A validated fast difference spectrophotometric method for 5-hydroxymethyl-2-furfural (HMF) determination in corn syrups

Food Chemistry 228 (2017) 197–203

Contents lists available at ScienceDirect

Food Chemistry journal homepage: www.elsevier.com/locate/foodchem

Analytical Methods

A validated fast difference spectrophotometric method for 5-hydroxymethyl-2-furfural (HMF) determination in corn syrups Jucimara Kulek de Andrade a,⇑, Camila Kulek de Andrade a, Emy Komatsu b, Hélène Perreault b, Yohandra Reyes Torres a, Marcos Roberto da Rosa a, Maria Lurdes Felsner a a b

Departamento de Química, Universidade Estadual do Centro-Oeste – UNICENTRO, 85040-080 Guarapuava, PR, Brazil Chemistry Department, University of Manitoba, 144 Dysart Road, R3T 2N2 Winnipeg, MB, Canada

a r t i c l e

i n f o

Article history: Received 19 April 2016 Received in revised form 28 July 2016 Accepted 31 January 2017 Available online 2 February 2017 Keywords: HMF Corn syrups Difference spectrophotometry In house validation Nested design

a b s t r a c t Corn syrups, important ingredients used in food and beverage industries, often contain high levels of 5-hydroxymethyl-2-furfural (HMF), a toxic contaminant. In this work, an in house validation of a difference spectrophotometric method for HMF analysis in corn syrups was developed using sophisticated statistical tools by the first time. The methodology showed excellent analytical performance with good selectivity, linearity (R2 = 99.9%, r > 0.99), accuracy and low limits (LOD = 0.10 mg L
1 and LOQ = 0.34 mg L
1). An excellent precision was confirmed by repeatability (RSD (%) = 0.30) and intermediate precision (RSD (%) = 0.36) estimates and by Horrat value (0.07). A detailed study of method precision using a nested design demonstrated that variation sources such as instruments, operators and time did not interfere in the variability of results within laboratory and consequently in its intermediate precision. The developed method is environmentally friendly, fast, cheap and easy to implement resulting in an attractive alternative for corn syrups quality control in industries and official laboratories. Ó 2017 Elsevier Ltd. All rights reserved.

1. Introduction Corn syrup, also known as glucose syrup, is defined as the purified, concentrated, aqueous solution of nutritive saccharides obtained from edible starch with a dextrose equivalent of 20 or more (Caballero, Trugo, & Finglas, 2003). Due to its physicochemical properties, they are widely used in the food and beverage industries (Belitz, Grosch, & Schieberle, 2009; Caballero et al., 2003). Corn syrups can be obtained by corn starch hydrolysis in acidic medium under high temperature and pressure, by enzymatic process or still by a combination of both processes (Almandoz, Pagliero, Ochoa, & Marchese, 2010; Caballero et al., 2003; Newton, 2007). Factors such as thermal processing, long periods of storage and the chemical composition of corn syrups favour the formation of high levels of toxic contaminants such as 5-hydroxy-2-methylfurfural (HMF) (Belitz et al., 2009; De Andrade et al., 2016; Kavousi, Mirhosseini, Ghazali, & Ariffin, 2015; Spano et al., 2009). Recently, a great interest in HMF quantification and distribution in different foods and beverages has been observed due to its ⇑ Corresponding author. E-mail address: [email protected] (J.K. de Andrade). http://dx.doi.org/10.1016/j.foodchem.2017.01.158 0308-8146/Ó 2017 Elsevier Ltd. All rights reserved.

possible toxicological effects (carcinogenicity, mutagenicity, cytotoxicity and genotoxicity) (Capuano & Fogliano, 2011; Kowalski, Lukasiewicz, Duda-Chodak, & Ziec´, 2013; Truzzi et al., 2012). Taking this into account it is important to monitor HMF concentration in foodstuffs, especially in corn syrups, which can have high levels of this contaminant. Different analytical methods have been developed for HMF analysis in foods and beverages. Liquid chromatography with UV detection (HPLC-UV) remains the most widely used method (Chávez-Servín, de la Carbot, García-Gasca, Castellote, & LópezSabater, 2015; De Andrade et al., 2016; Er Demirhan et al., 2015; Lee, Sakai, Manaf, Rodhi, & Saad, 2014; Spano et al., 2009; Teixidó, Núñez, Santos, & Galceran, 2011). In addition, UV–Vis spectrophotometric methods are considered also official methods for HMF determination in different matrices (Gürkan & Altunay, 2015; Truzzi et al., 2012; Zappalà, Fallico, Arena, & Verzera, 2005; Zhang, Wei, Liu, Lin, & Yuan, 2014). On the other hand, a growing interest in green methodologies that minimize or eliminate the production of by-products, the use of organic solvents and toxic reagents, which are hazardous to human health or to the environment (Armenta, Garrigues, & De La Guardia, 2008), has made that simple techniques such as UV–VIS Spectrophotometry has called the attention of researchers. UV–VIS Spectrophotometry can be considered a green methodology,

198

J.K. de Andrade et al. / Food Chemistry 228 (2017) 197–203

because it is fast, makes use of inexpensive instrumentation, does not require specialized training and does not use a large amount of toxic solvents as chromatographic methods do (Armenta et al., 2008; Gürkan & Altunay, 2015; Truzzi et al., 2012; Zhang et al., 2014). For this reason, spectrophotometric methods are still widely used in research, regulatory bodies and industry laboratories for HMF determination. White (1979) developed a difference spectrophotometric method for HMF determination in honey. This method is based on reacting HMF with sulphite ion (HSO
3 ) (Teixidó et al., 2011; Truzzi et al., 2012; White, 1979). It uses low cost reagents and does not generate wastes, thus the difference spectrophotometric method is a green alternative to other analytical methods, a fact that is very important for laboratories that aim to get ISO standards (Armenta et al., 2008). Nevertheless, in order to apply White’s spectrophotometric method to other foods, such as corn syrups, the analytical methodology must be appraised for accuracy, and precision in HMF determination. It is known that factors such as operator, instrument, time and random errors might affect the analytical precision, and contribute to the variability of measurements carried out within a laboratory (ISO 5725-6, 1994; Kuttatharmmakul, Massart, & Smeyers-Verbeke, 1999). Additionally, the selectivity and sensitivity of the spectrophotometric method should be assessed when the analysis is performed directly in the presence of complex matrixes (Spano et al., 2009). Thus to guarantee the suitability of spectrophotometric methods to accomplish the desired purpose, rigorous validation studies must be undertaken. In the last few years, numerous analytical methods have been developed for HMF analysis in food and beverages (ChávezServín et al., 2015; Er Demirhan et al., 2015; Lee et al., 2014). However, reports of rigorous validation studies about the appropriateness of these methodologies and using sophisticated statistical tools, as those recommended by ISO, IUPAC and AOAC guidelines are still scarce (AOAC, 2012; Araujo, 2009; De Andrade et al., 2016; Kuttatharmmakul et al., 1999). Considering the aforementioned facts, this work aimed to develop a fast, cheap, simple, green and direct methodology for HMF determination in corn syrup samples by difference spectrophotometry. To verify the analytical performance of the developed method a rigorous and sophisticated validation process based on AOAC, ISO and IUPAC guidelines was carried out. Additionally, a detailed precision study using nested design was carried out to evaluate the effect of factors such as different operators, instruments and time on intermediate precision. To our knowledge, no previous efforts have been made to optimize and validate the White’s method adapted for HMF analysis in corn syrup samples. 2. Materials and methods 2.1. Reagents and samples HMF (99%) and formic acid (95%) were purchased from SigmaAldrich (Brazil). Sodium hydrogen sulphite (97%) was purchased from Biotec (Brazil) and acetonitrile (HPLC grade) from EMD (Canada). Ultrapure water (0.055 lS cm
1) was obtained by a TKA-GenPure system from Thermo Scientific (Brazil) and a Simplicity UV Milli-Q system from Millipore (Canada). Commercial corn syrup samples were acquired in 2013 in supermarkets from Guarapuava (Brazil) (CS1–CS5) and from Winnipeg (Canada) (CS6) cities. 2.2. Standard and sample solutions Stock standard solutions of HMF (1000 mg L
1) and sodium hydrogen sulphite (0.40% (w/v)) were daily prepared in ultrapure

water. In order to apply the developed difference spectrophotometric method to HMF quantitation in corn syrup samples, aqueous solutions at 5.0 mg mL
1 and at 160.0 mg mL
1 were made for Brazilian and Canada corn syrup samples, respectively. For chromatographic analysis by HPLC-UV, aqueous solutions of corn syrup samples were prepared as described by De Andrade et al. (2016). 2.3. High-performance liquid chromatography (HPLC) HPLC analysis of corn syrups samples was carried out following the methodology described by De Andrade et al. (2016). 2.4. Difference spectrophotometry UV Spectrophotometric determinations were carried out in 1 cm quartz cells in a single beam spectrophotometer Cary 50 Bio from Varian. For the precision study and to record spectra, a double beam spectrophotometer SP-2000 UV from Spectrum Meter was utilized. Spectra were obtained with a bandwidth of 2 nm and a scan rate of 1 nm s
1. Absorbance values were read at 285 nm. 2.5. Development of the difference spectrophotometric method Firstly, to adapt White’s difference spectrophotometric method (1979) to HMF analysis in corn syrups the optimal concentration of sodium hydrogen sulphite reagent was determined. This concentration was defined as the concentration that produces the maximum reduction in HMF (4.0 mg L
1) absorption band at 285 nm with minimal interference. Concentrations of sodium hydrogen sulphite were varied from 0.05 to 0.40% (m/v). Two solutions were prepared for each concentration of sulphite reagent evaluated. The first solution (S1) was prepared by mixing 5.0 mL of HMF standard solution (4.0 mg L
1) and 5.0 mL of water. The second solution (S2) was obtained by adding 5.0 mL of HMF standard solution (4.0 mg L
1) and 5.0 mL of the sodium hydrogen sulphite solution under evaluation (varied concentration from 0.05 to 0.40% (m/v)). Afterwards, the absorbance was read at 285 nm. Ultrapure water was used as blank. When sulphite is added to HMF solution a reaction occurred and a reduction of HMF absorption band at 285 nm is observed. The percentage of reduction was determined by Eq. (1).

%Reduction ¼ ððAS1
AS2 Þ  100Þ=AS1

ð1Þ

AS1: absorbance of solution 1 (S1) at 285 nm (sample solution); AS2: absorbance of solution 2 (S2) at 285 nm (sample solution with addition of sulphite reagent). The optimal concentration of sodium hydrogen sulphite reagent was used in all HMF determinations in corn syrup samples by the difference spectrophotometric method described in 2.6. 2.6. HMF determination by difference spectrophotometry For HMF analysis in corn syrup samples two solutions were prepared. The first solution (S1) was prepared by addition of 5.0 mL of corn syrup sample solution (obtained as described in 2.2) and 5.0 mL of ultrapure water. The second solution (S2) was obtained by addition of 5.0 mL of corn syrup sample solution (prepared as described in 2.2) and 5.0 mL of sodium hydrogen sulphite solution at 0.1% (w/v). Both solutions were analyzed at 285 nm. The concentration of HMF was expressed as mg of HMF per kg of corn syrup and it was calculated using Eq. (2).
1

HMF ðmg kg Þ ¼ ð½HMF ðmg L
1 Þ  Vsample Þ=wsample

ð2Þ

J.K. de Andrade et al. / Food Chemistry 228 (2017) 197–203

[HMF (mg L
1)]: HMF concentration in (mg L
1) obtained by subtracting absorbance at 285 nm of solutions 1 and 2 (AS1
AS2) and by analytical calibration; Vsample: volume of corn syrup sample solution; wsample: corn syrup sample weight. All analyses were carried out within an hour immediately after mixing sodium hydrogen sulphite reagent at sample solution and HMF in order to avoid the sodium hydrogen sulphite solution degradation.

199

analyses were performed by each operator in different instruments and different days. An ANOVA characteristic of nested design at 95% confidence level was applied to verify the influence of the different factors (instruments, operators, days and replicates) on the variability of results within laboratory, i.e., in the intermediate precision. The variance components of each evaluated factor were also determined. ANOVA results allowed to obtain the estimates for repeatability and intermediate precision that were expressed in relative standard deviation (RSD%) and by Horrat values (AOAC, 2012).

2.7. Validation study The validation study was carried out using validation parameters such as linearity, detection limit (LOD), quantification limit (LOQ), accuracy and precision according to literature recommendations (AOAC, 2012; Araujo, 2009; ICH, 1996; Kuttatharmmakul et al., 1999). All statistical analyses were carried out using the software Minitab for Windows version 16.2.2 (Minitab, 2010). 2.7.1. Calibration and linearity Aqueous standard solutions of HMF at the concentration range of 2.0–10.0 mg L
1 were made from the HMF stock solution at 1000 mg L
1. Three aqueous standard solutions were prepared for each point of the analytical curve and the absorbance was read for each solution in triplicate. Linearity was checked by applying a linear regression analysis and a lack-of-fit test (Araujo, 2009; Barros Neto, Bruns, & Scarmínio, 2006; ISO 5725-6, 1994) to the experimental data at 95% confidence level. The significance of the coefficients in the regression equation (intercept and slope) was tested by applying a t-test. 2.7.2. Detection (LOD) and quantification (LOQ) limits Detection (LOD) and quantification (LOQ) limits were calculated based on ð3  SDÞ=m and ð10  SDÞ=m respectively, where m is the slope of the calibration curve and SD is the standard deviation on the intercept of the analytical curve (Araujo, 2009; ICH, 1996). 2.7.3. Precision Precision was evaluated by the repeatability and the intermediate precision estimates. The repeatability test was carried out with three corn syrup solutions prepared as described in 2.2, in the same day. The intermediate precision was evaluated at different levels (instruments, operators and time) and was estimated for each factor using a nested (hierarchical) design as recommended by ISO 5725-6 (1994) and by Kuttatharmmakul et al. (1999). The nested design considered four factors that could contribute to the variability of the results within the laboratory (operators, instruments, days and replicates) (Kuttatharmmakul et al., 1999). Analyses were carried out according to the nested design matrix (Fig. 1) where each of the m operators (m = 2) prepared n replicates (n = 3) of the corn syrup sample solution (CS1), obtained as described in 2.2, in p days (p = 5) and carried out the measurement of the absorbance values in q different spectrophotometers (q = 2). The

2.7.4. Accuracy Accuracy was evaluated by comparing HMF concentrations in corn syrups samples obtained by the developed difference spectrophotometric method and by a previously validated HPLC-UV method (De Andrade et al., 2016). A linear regression analysis at 95% confidence level was applied (AOAC, 2012; Araujo, 2009; Miller & Miller, 2005).

3. Results and discussion 3.1. Development of the difference spectrophotometric method To develop an accurate, precise and selective difference spectrophotometric method to determinate HMF in corn syrup samples, we first compared the UV absorption spectra of HMF standard solution (4.0 mg L
1) and corn syrup solution (CS1) (Fig. 2A). In both spectra, a broad and intense absorption band centered at 285 nm and another less intense band at 228 nm were observed. This UV absorption spectrum is characteristic of HMF as already reported (Taher & Cates, 1974). The absorption band at 285 nm is associated to the p–p⁄ transition, characteristic of the carbonyl group and the band at 228 nm is attributed to 3-acetylacrylic acid formed by oxidation of HMF in air (Taher & Cates, 1974). In this work, a rapid spectrophotometric method was developed using the difference spectrum generated by subtracting the spectrum obtained for corn syrup solution from those obtained after reaction with sodium hydrogen sulphite reagent (Fig. 2B). It is also possible to subtract the absorbance values at 285 nm (AS1
AS2) obtained for sample solution (S1) from that of sample solutions where sulphite reagent was added (S2). The reduction of HMF concentration by reaction with sodium hydrogen sulphite decreases the strong absorption band at 285 nm and increases more than three times the absorption band at 228 nm. The same behavior has been previously observed by White (1979) in honey clarified solutions, after addition of this reagent. The difference spectrum shows a maximum absorption at 285 nm and an isosbestic point at 240 nm (difference in absorbance is zero because the HMF and the product of reduction with sodium hydrogen sulphite reagent have equal absorptivity) (Fig. 2C). This change in spectral properties of HMF is due to the formation of a sulphite addition product that does not absorb at 285 nm but presents a strong absorption

Fig. 1. Nested design matrix applied for the intermediate precision study of the spectrophotometric method.

200

J.K. de Andrade et al. / Food Chemistry 228 (2017) 197–203

Fig. 2. (A) UV absorption spectra of (—) HMF standard and (- - -) corn syrup sample (CS1); (B) UV absorption spectra of corn syrup sample (CS1) before and after addition of sodium hydrogen sulphite reagent; (C) Difference spectrum generated by subtraction of corn syrup sample (CS1) spectrum from those obtained after reaction with sodium hydrogen sulphite; (D) Effect of sodium hydrogen sulphite concentration (%, w/v) on the HMF absorbance measured at 285 nm.

band around 200 nm (Fig. 2B) (Azevêdo, Reis, Da Silva, & De Andrade, 2007; Davidson & Dawodu, 1987; White, 1979). One drawback of using sodium hydrogen sulphite as reduction reagent is that it shows a weak absorption band at 284 nm (Fig. 2B) (Davidson & Dawodu, 1987; White, 1979). Therefore, it is necessary to find an acceptable ratio between efficiency and interference, evaluating the sulphite reagent concentration that results in the greatest reduction of the HMF absorption band at 285 nm with the lowest interference effect. Sodium hydrogen sulphite concentrations between 0.05% to 0.40% (w/v) were added to a HMF standard solution (4.0 mg L
1) (Fig. 2D) and the greatest reduction (91.6 and 92%) (Fig. 2D) in the absorption band at 285 nm was observed at concentrations equal to 0.10 and 0.15% (w/v), respectively. As pointed out by other researchers, the reaction between HMF and sodium hydrogen sulphite reagent is not complete (Davidson & Dawodu, 1987; White, 1979). A fraction of unreacted HMF is then still observed. Nevertheless, the reduction reagent chosen in this work has advantages over others employed in difference spectrophotometry, such as low cost and low toxicity (Davidson & Dawodu, 1987, 1988). Furthermore, in this work it was not necessary to add a clarifier agent (Carrez I and Carrez II solutions), as observed in the White’s method for HMF analysis in honey (White, 1979). A solution of sodium hydrogen sulphite at 0.10 (%, w/v) was chosen as the optimal condition, considering that the difference between the values of absorbance for this concentration and the concentration of 0.15 (%, w/v) was small and because the use of a smaller amount of reducing agent could decrease the effect of matrix interference. White (1979) also evaluated the effect of sulphite concentration on the quantification of HMF in honey and similarly, he chose 0.10 (%, w/v) as the best concentration for this reagent.

3.2. Validation study 3.2.1. Selectivity It is recognized that spectrophotometric methods are not always selective, especially because they show broad absorption bands and overlapping absorption of interferents and analytes may occur. To evaluate if the developed difference spectrophotometric method was selective, a corn syrup sample (CS1) and a HMF standard solution (3.0 mg L
1) were analyzed by a validated HPLC-UV method (De Andrade et al., 2016) with detection at 285 nm (Fig. 3). In both chromatograms, only one well defined peak at 4.6 min attributed to HMF was observed. These results suggest that the spectrophotometric method has good selectivity. In other words, other compounds absorbing at 285 nm, which could react with the sulphite reagent and interfere in the HMF analysis in corn syrup samples, were not observed. 3.2.2. Linearity Linearity is a validation parameter that allows determining the range in which the analytical signal is linearly proportional to the analyte concentration. This analysis is carried out by the calibration curve or analytical curve (de Ribeiro, Ferreira, Morano, Da Silva, & Schneider, 2008). To evaluate the linearity of the developed spectrophotometric method an analytical curve with aqueous HMF standard solutions in the range of 2.0–10.0 mg L
1 was built. The experimental calibration data (absorbance and concentration) were adjusted to a linear model. An ANOVA analysis characteristic of linear regression and a lack of fit test were performed at 95% confidence level in order to verify if the generated model was appropriate (Araujo, 2009; Barros Neto et al., 2006). The results demonstrated that the linear model is adequate to establish the relationship between the absorbance at 285 nm and

201

J.K. de Andrade et al. / Food Chemistry 228 (2017) 197–203

Fig. 3. HPLC-UV chromatograms at 285 nm of HMF standard and corn syrup sample (CS1).

the concentration of HMF, since the values for the lack of fit test were not significant at 95% confidence level (Flof = 0.04; p = 0.987). This fact is also highlighted by the very significant regression (Freg = 12732.7; p = 0.000). Additionally, the significance of the calibration coefficients (intercept and slope) was verified by applying a t-test. Results showed that only the slope was significant (tobs = 112.8 and p = 0.000 for slope and for intercept tobs = 0.12 and p = 0.908). Therefore, the calibration curve presents linearity in the range of concentrations studied, and can be represented by: absorbance = 0.00094 + 0.1359 HMF concentration (mg L
1). Furthermore, a good adjustment of the linear model to the experimental data is strengthened by the observed correlation coefficient (r = 0.9999), which is within the acceptable limits (r > 0.99) (Brasil., 2003), and by the determination coefficient (R2) equal to 99.9%, suggesting that only 0.1% of the variation in the data is explained by residuals. 3.2.3. Limits of detection (LOD) and quantification (LOQ) In order to verify the possibility of detecting and quantitating HMF at low concentrations in corn syrup samples, the LOD and LOQ values were calculated using the analytical curve method (ICH, 1996; Ribani, Bottoli, Collins, Jardim, & Melo, 2004). LOD

and LOQ values of 0.10 mg L
1 and 0.34 mg L
1, respectively, were obtained. These limits were lower than those observed for the determination of HMF in honey samples using the difference spectrophotometry and the sulphite reagent, and calculating as described in this work, using analytical curve method (LOD and LOQ values of 0.22 mg L
1 and 0.67 mg L
1, respectively) (Truzzi et al., 2012). Consequently, LOD and LOQ values obtained in our study were considered adequate. 3.2.4. Precision A sophisticated precision study was carried out at two levels (repeatability and intermediate precision) and following the adaptation of ISO 5725-6 (1994) for intralaboratorial validation studies suggested by Kuttatharmmakul et al. (1999). According to these researchers, the precision within a laboratory can be evaluated by considering different conditions of intermediate precision (equipment, operator and time). To evaluate the contribution of these variation sources, ISO 5725-6 (1994) recommended the use of a nested design, which was applied in this work (Table 1). The Fobserved values (Table 1) determined for different variation sources (instruments, operators and days) were lower than the Fcritical values suggesting that these sources do not contribute

Table 1 Results for precision study, ANOVA for nested design used to obtain the precision estimates and variance component (VC) estimates for the intralaboratorial precision study. Operator

Instrument 1

Instrument 2

Day

C1*

C2*

C3*

**

C123

1

1 2 3 4 5

1660.60 1663.52 1666.78 1648.43 1648.60

1661.72 1657.34 1659.54 1649.93 1658.21

1656.74 1660.64 1655.58 1657.74 1654.70

1659.69 1660.50 1660.63 1652.03 1653.84

2

1 2 3 4 5

1646.31 1651.92 1654.28 1656.74 1656.55

1658.69 1649.35 1659.87 1649.78 1648.50

1656.06 1654.25 1649.64 1647.44 1653.80

1653.69 1651.84 1654.60 1651.32 1652.95

Day

C1*

C2*

C3*

C123**

C mean***

1657.34

1 2 3 4 5

1660.66 1668.80 1659.94 1647.22 1664.36

1658.67 1659.87 1659.60 1650.66 1664.50

1653.71 1663.58 1662.51 1654.04 1660.99

1657.68 1664.08 1660.68 1650.64 1663.28

1659.27

1652.88

1 2 3 4 5

1648.70 1668.07 1657.48 1647.49 1650.80

1658.14 1658.81 1669.27 1656.76 1658.60

1658.61 1660.37 1648.52 1656.89 1660.13

1655.14 1662.42 1658.42 1653.71 1656.51

1657.24

C mean

***

Global 1656.68 mg kg
1

Concentration****

*

Variance Sources

Df

SS

MS

Estimate Amount

Instrument (I)

1

148.71

148.71

Operator (O)

2

180.11

90.54

nO nT nr s2I nT nr s2O þ

Day (T)

16

713.09

44.68

nr s2T þ s2r

Replicate (r)

40

963.58

24.09

s2r

Total

59

200.49

þ nT nr s2O nr s2T þ s2r

þ

nr s2T

þ

s2r

Fobserved

p-value

VCE*****

1.65a

0.327

2.02b

0.165

s2r s2T

s2

s

s2(%)

¼ SMr

1.95

1.39

5.45

r ¼ SMTn
SM r

3.03

1.74

1.85c

0.058

8.45

T s2O ¼ SMnOT
SM nr

6.83

2.61

19.01

I
SM O s2I ¼ SM nO nT nr

24.09

4.91

67.10

35.90

5.99

–

s2IPðOITÞ
1

¼

s2O

þ

s2I

þ

s2T

þ

s2r

C1 to C3: replicates of the HMF concentration (mg kg ) obtained in the corn syrup solution in each day. C123: average HMF concentration (mg kg
1) obtained by replicates (C1 to C3) in each day. *** C mean: average HMF concentration (mg kg
1) obtained from the averages of five days (C123 for each operator in each instrument). **** Global concentration: average HMF concentration (mg kg
1) obtained by different operators, instruments and days nr : number of replicates; nT : number of days; nO : number of operators; aF critical (0.05; 1, 2) = 18.5; bF critical (0.05; 2, 16) = 3.63; cF critical (0.05; 16, 40) = 1.90. ***** VCE: Variance components equation. **

202

J.K. de Andrade et al. / Food Chemistry 228 (2017) 197–203

imperfections and finite slit width effect, among others (Rouessac & Rouessac, 2007). Repeatability and intermediate precision estimates were obtained by RSD (%) values and the Horrat value was calculated to evaluate the suitability of intermediate precision (AOAC, 2012). For repeatability, a RSD (%) = 0.30 was obtained which it is much smaller than the maximum limit (RSD (%) < 3.7) recommended by the AOAC in the investigated concentration range (AOAC, 2012). Similarly, a small RSD (%) = 0.36 was calculated for intermediate precision considering different operators, instruments and days. These results are highlighted by a Horrat value equal to 0.07, which was lower than the recommended limit for precision of intralaboratorial analysis (<1.3) (González, Herrador, & Asuero, 2010). Similar difference spectrophotometric methods using sulphite reagent for HMF analysis in honey showed lower precision than the observed in this study with RSD values varying from 6.0 to 7.5% (Truzzi et al., 2012; White, 1979; Zappalà et al., 2005). Consequently, our results indicated that the developed difference spectrophotometric method presents an excellent precision. No similar reports were found about the evaluation of intermediate precision considering the effect of different instruments, operators and days on the determination of HMF in corn syrups. This fact shows the relevance of the work herein reported since these factors can cause variability in the results within laboratory. Besides this, the results obtained by nested design application could be considered as a truly intermediate precision study (González et al., 2010).

Table 2 Averages and standard deviations for HMF concentration (mg kg
1) in corn syrup samples determined by difference spectrophotometric and HPLC-UV methods. Samples*

CS1 CS2 CS3 CS4 CS5 CS6 CS7 *

HMF (mg kg
1) HPLC-UV

Difference Spectrophotometry

1655.9 ± 7.1 1086.9 ± 3.6 1627.9 ± 3.4 2121.3 ± 2.5 406.6 ± 2.3 1642.6 ± 5.1 50.4 ± 0.6

1654.6 ± 5.1 1087.4 ± 1.8 1623.7 ± 2.9 2114.4 ± 4.2 397.1 ± 4.2 1635.9 ± 6.2 46.6 ± 0.7

Corn syrup samples CS1 to CS6 were from Brazil and CS7 from Canada.

significantly to the dispersion of results within the laboratory. Thus, there are no significant differences among the HMF concentrations in corn syrup samples determined in different instruments, operators and days. In other words, the application of the developed difference spectrophotometric method in different instruments, in different days or by different operators does not affect its precision. Despite the fact that the factors investigated in the intermediate precision study do not show significant influence on the method’s precision, it is important to know their contribution to the variability of the results obtained from the intralaboratorial analyses. Thus, the variance components for each factor (s2O , s2I , s2T , s2r , s2IPðOITÞ ) and their estimates (s2, s, s2 (%)) extracted from the ANOVA characteristic of the nested design were computed (Table 1). The magnitude of the estimates of the variance components can indicate which of the evaluated sources (instrument, operator, time and replicates) need more attention when applying the difference spectrophotometric method and could contribute to greater variability in results and consequently decrease the precision of the analytical methodology. The highest estimates for the variance components studied (Table 1) were identified for replicates (67.10%) and days (19.01%), suggesting that these components make the greatest contribution to the variance in the results within laboratory and thus, present the highest influence in the method’s precision. Variability in replicates can be associated to errors in steps, such as weighing and dilution, during the sulphite reagent and sample solutions preparation. Moreover, the variability in results due to time (different days) can be attributed to instrumental deviations such as fluctuation in the lamp intensity, stray radiation, monochromator

3.2.5. Accuracy The accuracy of the method was evaluated by comparing HMF (mg kg
1) levels determined by the difference spectrophotometric method with the values obtained by HPLC-UV as reference method (Table 2) and applying a linear regression analysis at 95% confidence level. This approach allows identifying the presence of systematic and random errors, which can affect the accuracy of both methods, when the slope and intercept of linear model equation is significantly different from 1 and 0, respectively (Miller & Miller, 2005). Results obtained by difference spectrophotometry were slightly lower than those determined by HPLC-UV (Table 2). This can be attributed to the fact that the reaction between HMF and sulphite reagent is not complete, as described previously. Nevertheless, linear regression results (Fig. 4) demonstrated that there is a linear

S 6.86039 R-Sq 100.0% R-Sq(adj) 100.0%

Spectrophotometric method

2000

1500

1000

500

0

0

500

1000

1500

2000

HPLC-UV method Fig. 4. Comparison of spectrophotometric and HPLC-UV methods by linear regression at 95% confidence level.

J.K. de Andrade et al. / Food Chemistry 228 (2017) 197–203

functional relationship between the HMF concentrations (mg kg
1) determined by the two methods (Freg = 216,423; p = 0.000 and R2 = 100%; r = 1.000). A t-test was applied on the linear and angular coefficients of this new linear equation to verify if there were significant random and systematic errors in each of the investigated methods. Results showed that the slope is equal to 1.000 (tobs = 465.2; p = 0.000) and the intercept is equal to 0 (tobs = -1.53; p = 0.142). In other words, both methods (difference spectrophotometry and HPLCUV) are not affected significantly by systematic errors (Fig. 4). Truzzi et al. (2012) compared HMF concentrations in honey samples obtained using HPLC-UV methods and those obtained by White (1979). Similarly to our current study, these authors verified that the intercept of calibration curve was not significantly different from 0, the slope was not significantly different from 1, and the correlation coefficient (r = 0.9891) indicated a relatively strong relationship between the variables. Thus, it can be suggested that the developed difference spectrophotometric method presents good accuracy for HMF determination in corn syrup samples and the reference method (HPLC-UV) is not significantly affected by random and systematic errors. 4. Conclusions A simple, fast, cheap and green method for HMF analysis in corn syrups was developed. The method suitability was demonstrated through a rigorous validation process following recommendations of ISO, IUPAC and AOAC and by using sophisticated statistical tools. The developed difference spectrophotometric methodology showed excellent analytical performance with good selectivity, linearity and low limits of detection and quantification. Moreover, a detailed study of method precision demonstrated that variation sources such as instruments, operators and time did not interfere in the variability of results within laboratory and consequently in its intermediate precision. A good accuracy was also verified when HMF concentrations determined by difference spectrophotometry were compared with those obtained by a reference method (HPLC-UV) for different corn syrups samples. Features such as speed, low cost, ease of implementation and absence of costly and toxic solvents makes the spectrophotometric method an useful and attractive alternative for routine quality control of corn syrup samples by industry and official laboratories. Acknowledgments Authors would like to thank the National Counsel of Technological and Scientific Development (CNPq), Araucária Foundation, CAPES Foundation, the Canadian Foundation for Innovation (CFI), and the Natural Sciences and Engineering Research Council (NSERC) for financial support. J.K.A. thanks CAPES and Araucária Foundation for scholarships. References Almandoz, C., Pagliero, C., Ochoa, A., & Marchese, J. (2010). Corn syrup clarification by microfiltration with ceramic membranes. Journal of Membrane Science, 363, 87–95. Araujo, P. (2009). Key aspects of analytical method validation and evaluation. Journal of Chromatography B, 877, 2224–2234. Armenta, S., Garrigues, S., & De La Guardia, M. (2008). Green analytical chemistry. Trends in Analytical Chemistry, 27, 497–511. Association of Official Analytical Chemists (2012). Official methods of analysis of AOAC 19th Appendix F Washington. Azevêdo, L. C., Reis, M. M., Da Silva, L. A., & De Andrade, J. B. (2007). Efeito da presença e concentração de compostos carbonílicos na qualidade de vinhos. Química Nova, 30, 1968–1975. Barros Neto, B., Bruns, R. E., & Scarmínio, I. S. (2006). Statistical designs: Chemometrics (1st ed.). London: Elsevier. Belitz, H. D., Grosch, W., & Schieberle, P. (2009). Food chemistry (4th ed.). Lichtenbergstraße: Springer.

203

Brasil (2003). Ministério da Saúde. Agência Nacional de Vigilância Sanitária (ANVISA) RE n° 899, de 29 de maio de 2003. Guia para validação de métodos analíticos e bioanalíticos. Caballero, B., Trugo, L. C., & Finglas, P. M. (2003). Encyclopedia of food sciences and nutrition (2nd ed.). Maryland: Elsevier. Capuano, E., & Fogliano, V. (2011). Acrylamide and 5-hydroxymethylfurfural (HMF): A review on metabolism, toxicity, occurrence in food and mitigation strategies. LWT Food Science and Technology, 44, 793–810. Chávez-Servín, J. L., de la Carbot, K. T., García-Gasca, T., Castellote, A. I., & LópezSabater, M. C. (2015). Content and evolution of potential furfural compounds in commercial milk-based infant formula powder after opening the packet. Food Chemistry, 166, 486–491. Davidson, A. G., & Dawodu, T. O. (1987). Difference spectrophotometric assay of 5-hydroxymethylfurfuraldehyde in hydrolysed pharmaceutical syrups – I. Sodium borohydride reagent. Journal of Pharmaceutical & Biomedical Analysis, 5, 213–222. Davidson, A. G., & Dawodu, T. O. (1988). Difference spectrophotometric assay of 5-hydroxymethylfurfuraldehyde in hydrolysed pharmaceutical syrups – II. Isoniazid reagent. Journal of Pharmaceutical & Biomedical Analysis, 6, 61–66. De Andrade, J. K., Komatsu, E., Perreault, H., Torres, Y. R., Da Rosa, M. R., & Felsner, M. L. (2016). In house validation from direct determination of 5-hydroxymethyl-2furfural (HMF) in Brazilian corn and cane syrups samples by HPLC-UV. Food Chemistry, 190, 481–486. de Ribeiro, F. A. L., Ferreira, M. M. C., Morano, S. C., Da Silva, L. R., & Schneider, R. P. (2008). Planilha de validação: Uma nova ferramenta para estimar figuras de mérito na validação de métodos analíticos univariados. Química Nova, 31, 164–171. Er Demirhan, B., Demirhan, B., Sönmez, C., Torul, H., Tamer, U., & Yentür, G. (2015). Short communication: Determination of potential 5-hydroxymethyl-2furaldehyde and 2-furaldehyde compounds in follow-on milks and infant formulas using the high-performance liquid chromatography method. Journal of Dairy Science, 98, 818–822. González, A. G., Herrador, M. Á., & Asuero, A. G. (2010). Intra-laboratory assessment of method accuracy (trueness and precision) by using validation standards. Talanta, 82, 1995–1998. Gürkan, R., & Altunay, N. (2015). Quantification of 5-hydroxymethylfurfural in honey samples and acidic beverages using spectrophotometry coupled with ultrasonic-assisted cloud point extraction. Journal of Food Composition and Analysis, 42, 141–151. International Conference on Harmonisation (ICH). (1996). Validation of Analytical Procedures: Methodology, Q2B. International Standard, Accuracy (Trueness and Precision) of Measurements Methods and Results (1994). ISO 5725-6, Geneva. Kavousi, P., Mirhosseini, H., Ghazali, H., & Ariffin, A. A. (2015). Formation and reduction of 5-hydroxymethylfurfural at frying temperature in model system as a function of amino acid and sugar composition. Food Chemistry, 182, 164–170. Kowalski, S., Lukasiewicz, M., Duda-Chodak, A., & Ziec´, G. (2013). 5-Hydroxymethyl2-furfural (HMF) – heat-induced formation, occurrence in food and biotransformation: A review. Polish Journal of Food and Nutrition Science, 63. Kuttatharmmakul, S., Massart, D. L., & Smeyers-Verbeke, J. (1999). Comparison of alternative measurement methods. Analytica Chimica Acta, 391, 203–225. Lee, T. P., Sakai, R., Manaf, N. A., Rodhi, A. M., & Saad, B. (2014). High performance liquid chromatography method for the determination of patulin and 5hydroxymethylfurfural in fruit juices marketed in Malaysia. Food Control, 38, 142–149. Miller, J. N., & Miller, J. C. (2005). Statistics and chemometrics for analytical chemistry (5th ed.). Harrow: Pearson Education Limited. Minitab. Minitab Incorporation version 16.2.2 (2010). USA. Newton, D. E. (2007). The new chemistry: Food chemistry (1st ed.). New York: Facts on File Science Library. Ribani, M., Bottoli, C. B. G., Collins, C. H., Jardim, I. C. S. F., & Melo, L. F. C. (2004). Validação em métodos cromatográficos e eletroforéticos. Química Nova, 27, 771–780. Rouessac, F., & Rouessac, A. (2007). Chemical analysis: Modern instrumentation and methods and techniques (2nd ed.). Chichester: John Wiley & Sons Ltd.. Spano, N., Ciulu, M., Floris, I., Panzanelli, A., Pilo, M. I., Piu, P. C., ... Sanna, G. (2009). A direct RP-HPLC method for the determination of furanic aldehydes and acids in honey. Talanta, 78, 310–314. Taher, A. M., & Cates, D. M. (1974). A spectrophotometric investigation of the yellow color that accompanies the formation of furan derivatives in degraded-sugar solutions. Carbohydrate Research, 34, 249–261. Teixidó, E., Núñez, O., Santos, F. J., & Galceran, M. T. (2011). 5-Hydroxymethylfurfural content in foodstuffs determined by micellar electrokinetic chromatography. Food Chemistry, 126, 1902–1908. Truzzi, C., Annibaldi, A., Illuminati, S., Finale, C., Rossetti, M., & Scarponi, G. (2012). Determination of Very Low Levels of 5-(Hydroxymethyl)-2-furaldehyde (HMF) in Natural Honey: Comparison between the HPLC Technique and the Spectrophotometric White Method. Journal of Food Science, 77, 784–790. White, J. (1979). Spectrophotometric method for hydroxymethylfurfural in honey. Journal of the Association of Official Analytical Chemists, 62, 509–514. Zappalà, M., Fallico, B., Arena, E., & Verzera, A. (2005). Methods for the determination of HMF in honey: A comparison. Food Control, 16, 273–277. Zhang, H., Wei, L., Liu, J., Lin, S., & Yuan, Y. (2014). Detection of 5-hydroxymethyl-2furfural levels in selected Chinese foods by ultra-high-performance liquid chromatograph analytical method. Food Analytical Methods, 7, 181–188.