Optimization and validation for quantification for allulose of jelly candies using response surface methodology

Abstract

A simple, rapid and reliable extraction method for allulose content in jelly were optimized using response surface methodology. The extraction method was selected based on preliminary experiments, with a three-factor, three-level central complex design including 20 experimental runs to optimize the extraction parameters. The optimum extraction factors predicted were temperature of 66 °C, solvent of 74% (v/v) ethanol, and extraction time of 24 min under shaking water bath extraction. The measured parameters were in accordance with the predicted values. The developed analytical method was validated with regard to linearity, accuracy and precision presenting recovery level from 90.79 to 95.18% and detection limits varying from 0.53 to 1.62 mg/mL. Finally, the method will be potentially applicable to a commercial jelly food using optimum extraction.

Electronic supplementary material

The online version of this article (10.1007/s13197-020-04773-z) contains supplementary material, which is available to authorized users.

Introduction

Rare sugars are becoming potential raw materials for food industry (Wu and Birch 2005). d-allulose/psicose (D-ribo-2-hexulose), C-3 epimer of D-fructose, is a rare monosaccharides and defined by the International Society of Rare Sugars and the United States Department of Agriculture (USDA) as ‘generally recognized as safe’ (GRAS) for use as a food ingredient. Recently, a rare sugar has obtained great attention because of its increasing consumption as a noncaloric sweetener or as a raw material for the production of rare sugars (Chattopadhyay et al. 2014). In particular, d-allulose has been widely recognized as a nearly zero energy sweetener in nature. (Matsuo et al. 2002). d-allulose has 70% sweetness of sucrose and therefore is referred to as a rare sugar. d-allulose is found in wheat, Itea plants, steam treated coffee, processed cane and beet molasses. d-allulose has been demonstrated to provide several benefits such as anti-inflammatory effects, reactive oxygen species scavenging activity, and neuroprotective effects (Moller and Berger 2003; Zhang et al. 2016; Ploypetchara and Gohtani 2018). In addition, d-allulose improves the texture of food materials through gelling properties and provides a pleasant flavor through the Maillard reaction of proteins during food processing (Oshima et al. 2014; Zeng et al. 2013).

Typically, jelly candies are mainly manufactured by the coagulation of water-soluble copolymers, such as gelatin, pectin, and agar with fruit flavor (Burey et al. 2009; Figueroa and Genovese 2019). Although jelly candies have high content of gelatin, its shape changes easily depending on gelatin content. Furthermore, gelatin is a natural colloid with properties of gelling. But extraction and analysis of sugar in jelly candies are difficult because of gelatin. For example, Ouchemoukh et al. (2010) used water to extract sugar easily from Algerian honeys. McKee (1985) also extracted sugars of reducing and non‐reducing sugars from carrot and other storage root vegetables. On the other hand, Černá et al. (2003) employed 72% ethanol solution to extract fructose, glucose, sucrose, and maltose in jelly. Their method was based on the principle that pectin, one of the constituent ingredients of jelly, is precipitated when alcohol is added (May, 1990). Similarly, 80% ethanol was used as an extraction solvent to analyze sugar (sucrose, glucose, and fructose) in fruits rich in dietary fiber (Koesukwiwat et al. 2008).

d-allulose has been analyzed by several methods such as capillary electrophoresis, gas chromatography, and high-performance liquid chromatography (HPLC) but they have disadvantage such as complicated sample preparation and lengthy runtimes. (Haas et al. 2018; Patel et al. 2018; Surapureddi et al. 2019). Accordingly, simple and accurate method is required to determine of d-allulose. Also, studies using commercial as a jelly candies for the extraction and analysis of d-allulose have not yet been reported. Therefore, optimized extraction and analysis methods are needed that reflect the characteristics of d-allulose-containing jelly candies.

The aim of this study is to optimize the extraction conditions, and validate the method using a HPLC–refractive index detector (RID), which is suitable for the determination of allulose in jelly candies. The optimal extraction conditions were verified using a central complex design (CCD) of the response surface methodology (RSM). Satisfactory results were also obtained from the application experiments in jelly candies.

Materials and methods

Chemicals and reagents

d-allulose was supplied by Samyang Corporation (Seongnam, Republic of Korea). Ethanol (EtOH, analytical grade) and de-ionized water (analytical grade) were obtained from Merck (Darmstadt, Germany).

HPLC analysis

The d-allulose analysis was performed in accordance with the procedure reported by Patel et al. (2018). HPLC analysis was performed using a SHIMADZU LC solution (Kyoto, Japan), with a quaternary pump (LC-20AD), thermostated column oven (CTO-10A), 10 µL loop injector, and a SHIMADZU Chemstation for data analysis. Detection was performed using a SHIMADZU RID (RID-10A). A separation column was a Aminex HPX-87C (300 mm × 7.8 mm i.d., Bio-Rad, Hercules, CA, USA). The column temperature was held at 85 °C. The mobile phase for isocratic elution was de-ionized water. The flow rate was 0.6 mL/min.

Preliminary tests for extraction conditions

Jelly candies was purchased from a local market. The jelly candies were frozen for 2 h, and ground with an electric grinder (FM-909 T; Hanil Co., Seoul, Korea) and stored at − 18 °C in airtight polypropylene plastic containers for analysis. The initial step in the preliminary test was to select an appropriate extraction solvent for d-allulose in the jelly candies. A series of extraction solvents was tested, from 0% to 90% ethanol, based on the sugar extraction method provided by the Ministry of Food and Drug Safety in Korea. In detail, 5 g of jelly (blank sample) was weighed into a 500 mL centrifugation tube and dissolved in 25 mL of the extraction solvent. The sample was spiked with d-allulose (50% concentration) to determine the recovery rate (%). Subsequently, the jelly was extracted with water at 85 °C for 25 min. The extraction sample solution was cooled to room temperature and filled to the mark with the extraction solvent. The extraction sample solution was then centrifuged at 10,000 rpm for 10 min at room temperature using a high-speed refrigerated centrifuge (CR22N; Hitachi Koki Co., Ltd., Tokyo, Japan). Finally, the obtained supernatant was filtered through a 0.45 µm membrane filter prior to HPLC-RID analysis, and the best extraction solvent composition was chosen. The second step of the preliminary test was to determine the extraction temperature. The extraction temperature was varied from 40 to 80 °C for the chosen extraction solvent and a constant time course (25 min). Based on the results, three levels (low, intermediate, and high) were established for each process variable for the RSM.

Experimental design

To optimize the extraction of d-allulose from jelly in aqueous ethanol, RSM was carried out with the CCD. The operational factors (independent variables) in the design were the extraction temperature (X₁, °C), solvent concentration (X₂, %, v/v, water/ethanol), and time (X₃, min), while the dependent variable was the recovery rate (%). In the CCD, all process variables were studied at three levels (− 1, 0, and 1); each level is a code for the value of the original variable. Coding the variable levels involves the simple linear transformation of the original measurement scale; hence, values of − 1 and 1 are applied to determine the minimum and maximum values. The studied factors were the extraction temperature (40, 60, and 80 °C), solvent concentration (50, 70, and 90%) and time (10, 25, and 40 min). Supplementary Table 1 shows 20 different experimental runs of the CCD and the corresponding response data. After performing 20 different experiments, a quadratic model was fitted to the response data using Minitab 16 version (Minitab Inc., State college, PA, USA).

Data analysis

RSM analysis showed that the response variables Y (recovery, %) for the independent variables X_i and X_j are second-order polynomials, and k is the number of tested variables (k = 3) (Eq. (1)). The regression coefficients were β₀ (intercept), β_i (linear), β_ii (quadratic), and β_ij (cross-product terms).

$$Y={\beta }_0+\sum _{i=1}^k{\beta }_i{X}_i+\sum _{i=1}^k{\beta }_{\mathit{ii}}{X}_i^2+\sum _{i=1}^{k-1}\sum _{j>1}^k{\beta }_{\mathit{ij}}{X}_i{X}_j$$ 1

Analysis of variance (ANOVA) was performed to evaluate the individual linear, two-way, and interaction regression coefficients using Minitab 16. The coefficient of determination (R²) was used to estimate the fitness of the polynomial equation to the responses, and the significance of the dependent variables was statistically analyzed by computing the F value at statistical significance (p < 0.05). Furthermore, a quadratic model was developed, which can be expressed in three-dimensional plots. The CCD data include the contour and 3D response surface plots, which visualize the results of the experiment and allow the researcher to visually examine the relationships between the variables in the plot and the results. Statistical analyses were performed using SPSS v.20 software (IBM‐SPSS, Chicago, IL, USA), and the Duncan test was used to determine the statistical significance (p < 0.05 deemed as significant).

Method validation

The HPLC-RID method for d-allulose analysis in jelly candies was validated in terms of linearity, limits of detection (LOD) and quantification (LOQ), accuracy, and precision according to the International Conference on Harmonization harmonized tripartite guideline (Guideline 2005). Three calibration curves were plotted using five different concentrations (6.25, 12.5, 25, 50, and 100 mg/mL), which were prepared from a stock solution (500 mg/mL) in de-ionized water, to evaluate the linearity of the analytical method. The LOD and LOQ values were calculated to be 3.3 and 10 times, respectively, the standard deviation of the response divided by the slope of the calibration curve; these were used to evaluate the sensitivity of the method. The accuracy was estimated in jelly spiked with d-allulose at four different concentrations (10, 20, 40, and 80 mg/mL). The precision was measured by inter- and intra-day variation tests. The precision value was expressed as the relative standard deviation (RSD, %).

Results and discussion

Preliminary optimization experiments

Jelly confections are composed of high amounts of sugars together with gelatin (Pocan et al. 2019). Therefore, the affected by gelatin is important factor in the extraction and analysis for sugar in jelly candies. To choose the solvent, the preliminary experiments were conducted for determining the best extract solvent that would allow HPLC separation with high sensitivity and identification of d-allulose in jelly candies. The concentration of the extraction solvent was varied to determine the key factors for optimization to achieve the best extraction of d-allulose. From the obtained results, among the different concentrations, 70% ethanol was identified to be optimum compared with the other tested solvent concentrations (0, 30, 50, and 90%). As ethanol has a lower boiling point and can be easily vaporized during extraction, 70% ethanol was employed for further experiments. Among the tested temperatures, 60 °C yielded a significantly higher recovery rate of allulose (p < 0.05). On the other hand, an extraction temperature of 40 °C yielded relatively low recovery rates (86.98 and 86.82%) compared to other treatment temperatures (60 and 80 °C). Therefore, we chose 60 °C as the optimum extraction temperature.

Fitting the model

Three factor, extraction temperature (X₁: 40–80 °C), solvent concentration (X₂: 50–90%), and time (X₃: 10–40 h), was performed to optimize the interaction effects of the independent variables on the extraction of d-allulose. These factors were selected during the preliminary study for the highest extract of d-allulose. The RSM-CCD model has been similarly used to optimize the extraction of cinnamic acid and cinnamaldehyde in a selected design space (Lee et al. 2018). The results of 20 runs using CCD are shown in supplementary Table 1, which include the response surface design and the observed response. After obtaining the results from the CCD, a second-order polynomial equation reflecting the empirical relationship between the response variable and the independent variables was established as follows (Eq. (2)).

$$Y=55.65+0.660X_1+0.297X_2-0.023X_3-0.00394X_1^2-0.00147X_2^2-0.00195X_3^2-0.001825X_1{X}_2-0.00014X_1{X}_3+0.00172X_2{X}_3$$ 2

The results for ANOVA, the goodness-of-fit, and the adequacy of the regression model are summarized in Table 1. The model F-value was 6.69, and the model p-value was lower than 0.01, implying that the model was highly statistically significant. The F-value and p-value for the lack of fit were 1.30 and 0.389, respectively, which confirmed the good fit and suitability of the regression model. From Fig. 1a, a determination coefficient (R²) of 0.8576 was obtained by curve-fitting the predicted and experimentally obtained recovery rates of allulose. The adjusted determination coefficient (R_Adj² = 0.7295) further confirmed the minute difference between the experimental and predicted values. The standard deviation (S = 0.8895) indicates the degree of precision to carry out the experimentation. Table 1 shows that the linear coefficient (temperature, X₁), quadratic coefficient (temperature*temperature, X₁²), and cross product coefficient (temperature*solvent concentration, X₁X₂) are significant (p < 0.05), while the other terms’ coefficients are not significant (p > 0.05). The residual graph of the model in Fig. 1b indicates that the chosen model is suitable, because the prediction and scatter plot of the standardized residuals are spread randomly within a horizontal band of ± 1.5.

Table 1.

Analysis of variance of experimental data

Source		DF	Sum of square		Mean square		F-value		p-value
Model		9	47.6624		5.2958		6.69		0.003
Linear		3	14.8586		4.9529		6.26		0.012
X1		1	12.2922		12.2922		15.54		0.003
X2		1	2.3814		2.3814		3.01		0.113
X3		1	0.1850		0.1850		0.23		0.639
Square		3	26.4044		8.8015		11.12		0.002
X12		1	6.8300		6.8300		8.63		0.015
X22		1	0.9555		0.9555		1.21		0.298
X32		1	0.5311		0.5311		0.67		0.432
2-way interaction		3	6.3995		2.1332		2.70		0.102
X1X2		1	4.2632		4.2632		5.39		0.043
X1X3		1	0.0145		0.0145		0.02		0.895
X2X3		1	2.1218		2.1218		2.68		0.133
Error		10	7.9120		0.7912
Lack of fit		5	4.4756		0.8951		1.30		0.389
Pure error		5	3.4363		0.6873
Total		19	55.5743
S = 0.8895
R2 = 85.76%
R2(adj) = 72.95%

Fig. 1.

Plot of experimental values versus predicted values graph of the quadratic model (a); residual graph of the quadratic model (b); both obtained with the data of the central composition design assays

Optimization of d-allulose extraction conditions by RSM

Considering the validity of the predicted model, optimization of the conditions for d-allulose extraction was performed. Figure 2 depicts the influence of temperature, time and solvent concentration on d-allulose extraction. It can be seen that the recovery rate of d-allulose first increased with increasing temperature in the studied variable ranges from 40 to 66 °C, but decreased from 66 to 80 °C in Fig. 2a and b. This is in agreement with the results reported Giannoccaro et al. (2006), who compared the content of sugar extracted from soybean at different temperatures (25, 50, and 80 °C). The highest content of sugar was 12.14 − 12.54%, obtained at 50 °C. Increase in the temperature from 50 to 80 °C decreased the content of sugar. Time had no significant effect on d-allulose extraction in the selected range, whereas solvent concentration affects d-allulose extraction as shown in Fig. 2c and d. The response surface plot for d-allulose extraction versus the extraction solvent concentration and time is shown in Figs. 2e and f, where an extraction time of 24 min and 74% ethanol were effective conditions. This result is supported by previous studies (Černá et al. 2003; Koesukwiwat et al. 2008). In the study of Černá et al. (2003), 72% ethanol was effectively extracted sugar from jelly. In addition, Koesukwiwat et al. (2008) was extracted to analyze sugars in fruits rich in dietary fiber with 80% ethanol. The pectin is washed with alcohol to remove impurities. In other words, alcohol can effectively extract sugars from jelly components such as pectin (Sayah et al. 2016). Thus, the optimum d-allulose extraction from jelly is achieved with an extraction time of 24 min in a water bath maintained at 66 °C using 74% ethanol as the extraction solvent.

Fig. 2.

Response surface plots. (a, b) Interaction plot of temperature and concentration. (c, d) Interaction plot of temperature and time. (e, f) Interaction plot of concentration and time

Application of optimized d-allulose extraction conditions

The optimized sample preparation method was applied to jelly candies (supplementary Table 2). The experimental value (90.07) for the responses under this condition nearly agreed with the predicted value (89.10). The low coefficient of variation value (0.54%) represents a high accuracy and consistency of experiments performed (Liyana-Pathirana and Shahidi 2005). Therefore, the optimized model is well-fitted for the extraction of d-allulose from jelly candies.

Validation of the method

The method was validated for linearity, LOD and LOQ, accuracy, and precision. The linearity study was performed in triplicate with standard solutions corresponding to each point on the calibration curves. The calibration curve showed good linear regression (R = 0.999) at five concentrations ranging from 6.25 to 100 mg/mL using HPLC-RID. A similar linearity of d-allulose was found in the analysis of sugar using capillary electrophoresis (R = 0.999) (Surapureddi et al. 2019). The LOD and LOQ values were determined to be 0.53 and 1.62 mg/mL, respectively (Table 2) and were validated with satisfactory recovery and precision parameters of d-allulose in jelly. Table 3 indicates that the recovery values are in the range of 90.79–95.18%, indicating good accuracy of the optimized extraction conditions with the HPLC-RID analysis method. The precision expressed as % RSD was less than 1.98% and 2.07% for the intra- and inter-day analyses, respectively.

Table 2.

Regression equation, limits of detection and quantification for d-allulose

Sample	Linearity range (mg/mL)	Slope	Intercept	Correlation (R)	LOD (mg/mL)	LOQ (mg/mL)
Jelly	6.25–100	131,079.7	-39,910.4	0.999	0.53	1.62

Table 3.

Accuracy and precision for d-allulose in jelly

Sample	Spiked concentration (mg/mL)	Accuracy			Precision (RSD, %)
		Detected concentration (mg/mL)	Recovery (%)		Intra-day (n = 3)	Inter-day (n = 9)
Jelly	80	72.82	91.03 ± 0.25		0.40	0.28
	40	36.32	90.79 ± 1.87		0.82	2.07
	20	18.66	93.31 ± 0.60		0.55	0.64
	10	9.52	95.18 ± 1.53		1.98	1.61

Conclusions

The present study was designed to obtain optimal experimental conditions for maximizing the extraction efficiency of d-allulose from jelly. The extraction conditions were optimized with the CCD model of the RSM. The optimum extraction conditions of d-allulose were a temperature of 66 °C, solvent concentration of 74% (ethanol), and extraction time of 24 min. The experimental values for the responses were confirmed to closely agree (CV 0.54%) with the predicted values. Further, it was verified that the extraction of d-allulose from jelly was highly dependent on the temperature (X₁), and solvent concentration (X₂).

Using the optimized d-allulose extraction conditions, HPLC-RID analysis revealed good linearity with a correlation coefficient of 0.999. The estimated LOD and LOQ values were in the range of 0.53 and 1.62 mg/mL, respectively. Moreover, when the conditions were applied to spiked jelly samples, recovery rates of 90.79 − 95.18% were obtained, with under 5% of RSD. Consequently, it was concluded that the optimized extraction conditions and HPLC-RID method are suitable for d-allulose. The experimental approach in this study could serve as a reference for optimizing different food groups for d-allulose determination.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Compliance with ethical standards

Conflict of interest

The authors declare no conflict of interest.