You are on page 1of 8

Food Chemistry 365 (2021) 130514

Contents lists available at ScienceDirect

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

Comparison of ultraviolet and refractive index detections in the HPLC


analysis of sugars
Iqbal Jalaludin a, Jeongkwon Kim a, b, *
a
Department of Chemistry, Chungnam National University, Daejeon 34134, Republic of Korea
b
Graduate School of New Drug Discovery and Development, Chungnam National University, Daejeon, Republic of Korea

A R T I C L E I N F O A B S T R A C T

Keywords: Refractive index (RI) detection is the standard approach for quantitatively detecting sugars via high-performance
Honey liquid chromatography (HPLC), while ultraviolet (UV) absorbance detection is the most commonly used
HPLC detection method for general HPLC analysis. We compared the two detection approaches of UV and RI in the
Refractive index
HPLC analysis of small sugars to investigate whether UV detection could be an alternative method to RI
Sugar
Ultraviolet
detection. UV detection was performed using a photodiode array scanning from 190 to 400 nm. We obtained
comparable limit of detection (LOD) results for RI and UV detection in the HPLC analysis of monosaccharides,
while HPLC-RI provided better LOD results than HPLC-UV in disaccharide analysis. Both HPLC-RI and HPLC-UV
methods were applied to analyze a real honey sample, and similar results were obtained in terms of precision and
recovery. The study conclusively shows that the UV-based HPLC analysis of sugars offers a sufficient alternative
to RI-based HPLC analysis.

1. Introduction absence of UV chromophores in sugars, RI detection may be preferable


over UV-based detection for sugar analyses. However, this limitation
Ultraviolet (UV)-based detection is widely used in high-performance does not completely restrict UV-based detection, as UV absorption of
liquid chromatography (HPLC), primarily because it is easy to operate sugars is observed below 200 nm (Uchiho et al., 2015). To our knowl­
and relatively sensitive (Swartz, 2010). However, a refractive index (RI) edge, only a few studies have been performed using the UV-based
detector is typically used as a universal detector to detect compounds identification of unlabeled sugars, with sugars in fruits and juices
such as carbohydrates that do not have UV chromophores (Zhang, Li, & detected at 190 nm using HPLC-UV (Karkacier, Erbas, Uslu, & Aksu,
Yan, 2008). These detection methods share some similarities, but the 2003; Shaw & Wilson, 1983).
sensitivity and stability for a specific analyte vary when using different The most apparent superiority of UV-based detection in HPLC over
detectors (Scott, 1986). RI detection is its compatibility with gradient elution. Gradient elution
Carbohydrate analysis is a standard practice in many food industries can either shorten the acquisition time or increase the retention time
and is generally used for food labeling, quality assurance, and testing of difference between two adjacent peaks, ultimately improving the chro­
food additives (BeMiller, 2017). The use of chromatography improves matogram resolution. Moreover, gradient elution minimizes the use of
carbohydrate testing, and is considered a better alternative to colori­ solvents, which is undoubtedly beneficial to many industrial sectors.
metric tests (e.g., phenol-sulfur assay), which quantify total carbohy­ Stability is not always an issue with UV detection, unlike in RI detection
drates rather than individual sugar components (BeMiller, 2017). There where temperature and pressure must be precisely controlled to main­
are several benefits of quantifying individual sugars instead of an overall tain a consistent baseline (Hirata, Kawaguchi, & Funada, 1996). In
carbohydrate estimation, including the investigation of honey adulter­ general, the lack of chromophore-containing compounds, such as sugars,
ation (Se, Wahab, Syed Yaacob, & Ghoshal, 2019). Assessments for does not benefit greatly from UV detection, because individual sugars do
honey adulteration are typically conducted by evaluating the percentage not have unique UV absorptions. However, a detector such as a photo­
of monosaccharides and disaccharides, with the former expected to be diode array (PDA) can measure absorbance over a broad wavelength
within 70% and the latter about 5% (Wang et al., 2015). Due to the range while discerning sugars from other components whose UV

* Corresponding author at: Department of Chemistry, Chungnam National University, Daejeon 34134, Republic of Korea.
E-mail address: jkkim48105@cnu.ac.kr (J. Kim).

https://doi.org/10.1016/j.foodchem.2021.130514
Received 4 March 2021; Received in revised form 21 May 2021; Accepted 29 June 2021
Available online 1 July 2021
0308-8146/© 2021 Elsevier Ltd. All rights reserved.
I. Jalaludin and J. Kim Food Chemistry 365 (2021) 130514

absorbance is beyond the intended wavelength range. Consequently, a 2.4. Method validation
more reliable HPLC quantification can be performed for a complex
sample. For comparative purposes, all HPLC methods were validated using
In this study, we compared the performance of UV and RI detectors in the standard validation practice in compliance with the International
carbohydrate analysis, where UV detection was performed using a PDA Conference on Harmonization (ICH) Q2(R1) guidelines (ICH, 1995). The
detector. Two elution modes, i.e., isocratic and gradient, were compared assessment included linearity, sensitivity, accuracy, and precision of the
to assess the reliability of the approaches for their subsequent use on a methods.
real honey sample.
2.4.1. Linearity
2. Materials and methods The standard sugars were analyzed in triplicate at concentrations of
0.039, 0.078, 0.156, 0.313, 0.625, 1.25, 2.5, 5, and 10 mg/mL to obtain
2.1. Materials and reagents the mean and standard deviation and construct calibration curves by
plotting the peak area against the known sugar concentrations. Linear
Fructose was purchased from Kanto Chemical Co. Inc. (Tokyo, regression analysis was used to measure the slopes and coefficients of
Japan). Glucose, sucrose, and maltose monohydrate were purchased determination (R2) of the best-fitting lines.
from Sigma-Aldrich (St. Louis, MO, USA). Acetonitrile (HPLC-grade)
was obtained from Merck (Darmstadt, Germany). A commercial honey 2.4.2. Sensitivity
sample was obtained from a local market in South Korea. The honey The sensitivity of the method was measured using a linear regression
sample used in this study had previously passed the standard tests approach in which the limit of detection (LOD) and the limit of quan­
required by the Korea Food and Drug Administration (KFDA) for quality tification (LOQ) were calculated as follows:
control purposes. Water was purified using the Human Power I+ Water σ
Purification System (Seoul, South Korea). LOD = 3.3 ×
S

2.2. Standard and sample preparation LOQ = 10 ×


σ
S
A stock solution (10 mL) containing 100 mg each of fructose,
where σ is the standard deviation of the y-intercept and S is the standard
glucose, sucrose, and maltose was prepared by homogeneously dis­
curve slope.
solving each sugar in a small amount of water with mild heating,
combining all sugar solutions into a volumetric flask, and finally adding
2.4.3. Accuracy and precision
water to make up a 10-mL solution. Standard working solutions for
Accuracy and precision were measured using the standard sugar
generating a calibration curve were prepared from the stock solution by
solutions at high, medium, and low concentrations of 10, 5, and 0.625
two-fold serial dilutions (10–0.039 mg/mL).
mg/mL, respectively. Accuracy was expressed as percent recovery (%
To prepare the honey sample solution, 100 μL of pure honey was
recovery) by comparing the obtained value with the theoretical value,
diluted with water to prepare a 10-mL solution (100-fold dilution). The
while, precision was expressed as percent relative standard deviation (%
mass of the 100 μL of pure honey was measured using an analytical
RSD) to measure the closeness of five replicates (Rao, 2018). Two pre­
balance (OHAUS Adventurer, Parsippany, NJ, USA) for the accurate
cision procedures were considered: repeatability (inter-day precision
determination of the sugar content in honey. The honey was allowed to
analysis) and intermediate precision (inter-day precision analysis). The
immerse in the water for 1 h and the solution was then vortexed to
%recovery and %RSD were calculated as follows:
ensure complete dissolution. Both standards and samples were filtered
using a 0.2-µm pore size syringe filter (Whatman™ Puradisc, Buck­ Measuredconcentration
%recovery = × 100
inghamshire, UK) before HPLC analysis. Actualpreparedconcentration

2.3. HPLC operating conditions %RSD =


Standarddeviationofthemeasuredconcentration
× 100
Meanofthemeasuredconcentration
All HPLC analyses in this study were conducted using an Alliance
HPLC system (Waters, Milford, MA, USA), which consisted of an e2695 2.4.4. Robustness
separation module, 2998 PDA detector, and 2414 RI detector. Both Robustness was tested using a full factorial experimental design to
detector units were used simultaneously in isocratic mode, but only the see how small chromatographic changes in the experimental conditions
UV detector was used in gradient mode. The HPLC system was equipped affected the quantification and system compatibility. Three parameters
with a Shodex Asahipak NH2P-50 4E column (4.6 mm internal diameter (flow rate, column temperature, and solvent composition) were tested at
× 250 mm; Showa Denko KK, Tokyo, Japan). The system was semi- two levels: low (− 1) and high (1), as well as one center point (0),
controlled with the Waters Empower 3 Chromatography software. generating nine random experiments (Supplementary Table S1). The
In isocratic mode, the mobile phase was acetonitrile:water (75:25, v/ experiments were carried out sequentially using the fructose sample
v) (Tihomirova, Dalecka, & Mezule, 2016), which was retained for 15 solution (10 mg/mL), and the resulted peak area and retention time
min at a flow rate of 1.0 mL/min. In gradient mode, the mobile phase were computed and analyzed by Minitab® 20.2 statistical software
consisted of acetonitrile (solvent A) and water (solvent B), from which (Minitab Inc., Pennsylvania, USA).
the gradient procedure started at 35% of solvent B for 4 min and
decreased linearly to 20% from 4 to 10 min (Supplementary Fig. S1). 2.5. Sugar content in the honey sample
Other gradient conditions were also attempted (Supplementary Fig. S2).
In both isocratic and gradient elution modes, the sample injection vol­ The sugar content in the honey sample was measured in two ways.
ume was 20 μL and the column temperature was 35 ◦ C. The wavelength First, the linear equation from the calibration curve was used, with the
range of the UV detector was set from 190 to 400 nm for simultaneous sugar concentrations determined by substituting the peak area value at
detection. The peak area was measured at 190 nm. the respective elution time of the sugar in the honey sample to the linear
equation. Second, the amount of sugar in the honey was calculated using
an equation proposed by the KFDA to validate the sugar composition of
honey (Food Safety Korea, 2020):

2
I. Jalaludin and J. Kim Food Chemistry 365 (2021) 130514

AU V W’ are typically composed of carbon–carbon (C–C) single bonds, hydroxyl


%SC = × × × 100
AU’ V’ W (OH) groups, and ether groups. The λmax of C–C bonds was approxi­
mately 160 nm, while the λmax of the OH groups was about 150 nm
Here, AU and AU′ are the peak area of the sample and the standard, V (Higashi, Ikehata, & Ozaki, 2007; Tachibana et al., 2011). The cyclic
and V′ are the volume of the sample and the standard, and W and W′ are ether in the tetrahydrofuran of fructose and sucrose, and the tetrahy­
the weight of the sample and the standard, respectively. The calculated dropyran of glucose, sucrose, and maltose (Fig. 1d insert) have absorp­
result was expressed as a percentage, and the sugar concentration (%SC) tion peaks at about 58 × 103 cm− 1 (~172 nm) and 57 × 103 cm− 1
was measured using the 10 mg/mL sugar standard solution. (~175 nm), respectively (Pickett, Hoeflich, & Liu, 1951). The λmax for
both fructose and glucose has been reported to be 178 nm, while that of
3. Results and discussion sucrose is 175 nm (Uchiho et al., 2015). These observations indicate that
UV absorption by fructose and glucose is dependent on cyclic ether
3.1. HPLC-RI and UV analyses of sugars using isocratic elution groups, while that of sucrose is dependent on C–C bonds or OH groups.

Fructose, glucose, sucrose, and maltose are commonly observed in


3.2. HPLC-UV analysis of sugars using gradient elution
honey (Lee et al., 2020). Isocratic separation of the fructose, glucose,
sucrose, and maltose mixture was performed using an acetonitrile:water
One of the main advantages of UV detection over RI detection is that
(75:25, v/v) mobile phase, with sequential detection using RI and UV
both isocratic and gradient elution can be used in UV detection, while
detectors. A PDA detector was used to obtain the UV absorbance of the
only isocratic elution is possible in RI detection. Stability is the primary
sugars in the wavelength range from 190 to 400 nm. All four sugars were
reason that RI detection is undesirable for application in gradient elution
measured successfully using both detectors. All four sugar peaks in the
because temperature and pressure must be precisely regulated to
RI chromatogram appeared at a similar height (Fig. 1a). However, in the
maintain a stable baseline (Hirata et al., 1996).
UV chromatogram at 190 nm, the fructose peak was approximately five
Gradient elution was attempted and optimized for the HPLC-UV
times higher than the other peaks (Fig. 1b). The higher absorbance of
detection of sugars to shorten the analysis time while ensuring that
fructose than glucose or sucrose at 190 nm has been reported previously
the sugar peaks did not overlap. Supplementary Fig. S2 shows 12
(Uchiho et al., 2015). The UV contour map in Fig. 1c shows the fructose
different gradient elution profiles and their corresponding chromato­
UV absorption at wavelengths higher than 190 nm.
grams from the investigation. The optimized gradient profile alongside
The wavelength of maximum absorbance (λmax) for each sugar was
the chromatogram is shown in Supplementary Fig. S2L, where solvent B
below 190 nm, with the peaks increasing continuously at the extreme
(water) began at 35% for 4 min and decreased linearly to 20% from 4 to
left of the UV spectra, as shown in Fig. 1d. However, due to the limi­
10 min (Supplementary Fig. S1). The solvent composition with a
tations of typical commercial UV detectors, 190 nm is the lowest
stronger elution strength (solvent B) should be gradually increased in a
available wavelength. The low λmax values are related to the distinctive
conventional gradient elution procedure. However, the composition of
configurations of fructose and the other sugars studied. Neutral sugars
solvent B had to decrease to obtain better separation of the sugars in this

Fig. 1. Chromatograms and spectra of fructose, glucose, sucrose, and maltose: (a) RI chromatogram, (b) UV isocratic chromatogram at 190 nm, (c) UV contour
spectrum, and (d) UV absorption spectra.

3
I. Jalaludin and J. Kim Food Chemistry 365 (2021) 130514

study. Here, Δtr is the distance between the peaks and wav is the average width
Fig. 2 shows the UV chromatogram and the UV contour spectrum of of the two peaks in their respective units (both were calculated in time
the mixture of four sugars. Gradient elution reduced the total analysis units) (Harris, 2003). The resolution values obtained using the two
time from 15 min in the isocratic elution, as shown in Fig. 1, to less than neighboring peaks were equivalent to 1. Supplementary Table S2 shows
10 min. In general, all sugar peaks were shifted to earlier retention times the resolution values for the three methods (HPLC-RI isocratic, HPLC-UV
with gradient elution compared to isocratic elution. Furthermore, the isocratic, and HPLC-UV gradient).
baseline (Fig. 2a) of gradient elution was slightly wavier than that of the The performance and effectiveness of columns can be expressed
isocratic method (Fig. 1b), but this did not affect the consistency of the using the number of theoretical plates. The column used in this study
resulting peaks in terms of their areas and retention times. was the same as that used in many previous sugar analysis experiments
Peak symmetry and resolution are critical considerations that ensure (Barreira, Pereira, Oliveira, & Ferreira, 2010; Ma, Sun, Chen, Zhang, &
the gradient profile is optimized. The peaks in Fig. 2a are symmetric, Zhu, 2014). It was important to know how efficient the column was to
with no tailing or fronting. The resolution was determined based on the ensure that it was compatible with our research. The more theoretical
difference in elution time between the peaks and their widths. The plates in a column, the more likely it is to achieve an equilibrium be­
resolution was calculated as follows: tween the stationary and mobile phases and the higher the separation
Δtr efficiency is. The number of theoretical plates was determined as
Resolution = follows:
wav

Fig. 2. (a) HPLC-UV gradient chromatogram at 190 nm and (b) UV contour spectrum of fructose, glucose, sucrose, and maltose. Gradient elution shortened the
analysis time without compromising the quantitative and qualitative details of sugars.

4
I. Jalaludin and J. Kim Food Chemistry 365 (2021) 130514

( )2
tr Table 1
Numberoftheoreticalplates = 5.54
w0.5 The LOD, LOQ range, and R2 for the HPLC-RI, HPLC-UV isocratic, and HPLC-UV
gradient methods.
Here, tr is the retention time of the peaks, and w0.5 is the width at half of Methods Compounds LOD [mg/ Range (from LOQ) R2 a

the peak height. Supplementary Table S3 shows the number of theo­ mL] [mg/mL]
retical plates measured at the fructose peak for all three methods. The HPLC-RI Fructose 0.111 0.337–10 1.000
largest theoretical plate number was 2732, which was obtained from the Glucose 0.021 0.064–10 0.999
HPLC-UV gradient, where the value was almost double that of the Sucrose 0.019 0.058–10 0.999
Maltose 0.016 0.047–10 0.999
isocratic approaches.
HPLC-UV Fructose 0.044 0.134–10 1.000
isocratic Glucose 0.063 0.190–10 0.999
Sucrose 0.326 0.988–10 0.995
3.3. Preparation of the calibration curves
Maltose 0.475 1.440–10 0.994

To ensure that all three methods could be used on the actual sample, HPLC-UV Fructose 0.021 0.065–10 0.999
gradient Glucose 0.087 0.263–10 0.996
calibration curves were prepared to determine the linear relationship
Sucrose 0.125 0.378–10 0.990
between sugar concentration and response unit (Fig. 3). The y-axis of Maltose 0.522 1.581–10 0.998
each calibration curve was set as a peak area unit instead of the corre­ a
R2 was measured within the calibration curve range in Fig. 3 (0.039–10 mg/
sponding detectors’ responding units (RIU for RI and AU for UV) to
mL).
allow the comparison to be more visible. The RI detection calibration
curve (Fig. 3a) indicated a similar response for all sugars. In contrast, UV
detection for both the isocratic and gradient modes revealed very similar 3.5. Repeatability, intermediate precision, and recovery
profiles (Fig. 3b and c). The profiles revealed that fructose had a slope
nearly four times steeper than those of the other sugars. Supplementary Table S4 presents the precision and accuracy datasets
There were significant deviations at the lower sugar concentrations for the HPLC-RI isocratic, HPLC-UV isocratic, and HPLC-UV gradient
(0.625–0.0391 mg/mL) in all methods, particularly for RI detection methods. All %RSD values in the repeatability test (intra-day precision)
(Fig. 3a insert). The peak area variations were most likely due to the for the high (10 mg/mL), medium (5 mg/mL), and low (0.625 mg/mL)
occasionally inconsistent baseline. In addition, the Waters Empower sugar concentrations were less than 2.5%, except for HPLC-UV mea­
software’s automated peak area estimation can lead to fluctuations for surements of maltose with gradient elution at the low sugar concen­
diluted samples. Therefore, if the sample concentration is close to the tration, was found to be greater than 3.5% RSD. All three methods
quantitative limit, a manual assessment of the peak area is advised. generally had better repeatability for high and medium concentrations
than the low concentration. The inter-day precision %RSD values were
higher (range: 0.15–3.53%) than the intra-day precision %RSD values
3.4. Method validation (range: 0.33–5.86%).
In terms of accuracy, all three methods had %recovery values close to
The sensitivity of all three methods tested was expressed in terms of 100%. In particular, the HPLC-RI isocratic method collectively had the
the LOD and LOQ. Table 1 displays LOD, LOQ, and R2 values for each least deviation from 100%, except for the 5 mg/mL fructose sample,
sugar. The LOD and LOQ values of the RI detector for all sugars were which had a value of 109.7%. Because the isocratic elution for the RI and
generally comparable to those obtained previously using the same UV detectors always operated simultaneously, which will have reduced
detection method (Barreira et al., 2010); for instance, the LOD values of the disparity in the analyte environment, any inconsistency between the
fructose, glucose, and sucrose were 0.05, 0.08, and 0.06 mg/mL in the two HPLC methods could only be due to the detectors.
report and 0.111, 0.021, and 0.019 mg/mL in the current study,
respectively. 3.6. Robustness
In contrast, LOD and LOQ values of the UV detector showed com­
parable values to the RI detector for only the two monosaccharides. For Supplementary Fig. S3 summarizes the degree of chromatographic
the disaccharides, the LOD and LOQ values using UV detection were not variances caused by minor changes in three parameters: flow rate, col­
comparable to those using RI detection; in particular, the LOQ for umn temperature, and solvent composition. The parameters were
maltose was highest for the HPLC-UV isocratic and gradient methods, selected because they are more likely to vary when the method is
1.440 and 1.581 mg/mL, respectively, and only 0.047 mg/mL for the transferred across laboratories or when different operators handle the
HPLC-RI method. The LOD and LOQ values for the comparison of UV experiments (Vander Heyden, Nijhuis, Smeyers-Verbeke, Vandeginste,
and RI detectors were closely related based on R2, where only the su­ & Massart, 2001). As its number of variables was small, a two-level full
crose and maltose analyzed by the UV detector gave R2 < 0.996. factorial design was the preferred technique for evaluating robustness

Fig. 3. Calibration curves generated from serial dilution of standard working solutions for the (a) HPLC-RI, (b) HPLC-UV isocratic, and (c) HPLC-UV gradient
methods. All calibration curves had an acceptable linear range for all sugars (0.039–10 mg/mL) analyzed.

5
I. Jalaludin and J. Kim Food Chemistry 365 (2021) 130514

(Ferreira et al., 2017). Consequently, the peak area and retention time were too ambiguous to be defined; therefore, we did not include them in
were measured, as recovery and separation quality are the primary foci the final estimation.
of this investigation. The sugar concentrations in Fig. 5 were estimated from the calibra­
In response to the peak area, the flow rate displays a greater effect tion curve and expressed in mg/mL. To evaluate the differences among
than the other factors (Supplementary Fig. S3a). This is most likely the three methods, the %RSD values of the three methods for fructose,
owing to a longer period for the sugar to be maintained on the stationary glucose, and sucrose were calculated (Supplementary Table S5). Fruc­
phase when using a low flow rate, which causes the peak’s base width to tose and glucose had a low variation of 0.02–0.03% RSD, while
widen. As a result, the peak area corresponding to the concentration of considerable variation was apparent in sucrose, with a value of 0.3%
sugar increases, and the recovery might be substantially impacted. On RSD (Supplementary Table S5).
the other hand, the retention time was considerably impacted by two
factors: flow rate and solvent composition (Supplementary Fig. S3b).
Thus, the two aspects (flow rate and solvent composition) must be 3.8. Advantages of UV detection over RI detection
carefully evaluated to achieve a high quality of separation.
Compared to RI detection, UV detection enabled the use of gradient
elution, which provided the greatest benefits of decreasing the operating
3.7. Sugar content in a honey sample time and cost of analysis. In addition, UV detection had greater baseline
stability, with less time required to achieve a stable baseline. This was
Simple sugars such as fructose and glucose are the main components apparent from the calibration curve, where large deviations were found
of honey, with the remaining 17–20% being water (Kamal & Klein, at low sugar concentrations (0.625–0.0391 mg/mL) for RI detection
2011). To analyze the saccharides in the honey sample, a commercial (Fig. 3a insert).
honey sample was diluted 100-fold with water before the HPLC analyses UV detection is usually considered to have greater sensitivity than RI
to lessen its viscosity so that the friction between the column packed detection for compounds with a UV chromophore, with a detection
particles and samples decreased, allowing the flow rate to be sustained capability in the nanomole range (Swartz, 2010). However, the use of
without putting much pressure on the system. UV detection may be inefficient for non-UV-chromophore compounds,
The three methods evaluated in this study were able to measure such as sugars. Regardless, our results supported HPLC-UV analysis of
fructose, glucose, and sucrose in the honey sample because the three sugar as an alternative because the sensitivity of the two detectors was
sugars were within a quantifiable range (Fig. 4). Fig. 5 shows the sugar comparable for all sugars tested. If a more reliable detection is required
contents in the honey using the three methods, which were very similar to enhance sensitivity, labeling methods, such as 1-phenyl-3-methyl-5-
as a consequence of the calculation method used. The peaks at 6.5–7.5 pyrazolone labeling, are recommended to narrow the detection to the
min in Fig. 4d might be disaccharide variants, such as trehalose and picomole range (Honda et al., 1989). However, an unlabeled analysis
cellobiose, as reported previously (De La Fuente, Ruiz-Matute, Valencia- with UV-based detection is adequate for sugar analysis in the micromole
Barrera, Sanz, & Martínez Castro, 2011). However, because the peaks range.
were smaller than the quantification limit used in this analysis, they As indicated by previous studies, RI detection is the gold standard for

Fig. 4. Chromatograms and spectra of honey: (a) HPLC-RI chromatogram, (b) HPLC-UV isocratic chromatogram at 190 nm, (c) isocratic UV contour spectrum, (d)
HPLC-UV gradient chromatogram at 190 nm, and (e) gradient UV contour spectrum. a Concentration was estimated from three replications using the calibration
curve of the respective standard. b Percentage in honey was estimated from three replications using the KFDA formula based on the respective 10 mg/mL standards.

6
I. Jalaludin and J. Kim Food Chemistry 365 (2021) 130514

Fig. 5. Estimates of the sugar content in a honey sample using the three methods validated in this study.

detecting sugars (Chávez-Servín, Castellote, & López-Sabater, 2004). through the National Research Foundation of Korea (NRF) funded by the
However, the current study showed that the UV detection of mono­ Ministry of Education (2016R1D1A1B02008854) and by Basic Science
saccharides using both isocratic and gradient methods provided com­ Research Capacity Enhancement Project through the Korea Basic Sci­
parable results to RI detection in terms of the LOD and LOQ. For the ence Institute (National Research Facilities and Equipment Center) grant
monosaccharides tested, the peak areas between the RI and UV detectors funded by the Ministry of Education (Grant No. 2019R1A6C1010030).
were similar, although the two monosaccharides do not have chromo­
phores. Satisfactory sensitivity results were not obtained for the di­ Appendix A. Supplementary data
saccharides using HPLC-UV methods. However, HPLC-UV detection
methods may be useful for measuring high concentrations of Supplementary data to this article can be found online at https://doi.
disaccharides. org/10.1016/j.foodchem.2021.130514.

4. Conclusions References

Barreira, J. C. M., Pereira, J. A., Oliveira, M. B. P. P., & Ferreira, I. C. F. R. (2010). Sugars
The HPLC-RI and HPLC-UV analyses of sugars, particularly mono­
profiles of different chestnut (Castanea sativa Mill.) and almond (Prunus dulcis)
saccharides, were comparable in terms of performance, even though UV cultivars by HPLC-RI. Plant Foods for Human Nutrition, 65(1), 38–43. https://doi.org/
detection is primarily used to identify compounds with a UV chromo­ 10.1007/s11130-009-0147-7.
phore. It is important to note that the two detection methods for the BeMiller, J. N. (2017). Carbohydrate analysis. In S. S. Nielsen (Ed.), Food analysis (pp.
333–360). Cham: Springer International Publishing. https://doi.org/10.1007/978-3-
isocratic mode were reasonably validated by many previous studies and 319-45776-5_19.
have been used widely to determine the composition of sugars in foods. Chávez-Servín, J. L., Castellote, A. I., & López-Sabater, M. C. (2004). Analysis of mono-
We compared these methods with our newly developed gradient elution and disaccharides in milk-based formulae by high-performance liquid
chromatography with refractive index detection. Journal of Chromatography A, 1043
method for UV detection to confirm that all three methods could be used (2), 211–215. https://doi.org/10.1016/j.chroma.2004.06.002.
for sugar analysis. The significantly increased number of plates in the De La Fuente, E., Ruiz-Matute, A. I., Valencia-Barrera, R. M., Sanz, J., & Martínez
HPLC-UV gradient method suggested that more complex sugar samples Castro, I. (2011). Carbohydrate composition of Spanish unifloral honeys. Food
Chemistry, 129(4), 1483–1489. https://doi.org/10.1016/j.foodchem.2011.05.121.
could be successfully analyzed. Although RI detection is useful for sugar Ferreira, S. L. C., Caires, A. O., Borges, T. da S., Lima, A. M. D. S., Silva, L. O. B., & dos
quantification, direct UV analysis (with no pretreatment and no sample Santos, W. N. L. (2017). Robustness evaluation in analytical methods optimized
labeling) may also be a practical option given its versatility and time- using experimental designs. Microchemical Journal, 131(October 2017), 163–169.
10.1016/j.microc.2016.12.004.
and cost-effectiveness when paired with gradient elution. Food Safety Korea. (2020). 6.12.1 Honey. Retrieved November 17, 2020, from
https://www.foodsafetykorea.go.kr/foodcode/01_03.jsp?idx=11167.
CRediT authorship contribution statement Harris, D. C. (2003). Introduction to analytical separations. In Quantitative chemical
analysis (6th ed.). W. H. Freeman. Retrieved from https://books.google.co.kr/boo
ks?id=csTsQr-v0d0C.
Iqbal Jalaludin: Writing – original draft, Resources, Investigation, Higashi, N., Ikehata, A., & Ozaki, Y. (2007). An attenuated total reflectance far-UV
Data curation, Visualization. Jeongkwon Kim: Writing – review & spectrometer. Review of Scientific Instruments, 78(10), 103107. https://doi.org/
10.1063/1.2796928.
editing, Supervision, Funding acquisition.
Hirata, Y., Kawaguchi, Y., & Funada, Y. (1996). Refractive index detection using an
ultraviolet detector with a capillary flow cell in preparative SFC. Journal of
Chromatographic Science, 34(1), 58–62. https://doi.org/10.1093/chromsci/34.1.58.
Declaration of Competing Interest Honda, S., Akao, E., Suzuki, S., Okuda, M., Kakehi, K., & Nakamura, J. (1989). High-
performance liquid chromatography of reducing carbohydrates as strongly
The authors declare that they have no known competing financial ultraviolet-absorbing and electrochemically sensitive 1-phenyl-3-methyl5-pyrazo­
lone derivatives. Analytical Biochemistry, 180(2), 351–357. https://doi.org/10.1016/
interests or personal relationships that could have appeared to influence 0003-2697(89)90444-2.
the work reported in this paper. ICH. (1995). ICH topic Q2 (R1) validation of analytical procedures: Text and
methodology. Retrieved November 10, 2020, from https://www.ema.europa.eu
/en/documents/scientific-guideline/ich-q-2-r1-validation-analytical-procedures-text
Acknowledgements -methodology-step-5_en.pdf.

This research was supported by Basic Science Research Program

7
I. Jalaludin and J. Kim Food Chemistry 365 (2021) 130514

Kamal, M. A., & Klein, P. (2011). Determination of sugars in honey by liquid of the Science of Food and Agriculture, 34(1), 109–112. https://doi.org/10.1002/
chromatography. Saudi Journal of Biological Sciences, 18(1), 17–21. https://doi.org/ jsfa.2740340116.
10.1016/j.sjbs.2010.09.003. Swartz, M. (2010). HPLC detectors: A brief review. Journal of Liquid Chromatography and
Karkacier, M., Erbas, M., Uslu, M. K., & Aksu, M. (2003). Comparison of different Related Technologies, 33(9–12), 1130–1150. https://doi.org/10.1080/
extraction and detection methods for sugars using amino-bonded phase HPLC. 10826076.2010.484356.
Journal of Chromatographic Science, 41(6), 331–333. https://doi.org/10.1093/ Tachibana, S., Morisawa, Y., Ikehata, A., Sato, H., Higashi, N., & Ozaki, Y. (2011). Far-
chromsci/41.6.331. ultraviolet spectra of n-alkanes and branched alkanes in the liquid phase observed
Lee, D., Kim, Y., Jalaludin, I., Nguyen, H.-Q., Kim, M., Seo, J., … Kim, J. (2021). MALDI- using an attenuated total reflection—far ultraviolet (ATR-FUV) spectrometer.
MS analysis of disaccharide isomers using graphene oxide as MALDI matrix. Food Applied Spectroscopy, 65(2), 221–226. https://doi.org/10.1366/10-06036.
Chemistry, 342, 128356. https://doi.org/10.1016/j.foodchem.2020.128356. Tihomirova, K., Dalecka, B., & Mezule, L. (2016). Application of conventional HPLC RI
Ma, C., Sun, Z., Chen, C., Zhang, L., & Zhu, S. (2014). Simultaneous separation and technique for sugar analysis in hydrolysed hay. Agronomy Research, 14(5),
determination of fructose, sorbitol, glucose and sucrose in fruits by HPLC-ELSD. Food 1713–1719.
Chemistry, 145, 784–788. https://doi.org/10.1016/j.foodchem.2013.08.135. Uchiho, Y., Goto, Y., Kamahori, M., Aota, T., Morisaki, A., Hosen, Y., & Koda, K. (2015).
Pickett, L. W., Hoeflich, N. J., & Liu, T. C. (1951). The Vacuum Ultraviolet Absorption Far-ultraviolet absorbance detection of sugars and peptides by high-performance
Spectra of Cyclic Compounds. II. Tetrahydrofuran, Tetrahydropyran, 1,4-Dioxane liquid chromatography. Journal of Chromatography A, 1424, 86–91. https://doi.org/
and Furan. Journal of the American Chemical Society, 73(10), 4865–4869. https://doi. 10.1016/j.chroma.2015.11.006.
org/10.1021/ja01154a117. Vander Heyden, Y., Nijhuis, A., Smeyers-Verbeke, J., Vandeginste, B. G. M., &
Rao, T. N. (2018). In Calibration and Validation of Analytical Methods - A Sampling of Massart, D. L. (2001). Guidance for robustness/ruggedness tests in method
Current Approaches. InTech. https://doi.org/10.5772/intechopen.72087. validation. Journal of Pharmaceutical and Biomedical Analysis, 24(5–6), 723–753.
Scott, R. P. W. (Ed.). (1986). Chapter 8 The selection of the appropriate detector. In https://doi.org/10.1016/S0731-7085(00)00529-X.
Liquid chromatography detectors (Vol. 33, pp. 235–261). Elsevier. 10.1016/S0301- Wang, S., Guo, Q., Wang, L., Lin, L., Shi, H., Cao, H., & Cao, B. (2015). Detection of honey
4770(08)70046-0. adulteration with starch syrup by high performance liquid chromatography. Food
Se, K. W., Wahab, R. A., Syed Yaacob, S. N., & Ghoshal, S. K. (2019). Detection Chemistry, 172, 669–674. https://doi.org/10.1016/j.foodchem.2014.09.044.
techniques for adulterants in honey: Challenges and recent trends. Journal of Food Zhang, B., Li, X., & Yan, B. (2008). Advances in HPLC detection-towards universal
Composition and Analysis, 80(February), 16–32. https://doi.org/10.1016/j. detection. Analytical and Bioanalytical Chemistry, 390(1), 299–301. https://doi.org/
jfca.2019.04.001. 10.1007/s00216-007-1633-0.
Shaw, P. E., & Wilson, C. W. (1983). Separation of fructose, glucose and sucrose in fruit
by high performance liquid chromatography using UV detection at 190 nm. Journal

You might also like