Response surface methodology (RSM) identifies the lowest amount of chicken plasma protein (CPP) in surimi-based products with optimum protein solubility, cohesiveness, and whiteness

ABSTRACT

This research aimed to determine the minimum amounts of chicken blood plasma (X₁) and sorbitol (X₂) that affect the protein solubility (Y₁), cohesiveness (Y₂), and whiteness (Y₃) of surimi. Response surface methodology (RSM) generated 13 experimental designs ranging from 0.5% to 2.5% for chicken plasma protein (CPP) and 2% to 6% for sorbitol. The results show that optimal conditions for protein solubility, cohesiveness, and a whiteness of surimi were achieved with a CPP concentration of 0.79% and a sorbitol concentration of 4.68%. This resulted in a predicted protein solubility of approximately 49.09 mg/ml, cohesiveness of 0.654 and whiteness value of 75.55. Meanwhile, the optimal experimental values for protein solubility, cohesiveness, and whiteness were found to be 46.56 mg/ml, 0.65, and 75.55, respectively. The measured values and predicted values have no statistically significant difference (p > .05), indicating the validity of the RSM model. These findings are valuable for improving surimi production.

KEYWORDS:

• Chicken plasma protein
• cohesiveness
• cryoprotectant
• gelling agent
• optimization
• protein solubility
• RSM
• sorbitol
• surimi
• whiteness

1. Introduction

Animal blood is a by-product of slaughterhouses that can be used as a source of protein. It is the second most commonly used protein-based meat by-product in food (Mullen et al., 2017). Proteins like fibrinogen, albumin, and globulins are all part of the plasma, which makes up around 60% of the blood (Jin et al., 2021; Zou et al., 2019). At present and excluding soil fertiliser, animal feed, food products, and pharmaceuticals are the three primary added-value applications of blood proteins of major added value (Toldrà et al., 2019). Blood proteins are used not only as a protein source but also as a binding agent in meat products, a protein enhancer in pasta, a fat substitute, as well as a polyphosphate or caseinate replacement (Dick et al., 2019; Fernández et al., 2021; Hurtado et al., 2012; Kumar et al., 2021). Previous research studying the adaptability of plasma protein revealed that porcine plasma may enhance texture, inhibit lipid oxidation, and promote antibacterial growth during storage of frankfurters (Jin et al., 2021; Verma et al.,2018, 2022). Chicken blood plasma has been extensively studied because of its numerous useful properties including gelling, water-holding capacity, emulsifying, fat-binding, and foaming abilities and the presence of protease inhibitors (Chen et al., 2015; Gao et al., 2020; Zou et al., 2019, 2021).

Surimi is a processed deboned fish meat product with high protein content, low fat content, and mild flavour used to formulate a variety of food products (Zhang et al., 2022). Myofibrillar proteins have the ability to form a gel, which is essential for the production of high-quality surimi-based products, however surimi’s gel-forming ability can be affected by storage time and temperatures (Cao et al., 2022). Myofibrillar proteins are degraded irreversibly by proteolytic digestion at elevated temperatures (40–70°C), which has a negative impact on the strength of surimi gels (Yingchutrakul et al., 2022). The preservation of surimi’s myofibrillar proteins during frozen storage is essential to produce high-quality surimi and gels with optimal physicochemical characteristics (Walayat et al., 2022). Serine and cysteine proteases are two protease types responsible for the gel-weakening process (Sutloet et al., 2018). Gel softening in modori, a type of surimi that passes through the temperature zone of 40–60°C to form a gel, has been reduced using a variety of food-grade protease inhibitors. Plasma protein (from poultry, pigs, and cows), egg white, whey, and other sources such as protein isolates and concentrates derived from plants, notably legume seeds, are used as additional protein to enhance the physicochemical properties of surimi (Bashir et al., 2017; Ofori & Hsieh, 2014).

The importance of cryoprotectant has been highlighted too. The myofibrillar proteins in surimi are likely to get denatured and aggregated if cryoprotectants are not applied before freezing (Walayat et al., 2022). Several different chemical substances (amino acids, sugars, sugar alcohols like sorbitol) are used as cryoprotectants. These cryoprotectants were selected for their efficiency, affordability, and minimal tendency to trigger a Maillard reaction (Jommark et al., 2018). The addition of cryoprotectants to surimi inhibits protein denaturation and stabilises the structural and functional characteristics of the protein by forming bonds with the protein’s functional groups. Sorbitol has been shown to enhance gel texture during superchilling storage, prevent protein and lipid oxidation, decrease protein structural changes as well as reduce protein loss of emulsifying characteristics (Liu et al., 2014; Sharma et al., 2019; Sousa et al., 2022; Zhou et al., 2019). This impact is due to the fact that they are able to raise the surface tension of water, which slows down the process of water being removed from proteins and protects the hydrogen bonds that exist between protein – protein complexes and water – protein structures (Liu et al., 2014; Walayat et al., 2022).

Response surface methodology (RSM) is a set of statistical and mathematical approaches based on the fit of a polynomial model to data that must portray the behaviour of a data set in order to make statistical predictions (Yolmeh & Jafari, 2017). The method is effective for optimising, designing, developing, and refining processes in which a response or responses are influenced by several variables. RSM is also used to analyse and optimise complex food operations (Gaikwad et al., 2021). It also interacts with variables and one or more responses. The technique is used for complicated variable processes, and its main advantage is that it requires fewer experimental runs to collect enough information to deliver a statistically acceptable solution (Shehata et al., 2021). RSM has been applied to optimise the number of surimi washing cycles and the starch amount added in order to get the desired moisture content and shear stress value (Fogaça et al., 2013). RSM was employed in the analysis of the effect of percentages of surimi, water, and fat on the physical, textural, and sensory characteristics of pork sausage (Murphy et al., 2004). The optimisations of Konjac glucomannan addition, heating temperature, and heating time to obtain the desired gel strength have also been reported (Liu et al., 2013). However, there is a lack of studies that optimise the concentration of CPP in the production of surimi. The results presented here indicate the minimum amount (percentage) of plasma protein that is suitable to be added during production of surimi. Although there has been successful production of surimi using egg white, tapioca starch, and potato starch, there are still cases of using blood plasma (chicken, bovine, or porcine plasma protein) as indicated in Vladimirovich et al. (2010), Vasilevna et al. (2018) and Antipova et al. (1999). The use of CPP is favourable due to certain properties which possessing the proteolytic degradation, hence enhancing stronger gel-strength of surimi. The existing academic literature predominantly focuses on investigating diverse cooking techniques, with specific emphasis on factors such as the washing cycle, setting time, cooking duration, and temperature. Although individual studies have extensively explored the applications of sorbitol and chicken plasma protein as separate ingredients, their combined optimization remains relatively unexplored. The lack of information regarding the optimization of these two ingredients together can be attributed to the prevailing trend of conducting separate studies on each component. Despite this limited research, there exists significant potential for synergistically combining sorbitol and chicken plasma protein to enhance the overall quality and functional attributes of food products. Despite the numerous advantages blood plasma proteins offer in the production of surimi, consumer concerns persist. Concerns include potential allergic reactions to blood proteins and the belief that blood sourced from animals may contain harmful microorganisms, toxins, and prions, declaring surimi products unfavourable for consumption. Moreover, certain religious groups such as Muslims and Jews must abstain from consuming any materials derived from blood, while vegetarians and vegans refrain from such products for ethical reasons. Consequently, this research aims to ascertain the potential presence of plasma contaminants in surimi, shedding light on this aspect of its safety and suitability for consumption. We hypothesized that the plasma powder can be added lower than 1%, to produce good physical properties of surimi, while for sorbitol, it is hypothesized that more than 4% is needed.

2. Materials and methods

2.1. Materials

The fish paste was obtained from a local company in Selangor, Malaysia, and kept at −20°C until used. Chicken blood was collected at a local slaughterhouse (Selangor, Malaysia) and processed as in Section 2.1.1.

2.1.1. Extraction of chicken plasma protein and preparation of chicken plasma powder

Blood was collected directly from slaughtered chickens into a vacuum blood collection tube containing 3.2% sodium citrate (Vacutube) with a 9:1 ratio of blood to anticoagulant and vigorous shaking to prevent blood clotting. During collection and transport, the blood was kept at a temperature of 4°C in ice box. The blood was then centrifuged for 20 min at 2,300 g (Kubota, Japan) to remove cells and cellular debris. The blood plasma supernatant was collected in a sterile glass bottle and frozen at −35°C for 24 h. The frozen plasma is then freeze-dried until a constant weight is attained using a freeze dryer (ALPHA 11–4 LD plus, Christ, Germany) at −47°C and 0.133 bar. The powdered chicken blood plasma was kept at 4°C until further usage.

2.1.2. Surimi gel preparation

The surimi was prepared in accordance with the method by Benjakul et al. (2003) and Zhu et al. (2022), with a slight modification, and the fish paste was thawed for 12 h at a chilled temperature (4°C) before being ground and mixed with the CPP, sorbitol, NaCl (2%, w/w), sucrose (4%, w/w), and sodium tripolyphosphate (0.3%). Crushed ice was used to regulate the moisture level to between 80% and 85% in order to produce a uniform slurry (Fowler & Park, 2015). The surimi paste was put into a 2 cm × 2 cm × 2 cm ice cube mould and properly wrapped before being frozen at −18°C for at least 12 h. Prior to examination, the surimi cubes were thawed for 24 h until its core temperature reached 4°C before boiled at 90°C for 20 min and cooled to room temperature prior to the experiment (Moosavi-Nasab et al., 2019).

2.2. Experimental design for response surface methodology (RSM)

The percentage of CPP (X₁, % w/w) and percentage of sorbitol (X₂, % w/w) were investigated as parameters that influence the response in the experimental design. The experimental results of two factors, five levels, and single block and their upper and lower levels on the RSM design were applied.

Minitab Software version 19 was used to build the factorial design using central composite design (CCD). CCD designed thirteen experimental trials based on two independent variables, the percentage of CPP and the percentage of sorbitol, at five levels (-α −1, 0, 1, and +α), to determine the optimal formulation of surimi (Table 1). Table 1 displays the values of the test variables that were used to generate the experimental design. Results of the dependent factors that were included in this analysis were protein solubility (Y₁), cohesiveness (Y₂), and whiteness (Y₃).

Table 1. The coded and uncoded values used in optimisation of surimi production.

		Factor level
Independent variable	Code	-α	−1	0	1	+α
Percentage of chicken plasma protein (CPP)	X1	0.5	1.0	1.5	2.0	2.5
Percentage of sorbitol	X2	2	3	4	5	6

The quadratic response surface based on the central composite model was applied to determine the response variables (Y₁, Y₂, and Y₃) of the surimi as the results of the independent variables (X₁ and X₂). The following is the generalized second-order polynomial model used in the analysis.

(1) $\mathrm{Y}={\beta }_0+{\beta }_1{\mathrm{x}}_1+{\beta }_2{\mathrm{x}}_2+{\beta }_{11}{{\mathrm{x}}_1}^2+{\beta }_{22}{{\mathrm{x}}_2}^2+{\beta }_{12}{\mathrm{x}}_1{x}_2+\epsilon$(1)

where Y expressed the predicted response variables, β₀ is the constant value of response variables at the central point of the model, β₁ and β₂ known as the linear coefficient and β₁₁ and β₂₂ are the quadratic coefficient for the main factor; meanwhile, β₁₂ is the second-order interaction coefficient (Ruiz-Hernández et al., 2021). The response variable was set at 0 and 1, respectively, which is low and high to observe overall desirability values.

2.3. Analysis of protein solubility

The method developed by Rawdkuen et al. (2004a) was adapted to conduct the analysis of protein solubility, with a few necessary adjustments. Prior to analysis, the surimi was thawed for 24 h until its core temperature reached 4°C before being boiled at 90°C for 20 min. After draining the surimi, it was minced in a food processor for 5 min. One gram of minced surimi was dissolved in 20 mM Tris-HCl, pH 8.0, containing 1% SDS, 2% β-mercaptoethanol, and 6 M urea. The mixture was homogenised for 1 min with a homogenizer (IKA, Germany), then boiled for 2 min and stirred for 4 h at room temperature (28–30°C) with a magnetic stirrer. The mixture was centrifuged at 10,000 g for 30 min (Eppendorf AG, Germany). Two milliliters of 50% cold trichloroacetic acid (TCA) was added to 10 ml of supernatant, and the mixture was incubated at 4°C for 14 h before being centrifuged at 10,000 g for 20 min. The precipitate was dissolved in 0.5 M NaOH after being washed with 10% TCA and solubilized in 0.5 M NaOH. The concentration of the protein was determined using Biuret Method by following the Sigma manufacturer protocol and expressed as mg/ml.

2.4. Textural properties analysis (cohesiveness)

The surimi was defrosted at a chilled temperature for 12 h before being heated for 20 min at 90°C and cooled before analysis. The cohesiveness of 2 cm × 2 cm × 2 cm surimi samples was determined using a compression plate (P/75) on the TA.XT. T2i (Stable Micro System, United Kingdom) texture analyser. The speed of the probe and its return were both set to 10 mm/sec. The contact force was set at 10 g, and the compression was calibrated to 50% strain (Rawdkuen et al., 2007). Triplication was performed for each one.

2.5. Colour analysis (whiteness)

The colour characteristics of the cooked surimi samples were measured using a Chromameter, CR400 (Konica Minolta, Japan) in an inadequately lit area. The samples of various formulations were sealed in clear plastic and L*, a*, and b* readings were taken with a Chroma metre. Sample findings were acquired using a chromaticity diagram or colour chart to determine the sample’s brightness and hue based on L*, a*, and b* values. Triplication measurements were performed for each one. Whiteness index was computed using the following formula (Duangmal & Taluengphol, 2010):

(2) $\mathrm{W}\mathrm{h}\mathrm{i}\mathrm{t}\mathrm{e}\mathrm{n}\mathrm{e}\mathrm{s}\mathrm{s}={\mathrm{L}}^{\ast }-3{\mathrm{b}}^{\ast }$(2)

2.6. Statistical analysis

Minitab Software (19th version) was used to construct the RSM with 13 combinations (Table 2) as the results of CCD using multiple regression with respect to the Equation. In this research, the effect of two independent variables which are percentage of CPP (X₁) and percentage of sorbitol (X₂) was analysed on the response variables, protein solubility (Y₁), cohesiveness (Y₂), and whiteness (Y₃), of the surimi.

Table 2. Factors and comparison between experiment and predicted responses.

	Independent variables		Response variables
			Y1, (mg/ml)		Y2		Y3
Run	X1 (%)	X2 (%)	Experimental	Predicted	Experimental	Predicted	Experimental	Predicted
1	1.00	3.00	32.72 ± 1.03	32.99	0.618 ± 0.02	0.616	75.64 ± 0.01	75.51
2	2.00	3.00	52.19 ± 0.01	49.30	0.697 ± 0.14	0.694	69.32 ± 0.01	69.24
3	1.00	5.00	41.81 ± 0.01	43.82	0.624 ± 0.01	0.624	72.60 ± 0.01	72.62
4	2.00	5.00	29.49 ± 0.01	28.34	0.562 ± 0.01	0.563	71.40 ± 0.01	71.47
5	0.80	4.00	45.08 ± 0.59	43.29	0.632 ± 0.02	0.633	73.87 ± 0.01	73.93
6	2.21	4.00	41.21 ± 1.33	43.88	0.644 ± 0.01	0.645	68.69 ± 0.01	68.69
7	1.50	2.59	35.55 ± 1.53	37.22	0.650 ± 0.03	0.653	73.20 ± 0.01	73.34
8	1.50	5.41	30.84 ± 0.01	30.05	0.568 ± 0.04	0.566	72.96 ± 1.08	72.88
9	1.50	4.00	34.24 ± 0.01	33.96	0.570 ± 0.09	0.572	72.86 ± 0.01	73.21
10	1.50	4.00	31.51 ± 0.29	33.96	0.574 ± 0.06	0.572	73.14 ± 1.73	73.21
11	1.50	4.00	32.92 ± 0.01	33.96	0.594 ± 0.02	0.572	73.46 ± 1.12	73.21
12	1.50	4.00	35.95 ± 0.07	33.96	0.560 ± 0.01	0.572	73.17 ± 1.14	73.21
13	1.50	4.00	35.19 ± 0.01	33.96	0.564 ± 0.01	0.572	73.43 ± 0.58	73.21

where Y₁, protein solubility, Y₂, cohesiveness, and Y₃, whiteness, and X₁, percentage of CPP, and X₂, percentage of sorbitol.

An analysis of variance (ANOVA) was carried out to assess the lack of fit and the effect of optimization on the response variables of the surimi for the validity and fitness of the model. Besides, the determination of the regression coefficient (R²) and the probability value (p-value) as statistical indicators of the polynomial of the response variables. The quadratic regression model is displayed in 2D contour and 3D response surface plots to describe the effect of the independent variable on the response variables. The verification values of the experimental responses were compared to the predicted values as stated in Table 2. All the data were analysed using the t-test method using SPSS 26th version. A significant difference was set at the level p < .05 as the p-value was analysed statistically and computed at a 95.0% confidence level.

3. Results and discussion

3.1. Surimi production using response surface model

RSM was used to identify optimum conditions for the production of surimi with added CPP and sorbitol. Using five levels of CCD, MINITAB software version 19 was used to analyse the effects of the independent variables, X1% of CPP and X2% of sorbitol, on the protein solubility, cohesiveness, and whiteness of the surimi thus obtained.

Table 2 lists the treatments for protein solubility, cohesiveness, and surimi whiteness along with the corresponding experimental variable level combinations and the results. The experimental and predicted response values for each run are displayed in Table 2. Under specific conditions of variable factors (percentage of CPP at 2.00% and percentage of sorbitol at 3.00%), the highest experimental and predicted protein solubility were 52.19 mg/ml and 49.30 mg/ml, respectively. However, the experimental protein solubility showed a slight difference from the predicted value. On the other hand, at the specified variable factor condition (percentage of CPP 2.00% and percentage of sorbitol 5.00%), the lowest experimental and predicted protein solubility of surimi likewise showed modest discrepancies, being 29.49 mg/ml and 28.34 mg/ml, respectively.

Considering various variable combinations, the experimental cohesiveness of the surimi ranged from 0.560 to 0.697. The predicted values varied from 0.563 to 0.694, but the range was broader. The variables’ greatest experimental and predicted cohesiveness was attained when the CPP percentage was 2.00% and sorbitol percentage was 3.00%. In contrast to the lowest expected cohesiveness at 2.00% CPP and 5.00% sorbitol, lowest experimental cohesiveness was obtained when the variables were set at 1.50% CPP and 4.00% sorbitol.

The highest experimental and anticipated whiteness of surimi was found when the variables were set to 1% CPP and 3% sorbitol (Run 1), yielding whiteness values of 75.640 and 75.508, respectively. In contrast, the lowest experimental (68.690) and predicted (68.686) whiteness values were recorded at the run 6 combination of 2.21% CPP and 4.00% sorbitol.

A regression analysis was conducted to fit mathematical models to experimental data in order to identify the optimal response area. The imperceptible p-value indicates that the regression is suited to the experimental value. By employing multiple regression analysis, the empirical relationship between the independent variables and the response variables can be described in terms of uncoded values by the following quadratic, second-order polynomial equations:

(3) ${\mathrm{Y}}_1=-11.2+6.3{\mathrm{X}}_1+22.61{\mathrm{X}}_2+19.25{{\mathrm{X}}_1}^2-0.163{{\mathrm{X}}_2}^2-15.89{\mathrm{X}}_1{\mathrm{X}}_2$(3)

(4) ${\mathrm{Y}}_2=0.8580-0.1089{\mathrm{X}}_1-0.0744{\mathrm{X}}_2+0.1327{{\mathrm{X}}_1}^2+0.01863{{\mathrm{X}}_2}^2-0.0702{\mathrm{X}}_1{\mathrm{X}}_2$(4)

(5) ${\mathrm{Y}}_3=85.41-2.54{\mathrm{X}}_1-3.594{\mathrm{X}}_2-3.804{{\mathrm{X}}_1}^2-0.0510{{\mathrm{X}}_2}^2+2.560{\mathrm{X}}_1{\mathrm{X}}_2$(5)

where Y₁, protein solubility, Y₂, cohesiveness, and Y₃, whiteness, and X₁, percentage of CPP and X₂, percentage of sorbitol.

3.2. Fitting the model and analysis of experiment design

The effect of independent variables (X₁ = percentage of CPP, X₂ = percentage of sorbitol) has been studied by performing model CCD of RSM on the three response variables (Y₁ = protein solubility; Y₂ = cohesiveness; Y₃ = whiteness). The significance of the coefficient was indicated by both the F and p-values. The chance that the factors have no influence or an effect that is very small or inconsequential on the response is denoted by the p-value > .05. In addition, the results of each model’s F-value and p-value exhibit the significance of the model, and it demonstrates that there is a strong agreement between the experimental values and the predicted values (Song et al., 2021). According to the findings, the strong significance of the regression model is shown by a high F-value in conjunction with a low p-value (Shehata et al., 2021; Wangtueai & Noomhorm, 2009). In order for the model to be considered statistically significant, the p-value should be lower than 0.05 (Patel et al., 2011).

According to Wangtueai and Noomhorm (2009), the value of R² suggested the variability of the model in the observed data; thus, the analysis of variance showed that the predicted second-order models were statistically suitable. In addition, Mohamad Said and Mohamed Amin (2016) revealed that the efficacy of the regression model is defined by the closeness of R² values to 1. Moreover, high R² values (p < .05), a lack of fit that is not statistically significant (p < .05),and adjusted R² values that are close to R² suggest that the models can adequately explain the relationship between independent and dependent variables (Sharma & Sogi, 2022). Table 3 shows the ANOVA for optimisation of protein solubility, cohesiveness, and whiteness of surimi.

Table 3. Analysis of variance (ANOVA) for the fitted quadratic polynomial model for the response variable.

Source	Degree of freedom	Adjusted sum of square	Adjusted mean square	F-value	P-value	Status
Protein solubility (R^2 = 92.13; Adjusted R^2 = 86.51%)
Regression	5	469.878	93.976	16.39	.001	Significant
Linear	2	51.715	25.858	4.51	.055	Not significant
X1	1	0.352	0.352	0.06	.812	Not significant
X2	1	51.364	51.364	8.96	.020	Significant
Square (quadratic)	2	165.512	82.756	14.43	.003	Significant
X1* X1	1	161.089	161.089	28.09	.001	Significant
X2* X2	1	0.185	0.185	0.03	.863	Not significant
Interaction	1	252.651	252.651	44.05	.000	Significant
X1* X2	1	252.651	252.651	44.05	.000	Significant
Residual Error	7	40.147	5.735
Lack-of-Fit	3	27.511	9.170	2.90	.165	Not significant
Pure Error	4	12.635	3.159
Total	12	510.025
Cohesiveness (R^2 = 96.77; Adjusted R^2 = 94.46)
Regression	5	0.021716	0.004343	41.95	.000	Significant
Linear	2	0.007685	0.003843	37.12	.000	Significant
X1	1	0.000141	0.000141	1.36	.282	Not significant
X2	1	0.007544	0.007544	72.87	.000	Significant
Square (quadratic)	2	0.009103	0.004551	43.96	.000	Significant
X1* X1	1	0.007656	0.007656	73.95	.000	Significant
X2* X2	1	0.002413	0.002413	23.31	.002	Significant
Interaction	1	0.004928	0.004928	47.60	.000	Significant
X1* X2	1	0.004928	0.004928	47.60	.000	Significant
Residual Error	7	0.000725	0.000104
Lack-of-Fit	3	0.000026	0.000009	0.05	.984	Not significant
Pure Error	4	0.000699	0.000175
Total	12	0.022441
Whiteness (R^2 = 99.27; Adjusted R^2 = 98.74)
Regression	5	40.6430	8.1286	189.70	.000	Significant
Linear	2	27.7601	13.8801	323.93	.000	Significant
X1	1	27.5491	27.5491	642.93	.000	Significant
X2	1	0.2111	0.2111	4.93	.062	Not significant
Square (quadratic)	2	6.3292	3.1646	73.85	.000	Significant
X1* X1	1	6.2915	6.2915	146.83	.000	Significant
X2* X2	1	0.0181	0.0181	0.42	.537	Not significant
Interaction	1	6.5536	6.5536	152.95	.000	Significant
X1* X2	1	6.5536	6.5536	152.95	.000	Significant
Residual Error	7	0.2999	0.0428
Lack-of-Fit	3	0.0601	0.0200	0.33	.803	Not significant
Pure Error	4	0.2399	0.0600
Total	12	40.9429

The regression for protein solubility was determined to be significant, with a p-value of .001. The linear coefficient of X₁, the percentage of CPP, was insignificant with a p-value of .812, however for X₂, the percentage of sorbitol, contributed considerably to the model with a p-value of .020. In addition, the quadratic coefficient of X₁, the percentage of CPP, was found to be significant with a p-value of .001, unlike X₂, the percentage of sorbitol, did not contribute to the model with a p-value of 0.863. While the interaction of X₁, percentage of CPP, and X₂, percentage of sorbitol adds considerably to the model with a p-value of .000. The adjusted R², 86.51, was somewhat less than R², 92.13, yet was very close, indicating that the model’s overall predictive performance was adequate. Hence, the model is relevant.

With a p-value of .000, the model for cohesiveness was determined to be very significant. The linear coefficient of X₁, the percentage of CPP, was found to be non-significant with a p-value of .282, while for X₂, the percentage of sorbitol, it contributed to the model with a p-value of .020. With p-values of .000 and .002, respectively, the quadratic coefficients X₁, percentage of CPP, and X₂, percentage of sorbitol significantly contribute to the model. The p-value for the interaction between X₁, the percentage of CPP, and X₂, the percentage of sorbitol, is 0.000, indicating that this interaction is significantly produced by the model. The adjusted R², 94.46, was a bit lower than R², 96.77, but was relatively close, indicating that the model’s overall predictive accuracy was good.

The model of whiteness has a 0.000 p-value and was found to be very significant. The linear coefficient of X₁, the percentage of CPP, was highly significant (p-value = .000); however, X₂, the percentage of sorbitol, was not significant (p-value = .062). The quadratic coefficient of X₁, the percentage of CPP, was determined to be highly significant with a p-value of .000, whilst X₂, the percentage of sorbitol, did not contribute to the model with a p-value of .537. The model is significantly impacted by the interaction of X₁, the percentage of CPP, and X₂, the percentage of sorbitol, with a p-value of .000. The overall predictive performance of the model was satisfactory as revealed by the adjusted R², 98.74, which was barely different from R², 99.27. As a result, the overall model is very significant.

Significantly, all response variables yielded a resultant p-value of .000 for the interaction between X₁ and X₂, indicating a highly significant relationship with the model. This implies that the combined effect of the independent variables (CPP and sorbitol) produces improved outcomes compared to their individual contributions. This observation may be attributed to the inclusion of CPP in the surimi, which contains a protease inhibitor that effectively maintains the protein from autolysis, a process that naturally occurs in surimi and can negatively affect its texture (Benjakul & Visessanguan, 2000). Additionally, sorbitol plays a vital role in stabilizing the surimi structure during cold temperature storage (Arpi et al. 2018). Notably, the optimization of the surimi formulation exhibits remarkable improvements when these two ingredients work together synergistically. In other words, CPP works together with sorbitol, which they cannot be alone in the formulation. This finding is further corroborated by the interaction of the independent variables, which also yielded a p-value of 0.000 for each response variable.

The lack-of-fit test analyzes the variance of the data in relation to the model. Valid data should be used to ensure that the proposed model is a good fit. The lack-of-fit test will be statistically significant if the model is a poor fit for the data. Insignificant lack-of-fit is recommended, according to Sharma and Sogi (2022), because significant lack-of-fit would imply that the model failed to adequately reflect data in the experimental domain at points that were excluded from the regression. The results of this investigation suggested that the model accurately predicted the protein solubility, cohesiveness, and whiteness, as evidenced by an error analysis that revealed a non-significant lack-of-fit (p > .05) with a p-value of .165, .984 and .803, respectively. Therefore, the overall model is significant.

3.3. Optimum condition of response surface analysis

In this finding, the predicted optimal conditions of surimi for the target with X₁ = 0.79% and X₂ = 4.68% were feasible to be carried out (Table 4). Meanwhile, the optimum conditions for the maximum and minimum goals were not feasible to be carried out. The data displayed using contour and surface response of the model allow for a better understanding of the interpretation (Noor Eliza et al., 2021). The substantial link between the addition of the CPP and sorbitol to the concentration of protein solubility, cohesiveness, and whiteness of the surimi under ideal conditions was revealed in Figures 1–3.

Figure 1. Contour plot and surface plot for the effect of percentage of CPP and percentage of sorbitol on the concentration of protein solubility of surimi.

Figure 2. Contour plot and surface plot for the effect of percentage of CPP and percentage of sorbitol on the cohesiveness of surimi.

Figure 3. Contour plot and surface plot for the effect of percentage of CPP and percentage of sorbitol on the whiteness of surimi.

Table 4. Comparison values of target and predicted responses for different optimum conditions and experiment feasibilities.

					Optimum condition		Predicted response, Y
Goal		Lower	Target	Upper	X1	X2	FITS 1	FITS 2	FITS 3	F/NF
Target	Protein Solubility	29.49	52.18	52.19	0.7929	4.6750	49.09	0.6541	75.58	F
	FITS 1	28.34	49.29	49.30
	Cohesiveness	0.560	0.696	0.697
	FITS 2	0.563	0.693	0.694
	Whiteness	68.69	75.63	75.64
	FITS 3	68.68	75.50	75.51
Maximum	Protein Solubility	29.49	52.19	52.19	0.7929	5.0428	52.19	0.6728	71.82	NF
	FITS 1	28.34	49.30	49.30
	Cohesiveness	0.560	0.697	0.697
	FITS 2	0.563	0.694	0.694
	Whiteness	68.69	75.64	75.64
	FITS 3	68.68	75.51	75.51
Minimum	Protein Solubility	29.49	29.49	52.19	2.2071	5.1857	27.32	0.5756	70.57	NF
	FITS 1	28.34	28.34	49.30
	Cohesiveness	0.560	0.560	0.697
	FITS 2	0.563	0.563	0.694
	Whiteness	68.69	68.69	75.64
	FITS 3	68.68	68.68	75.51

Where X₁ = percentage of CPP (%), X₂ = percentage of sorbitol (%), FITS = predicted response, F = feasible, NF = not feasible.

Protein solubility is the measurement of the quantity of protein that can be dissolved in a solution under specific circumstances, which changes based on numerous criteria. The SDS, urea, and β- mercaptoethanol solution was employed to solubilize proteins by cleaving all bonds save the non-disulfide covalent bond, namely the Ɛ-(ỵ-glutamyl) lysine bond. Induced by endogenous TGase and plasma TGase, the increase in solubility implies the creation of a non-disulphide covalent link (Rawdkuen et al., 2004b). Generally, Figure 1 shows an increase in both factors will reduce the concentrations of protein solubility. According to Table 3, quadratic regression for sorbitol does not contribute much to the model; however, the interaction between CPP and sorbitol does contribute significantly to the model. Rawdkuen et al. (2004b) concurred that the addition of CPP to bigeye surimi decreased the protein’s solubility, hence altering the gel’s textural qualities. According to Benjakul et al. (2005), the addition of 0.5–2.5% porcine plasma protein decreased the protein solubility of the surimi, hence enhancing its breaking strength. The low solubility of protein shows that proteinase is inhibited in the surimi gels; hence, it acquires gel-enhancing abilities (Ding et al., 2022). The improvement of gel strength of surimi was obtained with the reduction of sulfhydryl concentration, protein solubility, and the decline of Ca^2±ATPase activity affected by the addition of gill protein hydrolysate (Lin et al., 2019; Tian et al., 2022). In addition, the effect of polyol cryoprotectant is not only observed in sorbitol, but the decrease in protein solubility effect was observed in surimi added with maltodextrin (Kulkarni et al., 2019).

The cohesiveness was empirically assessed as the ratio of the area under the curve of the second bite to that of the first bite in texture profile analysis (TPA) (Nishinari et al., 2019). Figure 2 illustrates that the addition of modest amounts of both variables has a bigger effect on the cohesiveness of surimi; nevertheless, the combination of a higher quantity of CPP and a low amount of sorbitol offers surimi an outstanding cohesion. A similar observation was made by Fowler and Park (2015), who noted that the formation of disulphide linkages in fish myosin (myosin-actomyosin) increased the gel’s strength; meanwhile, intramolecular disulphide bonds (amino acids within the protein) were converted to intermolecular bonds (protein–protein interaction) to produce stronger formation. The cohesiveness of the gels normally reduces at low quantities of bovine plasma protein, but progressively increases as the plasma concentration added (Jarmoluk & Pietrasik, 2003). Sorbitol can raise the surface tension of water Walayat et al. (2022), subsequently delaying the removal of water from proteins and preserving the hydrogen bonds between protein – protein and water – protein. Tian et al. (2022) concurred that surimi enhanced the textural and microstructural qualities of common carp by reducing protein denaturation and aggregation. In addition, the addition of an 8% combination of sucrose and sorbitol reduces oxidative alterations such as protein solubility, Ca^2±ATPase, and sulfhydryl concentrations. The addition of sucrose and sorbitol increased the functional properties of carp surimi gel, such as water-holding capacity and textural profile analysis, with minimal structural modifications (secondary and tertiary) (Walayat et al., 2022).

Surimi’s whiteness is a key quality indicator of surimi (Petcharat et al., 2021; Ramadhan et al., 2012). The colour parameter, particularly the whiteness of surimi gel, is one of the most essential qualities when evaluating the quality of surimi. Figure 3 indicates that the combination of CPP and sorbitol significantly reduces the whiteness of the surimi with increments of both variables. A study conducted by Fowler and Park (2015) found that the addition of salmon plasma protein significantly dulled the whiteness of the surimi. Pinkish-hued blood proteins that were degraded during processing led to a reduction in lightness (L* value). White colouring is preferred for surimi-based goods, and the inclusion of plasma protein influences the colour characteristics, especially the lightness and whiteness index of the final product (Escalante-Rodríguez et al., 2018; Fowler & Park, 2015; Rawdkuen et al., 2004a). Even after defibrination and diafiltration process, significant residual haemoglobin remained in the serum to be converted into nitrosyl-haemochrome (Parés et al., 2011; Ramadhan et al., 2012). Plasma, the liquid portion of blood, can be any shade from yellow to orange to pink. Hemoglobin is released from red blood cells, giving the plasma a darker hue. Osmotic shock following blood dilution with water, bacterial activity, and/or mechanical stress can all induce red blood cell membrane disruption (hemolysis), which releases haemoglobin and results in inferior quality plasma during blood collection (Lynch et al., 2017; Toldrà et al., 2019). According to Ramadhan et al. (2012), the whiteness of surimi was determined to be between 63.13 and 65.96. On the other hand, the possibility of Maillard browning to occur in samples treated with cryoprotectants is due to the presence of sugar, which interacted with proteins, causing browning and a reduction in whiteness. However, Roze et al. (2021) stated Maillard and caramelization reactions did not occur, resulting in a lighter colour of baked goods as sorbitol was used as a sugar substitute. Meanwhile, in the commercialization context, surimi is often preferred in a white colour because surimi paste is a secondary product after fish paste, and the whiteness is important for next step which is shaping or colouring to create aesthetic and eye-catching products. Thus, it is important to control the amount of CPP and sorbitol in formulations to achieve the desired physicochemical properties of surimi.

In this research, the surface plots demonstrated significant variations in protein solubility, cohesiveness, and whiteness of surimi as the initial levels of X₁ and X₂ variables were changed. The plots indicated that generally, an increase in the levels of both variables led to a decrease in protein solubility, cohesiveness, and whiteness. This observation highlights the statistical significance of the quadratic term. Moreover, the linear, quadratic, and interaction terms all exhibited significant effects on the protein solubility, cohesiveness, and whiteness of the surimi.

3.4. Optimisation of surimi production and validation of models

The validity of the response variables (Y₁, Y₂, and Y₃) was established by employing the predicted optimal conditions (X₁: 0.79% and X₂: 4.68%). The experimental findings of Y₁, Y₂, and Y₃ are presented in Table 5 with the following values: 46.56, 0.651, and 75.55. Comparatively, the model predicted results were 49.09, 0.654, and 75.58, respectively. The fact that there was not a significant difference (p > .05) between the experimental and predicted values for any of the response variables is shown in Table 5, it confirms that the regression model that was created is accurate. Hence, it shows that the models were competent to predict the response variables, which validated the validity and appropriateness of the final reduced model that was fitted by RSM. Furthermore, it was discovered that the models were adequate for forecasting the response variables.

Table 5. The difference values of experimental and predicted values of response variables under the optimal conditions.

Responses	Experimental value	Predicted value	p-value
Protein solubility (mg/ml)	46.56 ± 1.081a	49.09a	.63
Cohesiveness	.651 ± 0.045a	0.654a	.96
Whiteness	75.55 ± 0.397a	75.58a	.88

Values with same superscripts within row are not significantly different (p > .05).

Overall, 0.79% of CPP and 4.68% of sorbitol were found to produce surimi with acceptable protein solubility, cohesiveness, and whiteness.

4. Conclusions

Optimisation of protein solubility, cohesiveness, and whiteness were obtained as surimi was incorporated with 0.79% of CPP and 4.68% of sorbitol. It was observed that the minimum amount of protein derived from chicken plasma protein and sorbitol showed a good synergy due to their good physicochemical properties. Therefore, this suggests that the model is statistically significant and can reliably predict the optimization of surimi formulation within a 95% confidence interval. On the other hand, the quality of the surimi can be improved further by purifying the plasma protein. Ultimately, this research leads to more refined and superior surimi product, catering to the discerning tastes of consumers and contributing significantly to the ever-evolving landscape of food science and technology.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This research was supported by the Fundamental Research Grant Scheme of the Ministry of Higher Education [600-IRMI/FRGS 5/3 (032/2019)].