huang et al., 2024), while (Z)-3-nonen-1-ol contributes a green, cucumber-like aroma (Zhang et al., 2024). 2-Formyl-1-methylpyrrole is associated with roasted flavor (Dippong et al., 2022), and butyl-2-propenoate adds pungent, apple-like, and sweet-fruity notes (Jia, Li, Yang, & Su, 2025). Therefore, the characteristic VOCs identified in sugar-containing LPSSCs generally correspond to acceptable or desirable aroma qualities, supporting their compatibility with—and potential positive contribution to—the flavor profile of these low-protein products.

FSA and PCA were employed to assess the differences in volatile profiles among LPSSCs formulated with different sugar types (Fig. 7B and 7C). FSA results (Fig. 7B) revealed distinct VOC fingerprints across all samples, indicating significant compositional variation. PCA further supported this observation. The first two principal components—PC1 and PC2—explained 61% and 14% of the total variance, respectively, and clearly separated the samples in the score plot (Fig. 7C). Collectively, both multivariate analyses confirm that sugar type significantly influences the volatile compound profile of LPSSCs.

3.9. In vitro digestion behaviors

The in vitro starch digestibility of LPSSCs prepared with different fermentation times and sugar/ALPPSB ratios was evaluated using a simulated intestinal model (Guo, Dai, et al., 2022). As shown in Fig. 8A, undigested cake pieces remained visually intact both before enzyme addition and after 3 h of incubation with α-amylase and glucoamylase. This observation indicates that enzymatic hydrolysis did not substantially alter the morphologies of the LPSSC matrix under the tested conditions.

Fig. 8.

In vitro starch digestibility of LPSSCs prepared with different fermentation times and sugar/ALPPSB mass ratios. The raw ALPPSB powder was included as a reference. (A) Macroscopic appearance of the digestion mixtures before and after 3 h of enzymatic hydrolysis. (B) Starch hydrolysis kinetics of LPSSCs prepared with different fermentation times (0, 45, 90, 120 min). (C) Starch hydrolysis kinetics of LPSSCs prepared with different sucrose/ALPPSB mass ratios (15%, 25%, 35%). (D) Starch hydrolysis kinetics of LPSSCs prepared with different trehalose/ALPPSB mass ratios (15%, 25%, 35%). The sample marked with a red dashed square in (B) and (C) is the same formulation. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

The starch digestibility kinetics of LPSSCs prepared with different fermentation times and sugar/ALPPSB ratios are presented in Figs. 8B–D. The raw ALPPSB powder exhibited the lowest starch hydrolysis across all time points, which is expected as native (ungelatinized) starch is inherently less digestible than its gelatinized counterpart formed during steaming (Qi & Tester, 2023). As shown in Fig. 8B, fermentation significantly enhanced starch digestibility. LPSSCs fermented for 45, 90, and 120 min achieved substantially higher hydrolysis percentages (28.6–37.9% at 180 min) compared to the unfermented control (20.1% at 180 min). Notably, the fermentation times of 45 and 90 min induced higher starch hydrolysis percentages (36.4%–37.9% vs. 20.1%–28.6% at 180 min) of LPSSCs than other fermentation times. Fermentation reduces the average molecular weight of starch, erodes starch granule surfaces, and increases the crystalline layers within granules (Zhao et al., 2019). While the reduced molecular weight may enhance starch hydrolysis, the increased crystalline layers could inhibit it. Consequently, LPSSCs fermented for 45 or 90 min exhibited higher starch hydrolysis (36.4%–37.9% at 180 min) compared to other fermentation times (20.1%–28.6%). Further research on the effect of fermentation on the molecular weight of starch should be explored using gel permeation chromatography (Gong et al., 2024; Xiao et al., 2025).

As shown in Fig. 8C, LPSSCs formulated with higher sucrose/ALPPSB ratios (25% and 35%) exhibited significantly greater starch hydrolysis (36.2–42.8% at 180 min) compared to those with a 15% sucrose ratio (20.5% at 180 min), indicating that elevated sucrose levels enhance in vitro starch digestibility. Similarly, Fig. 8D shows that LPSSCs containing higher trehalose/ALPPSB ratios (25% and 35%) achieved slightly higher hydrolysis percentages (27.3–30.7% at 180 min) than the 15% trehalose formulation (23.2% at 180 min). This suggests that increasing trehalose content also promotes starch digestion, albeit to a lesser extent than sucrose.

A comparison of Fig. 8C and D indicates that LPSSCs prepared with Suc/ALPPSB mass ratios of 25% and 35% exhibited higher starch hydrolysis than those with equivalent Tre/ALPPSB ratios. Sugar-starch interactions can stabilize starch granules and modulate their hydrolysis (Allan, Chamberlain, & Mauer, 2020). This hydrolysis difference may be attributed to the distinct molecular structures of sucrose (a glucose-fructose disaccharide) and trehalose (an α,α-1,1-linked glucose disaccharide) (Li et al., 2024). Sucrose is known to be more readily hydrolyzed than trehalose (Wolfenden & Yuan, 2008), which likely influences the overall starch hydrolysis dynamics. Therefore, the higher starch hydrolysis observed with high sucrose addition (25% and 35%) compared to trehalose may result from a combination of sucrose's inherent hydrolyzability and its specific interactions with starch.

4. Conclusions

In this study, low-protein steamed sponge cakes (LPSSCs) with protein contents below 1.0 g/100 g were successfully developed, and the impacts of fermentation time and sugar type/level on their properties were systematically investigated. The protein contents of the optimized formulations were 0.68 g/100 g (0% sugar), 0.57 g/100 g (25% sucrose), and 0.55 g/100 g (25% trehalose), all of which satisfy the low-protein criteria for dietary management of amino acid metabolism disorders and organic acidemias. Moreover, a 45-min fermentation combined with a 25% sugar/ALPPSB ratio yielded the highest overall sensory score (85.25/100) and a higher starch hydrolysis percentage (36.2%). Thus, the LPSSCs not only meet therapeutic protein-restriction requirements but also achieve sensory quality comparable to conventional foods, enhancing patient acceptability and dietary enjoyment. Both fermentation time and sugar addition significantly influenced processing behavior, frozen-storage textural stability, volatile compound profiles, and in vitro starch digestibility of the LPSSCs. These findings advance the understanding of composition-processing-property relationships in the design of palatable, low-protein foods for clinical nutrition.

To advance the fundamental science and practical application of LPSSCs, future research should be strategically directed along the following avenues. First, it is crucial to elucidate the mechanistic impact of key processing steps, particularly fermentation, on starch physicochemical properties. This includes investigating changes in molecular weight distribution, crystalline structure, and gelatinization behavior, which collectively govern the final product texture. Second, the specific interactions between starch and different sugars (e.g., sucrose vs. trehalose) within the low-protein matrix require in-depth study. Understanding these interactions at the molecular level is essential for rationally optimizing formula sweetness, texture, and shelf-life. Third, the functional role of hydrocolloids and other additives, such as sodium alginate and guar gum, merits systematic exploration to enhance structural integrity, moisture retention, and sensory acceptance of LPSSCs. Finally, establishing a comprehensive structure-property relationship necessitates a multi-level evaluation system that integrates advanced physicochemical characterization, dynamic sensory profiling, and in vitro/in vivo digestion models to predict and tailor product performance for specific dietary needs.

CRediT authorship contribution statement

Ye Zi: Writing – original draft, Methodology, Investigation, Data curation. Cuiping Shi: Investigation. Zhenfeng Liu: Investigation. Wei Cai: Supervision. Jian Zhong: Writing – review & editing, Supervision, Funding acquisition, Data curation, Conceptualization.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

Supplementary material

Table S1. The GC-IMS data of LPSSCs at different fermentation times.

Table S2. The GC-IMS data of LPSSCs at different sugar/LPP mass ratios.

Fig. S1. Fingerprints of volatile organic compounds (VOCs) of LPSSCs with different fermentation times using headspace-gas chromatography-ion mobility spectrometry (HS-GC-IMS).

Fig. S2. Fingerprints of VOCs of LPSSCs with different sugars using HS-GC-IMS.

Data availability

Data will be made available on request.