40 J. Giebułtowicz et al. / J. Chromatogr.
A 1423 (2015) 39–46
[16] and the only use of CPE and LC–MS combination was reported
[3] for APCI as the ionization source. LC–ESI–MS/MS is the state-
of-the-art bioanalytical technique and is more frequently used in
the bioanalysis of drugs than LC–APCI–MS/MS. However, the sur-
factants introduced to the sample may cause matrix effects and ESI
is more sensitive to this phenomena than APCI [17]. Therefore, as
regards further applications of CPE in the bioanalysis of drugs, it is
of critical importance that the possibility of its compatibility with
LC–ESI–MS/MS is evaluated.
Bisoprolol is a selective ␤-1-adrenoceptor antagonist bereft of
any partial agonist effect [18]. Taking into account the social impor-
tance of bisoprolol, novel methods for its rapid, convenient and
environmentally safe determination in human plasma are still
needed. Previously reported methods of LC–MS/MS detection of
bisoprolol [19–28] did not use CPE as serum or plasma prepara-
tion technique. The most common one was protein precipitation
by the addition of organic solvents (e.g. acetonitrile) [20,21,24–26].
Although this technique is rapid and simple, its disadvantages
result from insufficient sample clean-up and include a shorter HPLC
column life, increased risk of significant matrix effects and con-
tamination of a mass spectrometer [29]. The SPE procedures using
Oasis polymeric mix sorbent cartridges were also applied for the
determination of bisoprolol alone or among other beta-antagonists
and beta-agonists [28]. SPE based on the hydrophilic-lipophilic bal-
anced sorbent beds is successfully used in the multi-analysis of
compounds of different polarity [30]. However, this technique may
be considered time consuming and rather expensive, especially if
there is a need for analyzing large numbers of samples in a pharma-
cokinetic study. LLE was also successfully applied in the bioanalysis
of bisoprolol [19,22,23]. This technique is often used in bioanaly-
sis as it is fast and inexpensive. Our previous experience confirms
that its application allows to avoid the contamination of an MS
ion source and extends the duration of an HPLC column [29,31,32].
However, LLE requires the use of organic solvents which are toxic
to laboratory staff and the environment, as well as expensive to dis-
pose of. CPE is an alternative technique with all the advantages of
LLE. Additionally, CPE is regarded safer and more environmentally
friendly due to the extraction being based on surfactants which are
less toxic and less expensive to dispose of [33]. An extensive search
did not return any reports on the application of CPE to the biso-
prolol determination nor the statistical comparison of CPE and the
well-established sample preparation techniques.
In this paper we introduce a novel combination of sample prepa-
ration with the separation and detection techniques. The aim of the
study was to assess the compatibility of CPE and LC–ESI–MS/MS and
the applicability of CPE to the bisoprolol determination in spiked
human plasma and clinical samples.
2. Materials and methods
2.1. Chemicals
The reference standard of bisoprolol fumarate was a kind gift
from ICN Polfa Rzeszów S.A. (Poland). Metoprolol tartrate (internal
standard, IS) and Trition X-114 was purchased from Sigma-Aldrich
(St. Louis, US). The solvents, HPLC gradient grade methanol, acetoni-
trile and formic acid 98% were purchased from Merck (Darmstadt,
Germany). Ultrapure water was obtained from the Millipore water
Table 1
Optimized MRM conditions and ion transitions.
Substance [M + H] + ion DP [V] EP [V]
Bisoprolol 326 91 10
Metoprolol (IS) 268 86 10
DP—declustering potential, EP—entrance potential, CE—collision energy, CXP—cell exit potential.
purification system (Milli-Q, Billerica, US). Blank plasma was
obtained from the Regional Center of Blood Donation and Treat-
ment (Warsaw, Poland).
2.2. Chromatographic and mass spectrometric conditions
The instrumental analysis was performed using Agilent 1260
Infinity (Agilent Technologies, Santa Clara, CA, US), equipped with
a degasser, autosampler, binary pump coupled to a hybrid triple
quadrupole/linear ion trap mass spectrometer QTRAP 4000 (AB
Sciex, Framingham, MA, US). The Turbo Ion Spray source was oper-
ated in the positive mode. The curtain gas, ion source gas 1, ion
source gas 2 and collision gas (all high purity nitrogen) were set
at 345 kPa, 207 kPa, 276 kPa and “high” instrument units, respec-
tively. The ion spray voltage and source temperature were 5000 V
and 600 ◦ C, respectively. The target compounds were analyzed in
the MRM mode (Table 1).
Chromatographic separation was achieved with a Symmetry
C18 column (150 mm × 4.6 mm, 3.5 ␮m, Waters, Milford, US). The
column was maintained at 40 ◦ C at the flow rate of 0.5 mL min−1 .
The mobile phases consisted of HPLC grade water with 0.1% formic
acid as the eluent A and methanol with 0.1% formic acid as the elu-
ent B. The gradient (%B) was as follows: 0 min. 10%; 1 min. 10%;
8 min. 90%, 16 min. 90%. The re-equilibration of the column to the
initial conditions lasted 4 min. The injection volume was 10 ␮L.
2.3. Standard solutions, calibration standards and quality control
samples
The standard stock solutions of bisoprolol and IS (1 mg mL−1 )
were prepared by dissolving an appropriate amount of the sub-
stance in methanol. The working solutions (10 ng mL−1 ) were
prepared by diluting standard stock solutions with water. The solu-
tions were stored at 4 ◦ C. The calibration standards for bisoprolol
were made at the following concentrations: 0.3, 0.9, 3.0, 10, 20, 30,
45, 60 and 70 ng mL−1 . The quality control samples were prepared
at the following levels: 0.9, 30 and 60 ng mL−1 . The calibration stan-
dards and quality control samples were stored at −80 ◦ C until use.
Sodium citrate was used as an anticoagulant.
2.4. Plasma samples from the pharmacokinetic study
Plasma samples from the pharmacokinetic study in humans
were obtained from the Pharmaceutical Research Institute [34].
All samples were previously analyzed by LLE and the LC–MS/MS
method adapted from Ding et al. [35]. Briefly, plasma sample (1 mL)
was mixed with the IS (metoprolol, 50 ␮L), 1 M NaOH (0.1 mL) and
shaken for 5 min with ethyl acetate (5 mL). After centrifugation the
organic layer was evaporated to dryness. The residue was recon-
stituted in 60% aqueous methanol (200 ␮L) and 30 ␮L aliquot was
injected onto the LC–MS/MS consisting of Alliance 2695 liquid chro-
matograph (Waters, Milford, MA, USA) coupled to a Quattro micro
API triple quadrupole mass spectrometer equipped with an electro-
spray ion source (Waters, Manchester, UK). The method was fully
validated according to the international guidelines [36] and all the
analysis were performed in accordance with the Good Laboratory
Practice (GLP).
Quantitative CE [V] CXP [V] Qualitative CE [V] CXP [V]
product ion product ion
116 27 8 222 21 12
116 27 8 133 37 10
 J. Giebułtowicz et al. / J. Chromatogr. A 1423 (2015) 39–46
2.5. Sample preparation
2.5.1. CPE
50 ␮L of the IS working solution, 300 ␮L of 7% Triton X-114
solution and 400 ␮L of 0.2 M NaOH were added to 250 ␮L plasma
sample and vortexed for 5 min. Next, the sample was incubated in
a water bath at 55 ◦ C for 20 min and centrifuged at 10 000 rpm at
25 ◦ C for 5 min. Then the aqueous phase was separated by decant-
ing. After the CPE 700 ␮L of acetonitrile was added to the micellar
phase. Alternatively to create more environmental friendly method
ethanol can be used in that step as well. The sample was incubated
at −20 ◦ C for 20 min and centrifuged at 10 000 rpm at 25 ◦ C for
5 min. The supernatant (500 ␮L) was diluted with an equal volume
of water (500 ␮L) and transferred to the vial. Ten ␮L aliquot was
injected onto the LC–MS/MS. All CPE experiments were performed
at the Bioanalysis and Drugs Analysis Department of the Medical
University of Warsaw.
2.5.2. LLE (LLE2)
The LLE2 experiments were performed at the Bioanalysis and
Drugs Analysis Department of the Medical University of War-
saw. Plasma sample (250 ␮L) was mixed with the IS (metoprolol,
50 ␮L), 1 M NaOH (25 ␮L) and shaken for 5 min with ethyl acetate
(1.25 mL). After centrifugation the organic layer was evaporated to
dryness. The residue was reconstituted in 60% aqueous methanol
(250 ␮L) and 10 ␮L aliquot was injected onto the LC–MS/MS.
2.6. Method validation
The method previously developed at the Pharmacology Depart-
ment of the Pharmaceutical Research Institute was fully validated
and the stability of bisoprolol was confirmed in various conditions
[34]. Therefore, only partial validation was performed for this study
according to the European Medicines Agency (EMA) [37] and US
Food and Drug Administration (FDA) [36] guidelines. Briefly, the lin-
earity range was selected as 0.3–70 ng mL−1 as described previously
[34]. The accuracy and precision of the method were determined
within-run and between-run using LLOQ (0.3 ng mL−1 ) and qual-
ity control (QC) samples (0.9, 30 and 60 ng mL−1 ). Carry over and
recovery was also studied.
As the aim of the study was to evaluate the compatibility of
CPE and LC–ESI–MS/MS, particular attention was paid to the matrix
effect assessment. Ion suppression or enhancement was initially
studied by the steady post-column infusion of the analyte, i.e. biso-
prolol or the IS at the concentration 60 ng mL−1 and the injection
of the extracted blank plasma sample. The second approach was
based on the method by Matuszewski et al. and allowed the quan-
titative assessment of the absolute and relative matrix effect by the
comparison of the analyte peak areas in the standard solutions and
Fig. 1. Optimization of the CPE method (n = 5). Experimental conditions: (A) 7% Triton X-114, 0.5 M NaOH; (B) extraction temperature at 55 ◦ C, 0.5 M NaOH; (C) extraction
temperature at 55 ◦ C, 7% Triton X-114.
41
post-extraction spiked samples. Various lots of matrix used for the
test included haemolyzed and hyperlipidaemic plasma samples.
2.7. Statistical analysis
The statistical evaluation of the results was performed with the
SAS software version 9.4. for Windows. The Shapiro–Wilk test was
used to evaluate normal distribution of QC results. The hypothesis
on the equality of variances was verified by the F-test (p ≥ 0.05).
The linear regression for bisoprolol concentration measured using
the CPE and LLE methods was calculated using the least squares
method, the 95% confidence limit band of the mean response at
particular LLE points and the 95% prediction interval limits.
3. Results and discussion
3.1. Optimization of the CPE procedure
Triton X-114 is more frequently chosen for the analysis of bio-
logical fluids [1,8–10,12,16,20,38–40] than other surfactants such
as Triton X-100 and Genapol X-080. All experiments were per-
formed using Triton X-114 due to its main advantages namely low
cloud-point temperature (23 ◦ C) and high density which facilitate
phase separation by centrifugation [16]. During the optimization of
the CPE conditions the following changes in the sample preparation
protocol were made: plasma samples were fortified with bisoprolol
and water instead of the IS working solution. The IS working solu-
tion was added in the last step of the extraction process instead of
water.
3.1.1. Effect of the equilibrium temperature
The CPE is based on the separation of two phases, when the
aqueous solutions of the surfactant are heated above the cloud-
point temperature. Theoretically, the optimal temperature for the
extraction is 15–20 ◦ C greater than the cloud-point of the surfac-
tant. With the increase of the equilibrium temperature, the volume
of the surfactant-rich phase decreases due to the disruption of the
hydrogen bonds and the dehydration of the phase [16]. However,
too high temperature may reduce the recovery—because of both the
decomposition of the heat labile compounds as well as the thermal
instability of the surfactant aggregates [8]. The range of 40–55 ◦ C
was selected for the optimization experiments. Due to a possi-
ble analyte instability, thermal stability problems of the surfactant
aggregates and denaturation of plasma proteins it was decided
not to perform experiments in higher temperatures. The analyti-
cal signal of bisoprolol increased with increasing the equilibrium
temperature, thus 55 ◦ C was selected as the working equilibrium
temperature (Fig. 1A). One can observe that standard deviation
for selected temperature was the highest one, still during method
validation acceptance criteria for precision were met.
42 J. Giebułtowicz et al. / J. Chromatogr. A 1423 (2015) 39–46

3.1.2. Effect of the surfactant concentration Table 2

Accuracy and precision of bisoprolol determination by CPE and LLE.
Increasing surfactant concentration improves the recovery, but
above a certain point it dilutes the compound of interest and Nominal bisoprolol concentration [ng/ml] 0.9 30.0 60.0
decreases the preconcentration rate [12]. The concentrations of Within-run accuracy [%]
Triton X-114 were optimized in the range of 5–20% which corre- CPE (n = 5) 89.9a 91.4 98.2
sponded to the concentration of 1.5–6% in the extraction mixture. LLE (n = 6) 107.9 97.6 95.2
No significant correlation between the bisoprolol peak area and the LLE2 (n = 5) 109.3 96.7 102.5
Within-run precision [%]
surfactant concentration was observed (Fig. 1B). It was observed
CPE (n = 5) 12.36a 3.55 3.98
that the surfactant concentration over 7% resulted in a lower signal LLE (n = 6) 3.62 4.08 2.70
of bisoprolol, probably due to a higher phase volume ratio (vol- LLE2 (n = 5) 8.67 6.42 2.13
ume of the surfactant-rich phase/volume of the aqueous solution) Between-run accuracy [%]
CPE (n = 15) 95.1b 93.1 98.4
resulting in the dilution of the extracted bisoprolol. Thus, based on
LLE (n = 9) 107.1 98.6 93.6
the results of the experiment, 7% of Triton X-114 was selected to LLE2 (n = 5) 104.4 93.3 100.5
achieve the best analytical signals. Between-run precision [%]
CPE (n = 15) 12.60b 6.30 3.80
3.1.3. Effect of the NaOH concentration LLE (n = 9) 3.67 5.03 6.50
LLE2 (n = 5) 10.44 8.24 4.36
The pH optimization is an important step in the CPE method
development. For organic molecules, especially ionizable ones, the CPE and LLE2 was performed in Medical University of Warsaw; LLE was performed
in Pharmaceutical Institute.
maximum extraction efficiency is achieved at those pH values a
n = 6.
where the uncharged form of the analyte prevails and therefore b
n = 16.
the analyte is favored to be partitioned into the micellar phase
[10]. Since bisoprolol is a weak base, the NaOH concentrations were concentration of bisoprolol. The values of regression parameters
tested in the range of 0.05–0.5 M that corresponded to the concen- for the curve, described by the equation: y = ax + b, were calculated
tration of 0.02–0.2 M in the extraction mixture. The best result was as: a = 0.00509 (0.00012), b = 0.00802 (0.00085) and r2 = 0.999. All
obtained when 0.2 M NaOH was added (Fig. 1C), however, the dif- regression parameters were statistically significant (p < 0.05). The
ferences between the analyte signal in 0.2 and 0.5 M NaOH were accuracy for LLOQ was 105% (RSD = 12%, n = 5) within one day and
within the measurement uncertainty. The phenomenon of a lower 107% (RSD = 9.8%, n = 15) within three runs. The accuracy for the QC
bisoprolol signal in higher NaOH concentrations might be caused samples (Table 2) within one day ranged 89–98% (RSD 3.6–13%)
by the co-extraction of the interfering compounds at a higher pH. while between-runs ranged 93–98% (RSD 3.8–13%). The carry-over
experiment in which blank human plasma samples were analyzed
3.2. Influence of the surfactants and selection of internal standard after the highest calibration standards did not show any signals
influencing quantification.
It is well known that the surfactants might seriously influence The recovery for bisoprolol in CPE was moderate and was
the accuracy and precision of bioanalytical methods. One of the rea- 46% (RSD = 12%) for high concentrations (60 ng mL−1 ) and 61%
sons might be difficulty of pipetting viscous liquids which results (RSD = 14%) for low concentrations (0.9 ng mL−1 ). The correspond-
in a relatively high uncertainty of the extraction yield of the com- ing recovery for LLE was 86% for high concentrations (60 ng mL−1 )
pound. Too low volume of the surfactant-rich phase might also and 74% for low concentrations (0.9 ng mL−1 ). In CPE the extraction
result in an unsatisfactory accuracy of the analytical method. This efficiency can be improved using salts like NaCl or Na2 SO4. These
can be overcome by the use of an internal standard. However, in salts are not compatible with LC–MS/MS technique and may cause
the case of methods coupling CPE and mass spectrometry there high matrix effect. Moreover, nonvolatile salts can deposit in source
is a problem of the surfactant’s negative effect on such ionization and plug capillaries thus requiring more cleaning and maintenance
types as MALDI or ESI [41]. Therefore, in CPE a very careful eval- operations. Alternatively violate salts like ammonium acetate or
uation of matrix effects should be an essential part of the method formate can be used, when low recoveries result in failing the vali-
validation. In order to reduce the matrix effect the samples were dation requirements for LLOQ. However, these salts has never been
diluted with acetonitrile and water after the extraction and the gra- used previously in CPE.
dient chromatographic separation was developed. No changes in During the evaluation of the matrix effect by the post-column
the quality of MS spectra nor in chromatograms after CPE exper- infusion of the analyte, the injection of the extracted matrix
iments were observed. There was no indication for additional ESI resulted in similar changes of the signal for both bisoprolol and
source cleaning. metoprolol (Fig. 2). Moreover, near the retention time of the com-
Generally, isotope-labeled compounds are recommended for pounds the signal was barely affected by the matrix and rather sta-
the LC–MS bioanalytical methods as internal standards to reduce ble. The observation was confirmed by the quantitative evaluation
matrix effects and increase the reliability of the results [37]. For this of the matrix effect studied in the bisoprolol concentrations of 1.0
study a compound of similar structure, i.e. metoprolol, was selected and 50 ng mL−1 using the calculation method by Matuszewski et al..
as the internal standard. This internal standard was proved to be an The absolute matrix effect for bisoprolol and the IS were 98% and
appropriate standard both in the SPE [27] and LLE extraction [22,34] 101% for the lower concentration and 105% and 108% for the higher
of bisoprolol from plasma. The structural analogue was selected for concentration, respectively. The relative matrix effect was assessed
this study to assess the compliance of CPE and LC–MS in more dif- as 11.5% for 1.0 ng mL−1 and 4.4% for 50 ng mL−1 , respectively.
ficult conditions than when the isotope labeled internal standard During the experiment we have also tested the influence of
is applied. NaOH and Triton concentrations on the absolute matrix effect. The
absolute matrix effect varied from 97% to 112% (RSD = 4.3%), but
3.3. Method validation no significant trends were observed. The phenomenon might be
a results of adequate chromatographic separation of analytes and
All validation experiments met the acceptance criteria. The inferences like, e.g., Triton. Thus, in CPE this aspect of analysis is
calibration curve obtained by the weighted linear regression probably more important than in other extraction techniques. In
analysis was linear in the range of 0.3–70 ng mL−1 regarding further study we aim to explore the effect of pH and chromato-
the peak area ratio of bisoprolol and the IS versus the nominal graphic separation on the absolute matrix effect in CPE.
 J. Giebułtowicz et al. / J. Chromatogr. A 1423 (2015) 39–46 43

XIC of +MRM (7 pairs): 268.133/116.100 Da ID: metoprolol 8 from Sample 3 (matryca i 1' 60ng/ml bisopr i metoprolol) of 2014-04-30bisoprolol i m... Max. 3341.7 cps.

18.98
3342

3000 OH 12.19 13.87

H 17.93 18.59
O N
2500
+ 11.90
H
IS
Intensity, cps

2000 11.65 17.66

O 15.02
1500 11.26
16.98

1000 8.68 10.72

3.33 4.07 8.22 8.95
0.92 1.17 2.14 2.33 3.19 4.81 5.10
6.59 7.11 7.30
500

0
1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 11.0 12.0 13.0 14.0 15.0 16.0 17.0 18.0 19.0
Time, min
XIC of +MRM (7 pairs): 326.259/116.000 Da ID: bisoprolol 1 from Sample 3 (matryca i 1' 60ng/ml bisopr i metoprolol) of 2014-04-30bisoprolol i met... Max. 4.8e4 cps.

OH 13.87 19.24
4.8e4 H 12.84 18.98
4.5e4 O N
18.19
4.0e4 11.77
+
3.5e4 H

3.0e4
O
Bisoprolol 11.32
17.25
Intensity, cps

15.05
2.5e4 O 11.13 16.99

2.0e4 9.13 9.49

2.33 3.33 4.26 4.42 6.97 7.26 8.24 8.91
1.5e4 0.26 1.28 2.13 3.19
1.0e4

5000.0

0.0
1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 11.0 12.0 13.0 14.0 15.0 16.0 17.0 18.0 19.0
Time, min

Fig. 2. Chromatograms recorded during the evaluation of the matrix effect by the steady post-column infusion and the injection of the extracted blank plasma sample for
metoprolol (internal standard, top) and bisoprolol (bottom).

These results indicate that CPE is a suitable technique for sample results. The statistical evaluation by the F-test indicated that the
preparation prior to the LC–ESI–MS/MS analysis as no significant hypothesis of the equality of variances was rejected for the low QC
matrix effect was observed. However, the chromatographic sep- (p = 0.003), as opposed to the medium and high QCs (p = 0.647 and
aration is probably crucial to reduce the matrix effect caused by p = 0.106, respectively). Therefore, one can conclude that for lower
the surfactant. Similar results of a negligible matrix effect in the concentrations, around 0.9 ng mL−1 , the precision of a bioanalytical
CPE–LC–MS method were observed by Liu et al. [3] using the atmo- method is significantly lower when CPE is applied instead of LLE.
spheric pressure chemical ionization (APCI) source. The results of To confirm that hypothesis we performed additional analysis
the validation also revealed that the structural analogue may be an using LLE method in the same laboratory, where CPE experiments
appropriate internal standard when CPE is used as the extraction were conducted. Due to instrumental differences we modified LLE
technique. method (LLE2) as described in Section 2.5.2. The results confirmed
our previous observation that in lower concentration (low QC) CPE
has significantly lower precision than LLE2 method (p = 0.042).
3.4. Comparison of the CPE and LLE methods

Surprisingly, there were only two reports in literature regarding

3.4.2. Application of the CPE method to clinical samples
the comparison of the CPE and LLE methods for the determination of
The final assessment of the compatibility of CPE and
drugs in biological matrices. However, the comparisons were based
LC–ESI–MS/MS was performed using 28 clinical plasma sam-
on the presentation of a chromatogram only in Zhou et al. [40] or
ples selected from different subjects and representing both high
in the brief description (including linearity, accuracy, precision and
(around maximum plasma concentration Cmax ) and low (in the
recovery) by Zhang et al. [11]. An extensive search did not find any
elimination phase) bisoprolol concentrations. Representative chro-
statistical comparisons of both methods; the lack of such analyses
matogram is presented in Fig. 3. A strong linear correlation of the
is contrary to the current trends in chemical metrology. To fill this
LLE and CPE data is evidenced by the Pearson’s correlation coef-
niche, a cross-validation of the bioanalytical methods using CPE and
ficient of 0.9802 (n = 28, SD = 0.0076) as presented in Fig. 4. The
LLE, supported by a detailed statistical evaluation, was performed
compatibility of CPE and LLE was further supported by the fact
[37,42,43]. The difference in the sample preparation techniques
that only 2 of 28 (7%) observations were located outside the 95%
was assumed to be the main source of variability. Other factors pos-
prediction limits.
sibly influencing the method’s performance, such as using different
A more detailed evaluation was based on the approach similar to
laboratories and analysts, different sample preparation techniques
the incurred sample reanalysis. The LLE results were considered as
as well as different LC–MS/MS methods and instruments, were con-
initial values, while the CPE results were considered as repeat val-
sidered less significant. The original data from the bioanalytical
ues. The acceptance criteria of 67% of the CPE results within ±20%
method full validation and pharmacokinetic study obtained pre-
of the mean were defined according to the cross validation part
viously in a GLP certified laboratory for the regulatory purposes
of the EMA guideline [37]. It was observed that 75% of 28 results
were used as reference [34].
met the acceptance criteria, however, the majority of the results
with % difference exceeding ±20% were found among the lower
3.4.1. Statistical evaluation of the validation parameters concentration (Table 3, Fig. 5). When data was limited to the LLE
Both methods met the acceptance criteria for accuracy and pre- results with the measured bisoprolol concentration not lower than
cision at all QC concentration levels (Table 2), confirming that 1.0 ng mL−1 , the statistics were far better: 91% of 22 results met the
both sample preparation techniques are capable to deliver reliable acceptance criteria, the mean %difference was lower and the SD of
44 J. Giebułtowicz et al. / J. Chromatogr. A 1423 (2015) 39–46

XIC of +MRM (7 pairs): 326.259/116.000 Da ID: bisoprolol 1 from Sample 16 (18/15) of 2014-04-17bisoprolol pacjencji.wiff (Turbo Spray) Max. 2016.7 cps.

8.36
2000

1900 Bisoprolol
1800

1700

1600

1500

1400

1300

1200
In te ns ity , c p s

1100

1000

900

800

700

600

500

400

300

200

100

0
1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 11.0 12.0 13.0 14.0 15.0 16.0 17.0 18.0 19.0
Time, min

XIC of +MRM (7 pairs): 268.133/116.100 Da ID: metoprolol 8 from Sample 16 (18/15) of 2014-04-17bisoprolol pacjencji.wiff (Turbo Spray) Max. 7.6e4 cps.

7.67
7.5e4

7.0e4 IS
6.5e4

6.0e4

5.5e4

5.0e4

4.5e4
In te ns ity , c p s

4.0e4

3.5e4

3.0e4

2.5e4

2.0e4

1.5e4

1.0e4

5000.0

0.0
1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 11.0 12.0 13.0 14.0 15.0 16.0 17.0 18.0 19.0
Time, min

Fig. 3. Chromatogram of sample obtained from healthy volunteer after 5 h of oral administration of bisoprolol fumarate. The determined concentration was 2.4 ng mL−1 . The
blue line represents the main ion transitions m/z 326 → 116 and m/z 268 → 116 for bisoprolol (top) and metoprolol (internal standard, bottom). The red line represents the
second ion transition specific m/z 326 → 222 and m/z 268 → 133 for bisoprolol and metoprolol, respectively.
 J. Giebułtowicz et al. / J. Chromatogr. A 1423 (2015) 39–46 45

60 60%
y = 0.9646x + 1.2289
R² = 0.9608 40%
50

% difference
CPE result [ng/ml]

20%
40
0%
30
-20%
0 10 20 30 40 50 60
20 LLE result [ng/ml]

Fig. 5. The %difference, i.e. (CPE result—average result)/average result × 100%, plot-
10 ted against bisoprolol concentrations measured using LLE.

0 The observations from the clinical samples analysis were in

0 10 20 30 40 50 60 line with the result of the statistical evaluation of the meth-
LLE result [ng/ml] ods’ precision. Therefore, it was concluded that the determination
of bisoprolol concentration in human plasma in the range
Fig. 4. Correlation of bisoprolol concentrations in clinical samples measured using 1.0–70 ng mL−1 by the CPE method leads to the results which are
CPE and LLE (n = 28). equivalent to those obtained by the widely used LLE method.

4. Conclusion
%difference decreased below 15%. The positive value of the mean
%difference may indicate that the application of CPE may lead to
It has been experimentally proved that cloud-point extraction
higher concentrations measured than in the case of LLE; however,
is compatible with LC–ESI–MS/MS. This novel sample prepara-
this difference seems insignificant, as in both data sets, i.e. n = 28
tion technique was successfully applied for the first time for the
and n = 22, the mean %difference was below 15%.
bisoprolol determination in human plasma. The bisoprolol plasma
concentrations in the clinical samples obtained by the CPE and LLE
methods were comparable in the range 1.0–70 ng mL−1 . The results
Table 3
Comparison of LLE and CPE results for clinical plasma samples.
indicate that further research on combining CPE and LC–MS is rec-
ommended and may lead to significant advances in the bioanalysis
Sample no. LLE result CPE result %Difference of drugs. The study provides the first report on the statistical com-
[ng/ml] [ng/ml]
parison of CPE and the conventional extraction method. CPE offers
1 1.51 2.60 27% significant practical advantages and can have a positive impact on
2 0.30 0.54 28%
the environment, therefore its wider application in future pharma-
3 0.30 0.56 30%
4 38.82 45.10 7% cokinetic studies is justifiable.
5 33.71 42.90 12%
6 33.71 36.10 3%
Conflict of interest statement
7 30.04 30.60 1%
8 3.26 4.14 12%
9 1.96 2.14 4% The authors declare no conflict of interest regarding the content
10 0.80 2.00 43% of this article.
11 24.31 23.10 −3%
12 36.44 36.30 0%
13 53.07 47.40 −6% Acknowledgments
14 43.43 47.50 4%
15 2.49 4.60 30% The authors are grateful to all investigators of the former biso-
16 6.90 7.02 1%
17 2.60 2.37 −5%
prolol bioequivalence study [34] for kindly providing their clinical
18 0.70 2.13 51% samples for the additional analysis.
19 46.44 42.60 −4%
20 42.73 47.50 5%
21 44.71 39.20 −7%
References
22 2.97 3.18 3%
23 0.89 1.30 19% [1] K. Madej, K. Persona, M. Wandas, E. Gomółka, Sequential cloud-point extraction
24 0.35 0.59 26% for toxicological screening analysis of medicaments in human plasma by high
pressure liquid chromatography with diode array detector, J. Chromatogr. A
25 13.34 15.10 6%
1312 (2013) 42–48.
26 28.11 33.00 8%
[2] K. Madej, K. Persona, Drug screening in human plasma by cloud-point extrac-
27 44.20 33.80 −13%
tion and HPLC, Cent. Eur. J. Chem. 11 (2013) 94–100.
28 37.95 36.70 −2% [3] W. Liu, K. Bi, X. Liu, J. Zhao, X. Chen, Cloud-point extraction combined with
All data n= 28 LC–MS for analysis of memantine in rat plasma, Chromatographia 69 (2009)
837–842.
% of samples meeting acceptance criteria 75%
[4] N. Gao, H. Wu, Y. Chang, X. Guo, L. Zhang, L. Du, Y. Fu, Mixed micelle cloud
mean difference 10%
point-magnetic dispersive ␮-solid phase extraction of doxazosin and alfuzosin,
SD of difference 16%
Spectrochim. Acta A: Mol. Biomol. Spectrosc. 134 (2015) 10–16.
CPE result over 1 ng/ml n= 22 [5] R. Heydari, N.S. Elyasi, Ion-pair cloud-point extraction: a new method for the
% of samples meeting acceptance criteria 91% determination of water-soluble vitamins in plasma and urine, J. Sep. Sci. 37
(2014) 2724–2731.
Mean difference 4%
[6] F. Nazari Serenjeh, P. Hashemi, M. Safdarian, Z. Kheirollahi, Semi-automated
SD of difference 10%
cloud point extraction with cold column trapping of surfactant-rich phase for
46 J. Giebułtowicz et al. / J. Chromatogr. A 1423 (2015) 39–46
phenazopyridine determination in human serum, J. Iran. Chem. Soc. 11 (2014)
733–739.
[7] Z. Pourghobadi, R. Heydari, R. Pourghobadi, M. Rashidipour, Determination of
gabapentin in human plasma using simultaneous cloud point extraction and
precolumn derivatization by HPLC, Monatsh. Chem. Chem. Mon. 144 (2013)
773–779.
[8] H. Ma, J. You, Y. Liu, Cloud-point extraction combined with HPLC for deter-
mination of larotaxel in rat plasma: a pharmacokinetic study of liposome
formulation, J. Sep. Sci. 35 (2012) 1539–1546.
[9] W.-J. Zhao, W. Liu, J.-B. Chen, Z.-M. Zhou, M.-M. Yang, Use of cloud point extrac-
tion with derivatizing reagent for the extraction and determination of isoniazid,
J. Chromatogr. Sci. 49 (2011) 154–158.
[10] F. Han, R. Yin, X.-L. Shi, Q. Jia, H.-Z. Liu, H.-M. Yao, L. Xu, S.-M. Li, Cloud point
extraction-HPLC method for determination and pharmacokinetic study of flur-
biprofen in rat plasma after oral and transdermal administration, J. Chromatogr.
B 868 (2008) 64–69.
[11] H. Zhang, H.-K. Choi, Analysis of meloxicam by high-performance liquid chro-
matography with cloud-point extraction, Anal. Bioanal. Chem. 392 (2008)
947–953.
[12] X. Liu, X.-H. Chen, Y.-Y. Zhang, W.-T. Liu, K.-S. Bi, Determination of arbidol in rat
plasma by HPLC–UV using cloud-point extraction, J. Chromatogr. B 856 (2007)
273–277.
[13] M.D. Rukhadze, S.K. Tsagareli, N.S. Sidamonidze, V.R. Meyer, Cloud-point
extraction for the determination of the free fraction of antiepileptic drugs in
blood plasma and saliva, Anal. Biochem. 287 (2000) 279–283.
[14] A.B. Tabrizi, S. Naini, K. Parnian, S. Mohammadi, F.E. zad, S.P. Anvarian, A. Abdol-
lahi, Determination of triamterene in human plasma and urine after its cloud
point extraction, Quím. Nova 37 (2014) 1182–1187.
[15] H. Wu, G.-Y. Zhao, L.-M. Du, Determination of ofloxacin and gatifloxacin
by mixed micelle-mediated cloud point extraction-fluorimetry combined
methodology, Spectrochim. Acta, A: Mol. Biomol. Spectrosc. 75 (2010)
1624–1628.
[16] X.Y. Qin, J. Meng, X.Y. Li, J. Zhou, X.L. Sun, A.D. Wen, Determination of ven-
lafaxine in human plasma by high-performance liquid chromatography using
cloud-point extraction and spectrofluorimetric detection, J. Chromatogr. B 872
(2008) 38–42.
[17] B.K. Matuszewski, M.L. Constanzer, C.M. Chavez-Eng, Strategies for the
assessment of matrix effect in quantitative bioanalytical methods based on
HPLC−MS/MS, Anal. Chem. 75 (2003) 3019–3030.
[18] G.-F. Li, K. Wang, R. Chen, H.-R. Zhao, J. Yang, Q.-S. Zheng, Simulation of the phar-
macokinetics of bisoprolol in healthy adults and patients with impaired renal
function using whole-body physiologically based pharmacokinetic modeling,
Acta Pharmacol. Sin. 33 (2012) 1359–1371.
[19] H. Chang, J. Li, J. Li, X. Guan, F. Sun, Z. Qian, K. Bi, G. Fan, Simulta-
neous determination of amlodipine and bisoprolol in rat plasma by a liquid
chromatography/tandem mass spectrometry method and its application in
pharmacokinetic study, J. Pharm. Biomed. Anal. 71 (2012) 104–110.
[20] G.-Y. Liu, W. Wang, J.-Y. Jia, C. Lu, Y.-M. Liu, M.-Q. Zhang, Y. Liu, S.-J. Li, C. Yu,
Liquid chromatography tandem mass spectrometry method for determination
of bisoprolol in human plasma using d5-bisoprolol as the internal standard,
Biomed. Chromatogr. 24 (2010) 574–580.
[21] O. Gonzalez, G. Iriarte, E. Rico, N. Ferreirós, M.I. Maguregui, R.M. Alonso, R.M.
Jiménez, LC–MS/MS method for the determination of several drugs used in
combined cardiovascular therapy in human plasma, J. Chromatogr. B 878
(2010) 2685–2692.
[22] G. Peste, N. Bibire, M. Apostu, A. Vlase, C. Oniscu, A New Liquid
Chromatography-Tandem Mass Spectrometry Method for Determination of
Bisoprolol in Human Plasma Samples, J. Biomed. Biotechnol. 2009 (2009),
http://dx.doi.org/10.1155/2009/736327, Article ID 736327, 8 pages.
[23] G. Peste, C. Oniscu, A. Vlase, Experimental research for determination of
bisoprolol fumarate in human plasma samples using liquid chromatography-
tandem mass spectrometry (LC–MS/MS) technique, Rom. Biotechnol. Lett. 15
(2010) 5141–5145.
[24] M.F. Tutunji, H.M. Ibrahim, M.H. Khabbas, L.F. Tutunji, Simultaneous determina-
tion of bisoprolol and hydrochlorothiazide in human plasma by HPLC coupled
with tandem mass spectrometry, J. Chromatogr. B 877 (2009) 1689–1697.
[25] M. Liu, D. Zhang, Y. Sun, Y. Wang, Z. Liu, J. Gu, Liquid
chromatographic–electrospray tandem mass spectrometric determination of
bisoprolol in human plasma, Biomed. Chromatogr. 21 (2007) 508–513.
[26] S. Li, G. Liu, J. Jia, Y. Liu, C. Pan, C. Yu, Y. Cai, J. Ren, Simultaneous determina-
tion of ten antiarrhythic drugs and a metabolite in human plasma by liquid
chromatography—tandem mass spectrometry, J. Chromatogr. B 847 (2007)
174–181.
[27] J. Bhatt, G. Subbaiah, S. Kambli, B. Shah, M. Patel, A. Saxena, A. Baliga,
S. Nigam, H. Parekh, G. Yadav, A high throughput and sensitive liquid
chromatography–tandem mass spectrometry (LC–MS/MS) method for the
estimation of bisoprolol in human plasma using multiplexing technique, J.
Chromatogr. B 852 (2007) 374–381.
[28] M. Josefsson, A. Sabanovic, Sample preparation on polymeric solid phase
extraction sorbents for liquid chromatographic-tandem mass spectrometric
analysis of human whole blood—a study on a number of beta-agonists and
beta-antagonists, J. Chromatogr. A 1120 (2006) 1–12.
[29] J. Musijowski, M. Filist, P.J. Rudzki, Sensitive single quadrupole LC/MS method
for determination of lapatinib in human plasma, Acta Pol. Pharm. 71 (2014)
p7.
˛
[30] J. Giebułtowicz, G. Nałecz-Jawecki, Occurrence of antidepressant residues in
the sewage-impacted Vistula and Utrata rivers and in tap water in Warsaw
(Poland), Ecotoxicol. Environ. Safe. 104 (2014) 103–109.
[31] J. Musijowski, E. Piórkowska, P.J. Rudzki, Determination of sunitinib in human
plasma using liquid chromatography coupled with mass spectrometry, J. Sep.
Sci. 37 (2014) 2652–2658.
[32] M. Filist, K. Buś-Kwaśnik, H. Ksycińska, P.J. Rudzki, Simplified LC–MS/MS
method enabling the determination of azithromycin in human plasma after
a low 100 mg dose administration, J. Pharm. Biomed. Anal. 100 (2014)
184–189.
[33] H. Filik, S.D. Çekiç, Cloud point extraction of pesticide residues, in: M.
Stoytcheva (Ed.), Pesticides in the Modern World—Trends in Pesticides Analy-
sis, InTech, Rijeka, Croatia, 2011.
[34] K. Buś-Kwaśnik, H. Ksycińska, A. Leś, K. Serafin-Byczak, P.J. Rudzki, J. Raszek, T.
Łazowski, A. Bielak, A. Wybraniec, Bioequivalence and pharmacokinetics of two
10-mg bisoprolol formulations as film-coated tablets in healthy white volun-
teers: a randomized, crossover, open-label, 2-period, single-dose, fasting study,
Int. J. Clin. Pharmacol. Ther. 50 (2012) 909–919.
[35] L. Ding, X. Zhou, X. Guo, Q. Song, J. He, G. Xu, LC–ESI–MS method for the deter-
mination of bisoprolol in human plasma, J. Pharm. Biomed. Anal. 44 (2007)
520–525.
[36] FDA Guidance for Industry, Center for Drug Evaluation and Research
(CDER).Bioanalytical Method Validation, Food and Drug Administration,
Rockville, MD,U.S, 2001.
[37] Guideline on bioanalytical method validation, EMEA/CHMP/EWP/192217/
2009, European Medicines Agency. Committee for Medicinal Products for
Human Use (CHMP), 2011.
[38] Z.-M. Zhou, D.-Y. Zhao, J. Wang, W.-J. Zhao, M.-M. Yang, Study of cloud point
extraction and high-performance liquid chromatographic determination of iso-
niazid based on the formation of isonicotinylhydrazone, J. Chromatogr. A 1216
(2009) 30–35.
[39] G. Ren, Q. Huang, J. Wu, J. Yuan, G. Yang, Z. Yan, S. Yao, Cloud point
extraction—HPLC method for the determination and pharmacokinetic study
of aristolochic acids in rat plasma after oral administration of Aristolochiae
Fructus, J. Chromatogr. B 953–954 (2014) 73–79.
[40] J. Zhou, P. Zeng, H.H. Tu, F.Q. Wang, Development and application of
high-performance liquid chromatography for the study of ampelopsin phar-
macokinetics in rat plasma using cloud-point extraction, J. Sep. Sci. 34 (2011)
160–168.
[41] A. 43 Nordström, J.V. Apon, W. Uritboonthai, E.P. Go, G. Siuzdak, Surfactant-
enhanced desorption/ionization on silicon mass spectrometry, Anal. Chem. 78
(2005) 272–278.
[42] D. Gansser, Experience with a cross-validation approach, Chromatographia 55
(2002) S71–S74.
[43] M.T. Gilbert, I. Barinov-Colligon, J.R. Miksic, Cross-validation of bioana-
lytical methods between laboratories, J. Pharm. Biomed. Anal. 13 (1995)
385–394.
 Talanta 83 (2011) 980–987

Contents lists available at ScienceDirect

Talanta
journal homepage: www.elsevier.com/locate/talanta

Tyrosinase immobilized magnetic nanobeads for the amperometric assay of

enzyme inhibitors: Application to the skin whitening agents
Veronica Harceaga Sima a , Stephanie Patris b , Zeynep Aydogmus c , Ahmad Sarakbi b ,
Robert Sandulescu a , Jean-Michel Kauffmann b,∗
a
University of Medicine and Pharmacy, Faculty of Pharmacy, 400012 Cluj-Napoca, Romania
b
Université Libre de Bruxelles (ULB), Faculty of Pharmacy, 1050 Brussels, Belgium
c
Istanbul University1 , Faculty of Pharmacy, 34116 Istanbul, Turkey

a r t i c l e i n f o a b s t r a c t

Article history: The immobilization of tyrosinase onto glutaraldehyde activated streptavidine magnetic particles and
Received 22 September 2010 subsequent retention onto a magnetized carbon paste electrode for the amperometric assay of tyrosi-
Received in revised form 26 October 2010 nase inhibitors is described. Tyrosine was used as substrate as it is the first substrate in the melanogenesis
Accepted 1 November 2010
process. The sensing mode is based on monitoring the decrease of the amperometric signal correspond-
Available online 10 November 2010
ing to the electrochemical reduction of dopaquinone enzymatically generated. This current decrease is
due to the presence of inhibitors acting directly on the enzyme or inhibitors acting on the product of the
Keywords:
enzymatic reaction, i.e. dopaquinone. The methodology is designed for the evaluation of the inhibitory
Magnetic nanoparticles
Tyrosinase
potency of the most frequently used active substances in cosmetic marketed products against hyper-
Skin whitening agent pigmentation such as kojic acid, azelaic acid and benzoic acid. These compounds bind to the tyrosinase
Biosensor active center. Ascorbic acid is also investigated as it interrupts the synthesis pathway of melanin by reduc-
ing the melanin intermediate dopaquinone back to l-dopa. By comparing the obtained IC50 , under the
same experimental conditions, the order of their inhibitory potency was: kojic acid (IC50 = 3.7 × 10−6 M,
Ki = 8.6 × 10−7 M), ascorbic acid (IC50 = 1.2 × 10−5 M), benzoic acid (IC50 = 7.2 × 10−5 M, Ki = 2.0 × 10−5 M)
and azelaic acid (IC50 = 1.3 × 10−4 M, Ki = 4.2 × 10−5 M) in close agreement with literature spectrophoto-
metric inhibition data using the soluble tyrosinase.
© 2010 Elsevier B.V. All rights reserved.

1. Introduction Hyperpigmentation is a complex process. There are many fac-

Melanin, the dark pigment produced by skin cells in the most
inner layer of the epidermis, plays an important role in protect-
ing human skin from the harmful effects of the UV sun radiations.
Melanin also determines our phenotypic appearance. However,
abnormal accumulation of melanin in the basal layer of the dermis
is responsible for hyperpigmentation including melasma, freckles
and senile lentigines [1]. These undesirable conditions of the human
skin are serious esthetic problem for human beings nowadays since
both appearance and quality of life are becoming more and more
important. This problem is particularly prevalent in middle aged
and elderly individuals [2] and this has encouraged researchers to
seek for new potent inhibitors of the hyperpigmentation process.
Recently, a global market demand has developed on this subject.
For the past few decades, tyrosinase inhibitors have been a great
concern solely due to the key role of tyrosinase in mammalian
melanogenesis [3].
∗ Corresponding author.
E-mail address: jmkauf@ulb.ac.be (J.-M. Kauffmann).
1
On leave.
tors that influence the activity of melanocites, but it is well known
and unanimously accepted that tyrosinase activity plays the main
role in the pathway of melanin biosynthesis. It catalyzes the first
two reactions of the melanogenesis process namely: the hydroxy-
lation of l-tyrosine to l-dopa as monophenolase, using molecular
oxygen as co-substrate, and the oxidation of l-dopa to dopaquinone
as diphenolase. These steps, and mostly the first oxidizing step, are
the rate-limiting steps in melanin synthesis since the subsequent
reactions can proceed spontaneously at physiological pH [3].
A number of tyrosinase and melanogenesis inhibitors, from both
natural and synthetic sources, have been identified [4]. However,
the definition of “tyrosinase inhibitor” is sometimes misleading, the
terminology being also used to refer to melanogenesis inhibitors,
whose action mainly reside in some interference in melanin forma-
tion, regardless of any direct inhibitor–enzyme interaction [3]. Only
specific tyrosinase inactivators and inhibitors should be regarded
as “true inhibitors”, which bind to the enzyme and inhibit its activ-
ity. Some skin whitening agents that act as tyrosinase inhibitors,
such as kojic acid and azelaic acid, or as melanogenesis inhibitors
such as hydroquinone, arbutin and ascorbic acid are the most
well known to most dermatologists [5]. The interference with
melanin synthesis can occur in different ways. Kojic acid is a

0039-9140/$ – see front matter © 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.talanta.2010.11.005
 V.H. Sima et al. / Talanta 83 (2011) 980–987
chelator of the copper of the enzyme active site [6], arbutin is a
pro-drug of hydroquinone that is an alternative substrate, azelaic
acid blocks the tyrosine access to the active site and ascorbic acid
chemically reduces the dopaquinone back to l-dopa, thus avoid-
ing dopachrome and melanin formation [7]. Also benzoic acid, a
preservative very frequently used in cosmetic and food industry [8],
has been demonstrated to have inhibitory properties for tyrosinase
[9].
The need for the quantification and comparative analysis at high
throughput of the inhibitory activity of possible new products and
of the existing ones is becoming more and more of interest for the
evaluation and development of efficient inhibitors of the hyperpig-
mentation process.
Until now, several methods for the screening of skin-whitening
agents have been proposed. The method most frequently applied
exploited the spectrophotometric quantification of dopaquinone,
before and after the inhibition of soluble tyrosinase [6,10]. Another
method with tyrosinase in solution relied on electron spin reso-
nance spectrometry that measured the accumulation of eumelanin
radicals in melanin culture cells [11]. In vivo studies evaluated the
inhibitors potency by photometry [12]. Usually in each of the cur-
rently developed method, the inhibitory strength of the studied
compounds was compared to that of a standard inhibitor, namely
kojic acid [6]. Spectrophotometry is not a primary choice analyti-
cal method for the quantification of an on going high ratio reaction
which is the case of the enzymatic formation of dopaquinone due
to the low reproducibility of the obtained results. The use of spec-
trometry has the disadvantage of the need of culture cells and the
photometric one is applied to in vivo studies so it can be used just for
substances that have been approved for human use and the imple-
mentation of such study involves ethic problems. Thus they are not
suitable for rapid assays in scientific research of novel potential skin
whitening agents. The use of biosensors would be an interesting and
valuable alternative for the rapid screening and comparison of the
inhibitory potency of the existing compounds and the novel ones
due to its relative simplicity in use and the high reproducibility of
the analytical data.
To the best of our knowledge there have been very few tyrosi-
nase based biosensors fabricated for the evaluation of the inhibitory
potency of kojic acid. A Clark-type oxygen electrode that monitored
oxygen consumption during catechol conversion in the presence
of kojic acid inhibitor was reported [13]. Another biosensor for
kojic acid quantification used dopamine as substrate and it mon-
itored the oxygen consumption with a Clark type electrode [14].
There have been a couple of tyrosinase biosensors developed for
the evaluation of the inhibitory potency of benzoic acid [8,15]
but none of them used l-tyrosine as enzyme substrate. Tyrosinase
based biosensors usually used phenol or its derivatives as sub-
strate with the enzyme being combined with an electrochemical
transducer to sense either the oxygen consumption of the overall
enzymatic reaction [14] or the enzymatic production of the elec-
troactive quinone species [16]. There was one article reported of
a tyrosinase based biosensor for quantitative analysis of phenols.
Tyrosinase was immobilized to core–shell magnetic particles and
subsequently attached to the surface of a carbon paste electrode
with the help of a permanent magnet [17].
Tyrosinase is widely distributed in microorganisms, animals,
humans and plants. The enzyme extracted from the mushroom
Agaricus bisporus is highly homologous with the mammalian one
making it well suited as a model for studies on melanogenesis.
Practically all studies of inhibitors conducted so far have used
the mushroom tyrosinase likely because of its readily commercial
availability [3]. Yet alternative systems for testing can be obtained
like mammalian tyrosinase, melanocytic cultures, co-cultures of
keratinocytes and melanocytes, skin culture to finally in vivo appli-
cation to animal skin [5,18].
981
Different phenolic compounds can be enzymatically oxidized
by tyrosinase and several substrates have been proposed in the
literature for the evaluation of the enzyme inhibitors. Dopamine
was used as a substrate in the construction of an amperometric
biosensor for the detection of tyrosinase inhibitors such as kojic
acid, benzoic acid and SCN− ion [14]. Catechol, p-cresol, m-cresol,
phenol and p-chlorophenol were used as enzyme substrates for
the determination of benzoic acid based on its inhibition prop-
erties, using an amperometric biosensor device [19–21]. l-Dopa
was used as enzyme substrate in spectrophotometric evaluation of
skin-whitening agents [10]. Also l-tyrosine was used as a substrate
in spectrophotometric evaluation of different inhibitors [22,23].
Cellular tests with human melanocytes were used for the trans-
formation rate study of the tyrosinase substrate l-dopa using a
spectrophotometric method [24].
Recently, biosensors based on the inhibition of tyrosinase activ-
ity in the presence of a monophenol or o-diphenol substrate have
been proposed for the determination of triazine pesticides [25], car-
bamate and organophosphate pesticides [16], diuron and atrazine
[26], herbicides such as alachlor, diazinon and carbaryl [27], propyl
gallate [28], 2,4-dichlorophenol [29], fluoride [30], benzoic acid
[19], cinnamic acid and sorbic acid [31].
Besides choosing the adequate transducer and enzyme in the
construction of a biosensor, an important role is played by the
enzyme immobilization method. The use of magnetic particles
(beads) in biomedical and pharmaceutical sciences has increased
significantly in recent years [32]. The masterbeads are specially
designed for allowing simple and effective immobilization of lig-
ands such as enzyme through activation with bi-functionalized
reactants [33], like glutaraldehyde. Immobilization of biomolecules
onto the surface of magnetic particles can result in a number of
additional functionalities: on one hand the nanostructure of the
magnetic particles permits a high enzyme loading without affect-
ing its natural structure and consequently its activity and on the
other hand the beads can be attracted by a magnet and be strongly
retained in close proximity to the electrode surface and read-
ily washed away if needed [34]. The streptavidine masterbeads
(NMPS) are mainly designed for immunological analyses [35] but
they can very well be used for enzyme immobilization. Based on
the IUPAC definition [36] such a configuration is strictly speaking
not a biosensor, i.e. a self-contained integrated device, but since it
has the enzyme, during the assays, in close spatial contact with the
electrode we may consider it as a biosensor.
In the present work, a novel method for the evaluation of the
potency of the tyrosinase and melanogenesis inhibitors used as
skin whitening agents is proposed. The biosensor was fabricated
using tyrosinase immobilized via glutaraldehyde activated strep-
tavidine magnetic nanoparticles and retained onto a “magnetized”
carbon paste electrode, mCPE. The interest in using a carbon paste
electrode lies in its low background current, allowing for highly
sensitive assay to be performed in the applied potential range
close to 0.0 V versus an Ag/AgCl reference electrode. The employed
tyrosinase was the mushroom tyrosinase as it is highly homology
with the human enzyme. The substrate used was the l-tyrosine,
the enzyme’s natural substrate in the skin and the investigated
inhibitors were kojic acid, azelaic acid, ascorbic acid, the most
frequently used active substances in marketed products for hyper-
pigmentation, and benzoic acid, a well known preservative in the
cosmetic industry. The obtained configuration was used to moni-
tor the reduction current of the enzymatically generated oxidized
species of l-tyrosine, i.e. the dopaquinone, in the presence of molec-
ular oxygen (Fig. 1). The biosensor was applied to the screening
of inhibitors which induced a decrease of the reduction current
proportional to the inhibitor concentration and strength.
In practice, the steady state current for l-tyrosine was recorded
as I0 . The biosensor response in the case of the addition of an
982 V.H. Sima et al. / Talanta 83 (2011) 980–987
Fig. 1. Biosensor’s mechanism of action.
inhibitor was recorded as I1 . The concentration of inhibitors was
correlated with the percentage of inhibition (In %) which was calcu-
lated using the relationship: In (%) = [(I0 − I1 )/I0 ] × 100. With I0 and
I1 being the biosensor current signals before and after the addi-
tion of the inhibitor, respectively, I1 was expected to be smaller
comparing to I0 as the process was inhibited.
2. Materials and methods
2.1. Material
The mushroom tyrosinase (EC 1.14.18.1) 5370 Units/mg was
purchased from Sigma. l-Tyrosine and l-dopa were provided by
Janssen Chimica. The Masterbeads Streptavidine (500 nm) were
purchased from Ademtech France. Dopamine, glutaraldehyde 25%
wt and monosodium phosphate was provided by Sigma–Aldrich
and disodium phosphate was provided by Merck. The carbon paste
was from Metrohm. Kojic acid was purchased from Kreglinger
Europe, azelaic acid and ascorbic acid from Acros Organics and
benzoic acid from Fluka.
All reagents were of analytical grade and were used as received.
All solutions were prepared with distilled water and stored in the
refrigerator (4 ◦ C) when not in use. The phosphate buffer was pre-
pared by mixing two stock solutions of monosodium phosphate
0.1 M and disodium phosphate 0.1 M to the desired pH. The stock
solutions of l-tyrosine, l-dopa, kojic acid, benzoic acid, azelaic acid
and ascorbic acid were prepared in phosphate buffer.
2.2. Construction of the tyrosinase-NMPS biosensor
2.2.1. Enzyme immobilization
The Streptavidin Masterbeads were monodispersed and were
based on superparamagnetic particles composed of a magnetic core
encapsulated by a hydrophilic polymer shell coated with strepta-
vidin. The magnetic particles had a diameter of 500 nm (RSD max
20%), density of approx. 2.0 g/cm3 , magnetic susceptibility approx.
40 emu/g, specific surface area of 5 m2 /g, iron oxide content approx.
70% m/m and the solid content of 10 mg/mL [33].
An amount of 1 mL 10 mg/mL NMPSs was washed with 0.01 M
phosphate buffer pH 6.5 (5 mL) then reacted with 1 mL of 2.5%
wt glutaraldehyde solution (in 0.01 M phosphate buffer, pH 6.5)
at room temperature for 30 min. After washing with 0.01 M phos-
phate buffer pH 6.5 (5 mL), the glutaraldehyde-treated NMPSs
were reacted with 1 mL tyrosinase solution (3 mg/mL in phosphate
buffer 0.01 M, pH 6.5) during 1 h at room temperature to obtain
the tyrosinase-NMPSs. The resulting suspension was washed with
0.01 M phosphate buffer pH 6.5 (5 mL) and stored in 1 mL 0.01 M
phosphate buffer pH 6.5 at 4 ◦ C. During the washing steps, the
beads were trapped by magnetic forces by placing the reacting
Eppendorf tube close to a strong magnet and the supernatant was
carefully pipetted off. The final concentration of tyrosinase-NMPSs
was 10 mg/mL (referring to the amount of NMPS) [35].
A 10 mg/mL tyrosinase-bead suspension was used for l-
tyrosine response optimization and an approximately 1.25 mg/mL
tyrosinase-bead suspension was used in the inhibition assays.
The initial 1 mL amount of the tyrosinase-NMPSs was diluted and
it resulted finally in 8 mL suspension of 1.25 mg/mL tyrosinase-
NMPSs that allowed performing approximately 800 inhibition
assays.
Visible spectrophotometry was applied for the determination
of the efficiency of the immobilization method of tyrosinase onto
the NMPSs. The method consisted in determining the amount of
immobilized tyrosinase by measuring the absorbance of tyrosinase
in a 60 times diluted tyrosinase solution used in the immobiliza-
tion process, before and after the enzyme was immobilized onto the
activated NMPSs. The percentage of immobilization, calculated by
referring to a calibration curve (absorbance versus tyrosinase con-
centration) realized in 0.01 M phosphate buffer pH 6.5 ( = 280 nm),
was found to be around 40 ± 5%. Therefore it was inferred that
the inhibition assays were realized with approximately 8 Units of
tyrosinase immobilized at the electrode surface.
2.2.2. Biosensor fabrication
The working carbon paste electrode (mCPE) consisted of a home
made electrode: inside a plastic rod, a permanent magnet (Neody,
3 mm diameter) was inserted leaving a depression at the surface of
approximately 2 mm for housing the solid carbon paste layer, and
a copper wire was used as the electrical conductor.
Before each experiment the solid carbon paste was manually
poured in the hole and the resulting mCPE was smoothed on a clean
paper surface. The diameter of the active surface was 3 mm. A vol-
ume of 10 ␮L of the Tyrosinase-NMPSs suspension was spread over
the surface of the electrode (in position up side down) allowing
the particles to settle, being attracted within a few seconds by the
magnet. Subsequently, the tyrosinase-NMPSs immobilized mCPE
was oriented in the right position in a three-electrode cell. The
strong attraction of the Tyrosinase-NMPSs by the magnet housed
inside the working electrode allowed appropriate stirring during
the amperometric experiments.
2.3. Apparatus
The experiments were performed by using as potentiostat a
BASi EPSILON system with a C3 Cell Stand. Amperometry was per-
formed in a conventional three electrodes setup with the biosensor
as working, a Pt wire as auxiliary and an Ag/AgCl 3 M KCl as refer-
ence electrode, respectively. The cyclic voltammetry experiments
were performed in the same conventional three electrodes setup
used in amperometry. All the cyclic voltammetry experiments were
recorded at 100 mV s−1 or 10 mV s−1 in a potential window com-
prised between −0.6 V and +1.1 V versus Ag/AgCl 3 M KCl reference
electrode. The spectrophotometric assays were performed using
an Agilent Hewlett Packard Diode Array UV Vis Spectrophotometer
Model 8453. The pH of the solution was controlled with a Metrohm
827 pH Lab system. All experiments were performed at room tem-
perature (23 ◦ C).
2.4. Amperometric assay
During amperometric experiments the biosensor potential was
kept at −100 mV under continuous stirring conditions with a mag-
netic spinbar at 300 rpm. The working potential was imposed and
the background current was allowed to reach a steady state value
within approx. 10 min. Different amounts of tyrosinase standard
solution or inhibitor solution were added every 100 or 200 s and the
current was recorded as a function of time in 10 mL 0.1 M phosphate
buffer solution.
 V.H. Sima et al. / Talanta 83 (2011) 980–987
2.0E-06
I (A)
1.5E-06
b
1.0E-06
5.0E-07 a ence of l-tyrosine and l-dopa using 10 ␮L of the tyrosinase-NMPS
0.0E+00
-1.0 -0.5 0.0 0.5 1.0 1.5
-5.0E-07
E (V) vs Ag/AgCl
-1.0E-06
-1.5E-06
Fig. 2. Repetitive CV for l-tyrosine 4.76 × 10−5 M, 0.1 M phosphate buffer pH 6.5,
mCPE, scan rate, a: 10 mV s−1 , b: 100 mV s−1 .
After each amperometric inhibition assay, the tyrosinase-
NMPSs were readily removed by flushing a water burst over the
biosensor surface. The surface of the mCPE was renewed after each
inhibition assay.
3. Results and discussions
3.1. Working conditions
The working conditions for an optimal response to the sub-
strate of the tyrosinase based biosensor have been tested. Since
the biosensor was designed for testing inhibitors of the skin pig-
mentation process it was taken into account to operate as close as
possible to the natural environment of the skin tyrosinase which is
a transmembrane protein of melanosomes [37].
3.1.1. Selection of the applied potential
It is well known that l-tyrosine enzymatic oxidation pro-
duces dopaquinone and it was expected that this product could
be detected by electroreduction at the mCPE. Likewise the elec-
trochemical oxidation of l-tyrosine was supposed to generate
dopaquinone. Cyclic voltammetry (CV) performed at 10 and
100 mV/s (from 0.0 mV to 1.1 V, and back to −0.6 V) in 0.1 M phos-
phate buffer pH 6.5, showed the irreversible oxidation peak of
l-tyrosine (1 × 10−3 M and 4.76 × 10−5 M) since no reduction peak
was observed in the investigated potential domain (Fig. 2). Same
experiments were performed also for l-dopa which gave an oxida-
tion peak at 0.550 V but still no reduction peak was observed. The
absence of a reduction current and the progressive decrease of the
l-tyrosine peak during multiple scanning suggested that the gen-
erated dopaquinone escaped rapidly from the electrode solution
interface likely due its high reactivity giving rise to polymer-like
structures fouling the electrode surface [38]. Amperometry of l-
tyrosine (2.44 × 10−5 M) at the tyrosinase-NMPS mCPE, thanks to
the low background current compared to CV, permitted, however,
to detect a reduction current at potentials near 0 V versus Ag/AgCl
likely corresponding to the electroreduction of some enzymatically
generated dopaquinone remaining at the electrode solution inter-
face. Amperometric experiments were realized at 0, −50, −100,
−150, −200 and −250 mV. By decreasing the applied potential the
reduction current increased but the response steady-state became
less stable. An applied potential of −100 mV was considered opti-
mal taking into account the magnitude of the reduction current,
the ratio between signal and background current, the steady-state
of the response plateau and the possible interfering species at the
applied potential, in agreement with other tyrosinase based biosen-
sors described in the literature [21].
3.1.2. Selection of the substrate
To the best of our knowledge there is no tyrosinase based amper-
ometric biosensor for the evaluation of skin-whitening agents that
983
used l-tyrosine as a substrate, reported in the literature. Moreover
it has been demonstrated that the intensity of inhibition varied
considerably depending on the characteristics of the substrate [19].
The biosensor response was tested comparatively both in the pres-
suspension (10 mg/mL) deposited onto the surface of the mCPE. It
was observed that, in the concentration range 4.98 × 10−6 M and
2.91 × 10−5 M, the sensitivity to l-dopa (−5.56 × 10−4 A/M) was
almost 6 times higher comparing to l-tyrosine (−9.37 × 10−5 A/M)
and the response time (the time necessary for reaching 95% of the
maximum response) to l-dopa (56 s) was faster comparing to l-
tyrosine (86 s). l-Tyrosine, however, was chosen as model substrate
for subsequent experiments. This decision was based on the fact
that, in vivo, l-tyrosine is the natural substrate of tyrosinase and
also the starting point of melanin synthesis and its oxidation, to
l-dopa, is the rate-limiting step of the whole process [3]. The sensi-
tivity to l-tyrosine was considered high enough for assuring reliable
results and the lag period of its first enzymatic oxidizing step to l-
dopa was sufficiently short for further readily implementation of
the testing conditions.
3.1.3. Selection of the pH
The biosensor response was studied in the pH range comprised
between 5.5 and 7. It is well known that enzyme activity is highly pH
dependent and that the optimum pH for an enzymatic assay must
be determined empirically [39]. By decreasing the pH from 7.0 to
5.5, the amperometric signal increased but no stable steady-state
plateau was obtained at pH 6.0 and 5.5. For further experiments,
the pH chosen was 6.5. At this pH, the current and the steady-state
of the biosensor response were considered optimal, in agreement
with different literature data where the working pH of the tyrosi-
nase based assays vary between 6.0 and 7.0. [9,14,22,23]. The
selected pH corresponded to the Sigma declared tyrosinase stabil-
ity pH and is also close to the normal pH range of the skin which is
considered to vary between 4 and 6.5.
3.1.4. Amount of tyrosinase-NMPS spiked onto the mCPE
The amperometric response was determined as a function of l-
tyrosine, in the concentration range 4.98 × 10−6 and 2.91 × 10−5 M,
using different amounts (5 ␮L, 10 ␮L and 20 ␮L) of the 10 mg/mL
tyrosinase-NMPS suspension spiked onto the surface of the mCPE.
The response increased proportionally by raising the amount of
immobilized enzyme onto the surface of the electrode. The sen-
sitivities obtained were: −6.82 × 10−5 A/M, −9.37 × 10−4 A/M
and −1.68 × 10−4 A/M for 5, 10 and 20 ␮L spiked suspension,
respectively. The repeatability of 5 ␮L deposited beads was poor
comparing to 20 ␮L and 10 ␮L deposits. The latter was selected
in subsequent work since it offered good sensitivity and repeata-
bility (RSD = 2.8%, n = 3) and it consumed less beads then a 20 ␮L
deposit. For the subsequent inhibition assays it was decided to use
a diluted suspension of tyrosinase-NMPS (1.25 mg/mL) since it gave
a good response to l-tyrosine (3.33 × 10−4 M) and allowed to reduce
substantially the cost of the assays.
3.1.5. Selection of the substrate concentration for the inhibition
studies
The opinions about the substrate concentration to be consid-
ered in inhibition assays are quite contradictory. Kok et al. [40]
concluded, when measuring the inhibition potency of a competi-
tive inhibitor with an acetylcholinesterase and a choline oxidase
biosensor, that the inhibition percentage increased by raising the
substrate concentration. Therefore they worked in enzyme satu-
rated substrate conditions. Shan et al. [19] demonstrated that for
a tyrosinase biosensor for the determination of benzoic acid, the
concentration of the catechol substrate did not affect the maxi-
mum inhibition percentage but it affected the sensitivity of the
984 V.H. Sima et al. / Talanta 83 (2011) 980–987
method. It must be noted, however, that the use of a high sub-
strate concentration would not yield sensitive inhibition responses
when the quantification of a competitive inhibitor is performed by
simultaneous addition of the inhibitor and the substrate. Because
in competitive inhibition the substrate competes with the inhibitor
for the enzyme active site, and the inhibition, especially at low
inhibitor concentrations, would likely not be detected [40].
Two different l-tyrosine concentrations were studied;
3.33 × 10−4 M and 4.76 × 10−5 M, and the obtained IC50 for
kojic acid were found to be equal within the experimental error
(RSD = 2.3%). The percentage of inhibition was evaluated from
the response of the uninhibited form of the enzyme versus the
response of the enzyme partially inhibited. Taking into account that
our device was not designed for the quantification of inhibitors but
for the comparative screening and quantification of their inhibitory
potency we considered that all the immobilized enzyme molecules
must take part in the reaction and this could only be possible in
substrate concentration corresponding to the saturation portion
of the activity versus substrate curve [40]. Enzyme saturated
substrate concentration, namely 3.33 × 10−4 M l-tyrosine, was
then selected for further inhibition studies.
Since the concentration of substrate was quite high, it had to be
established if the quantity of the dissolved oxygen in the solution
was not a limiting factor for a 3.33 × 10−4 M l-tyrosine concentra-
tion. After 30 min of air bubbling into the working buffer solution
the biosensor response did not change, but after 30 min of nitrogen
bubbling the biosensor response decreased dramatically. The latter
was also a further indirect evidence of the generation of an elec-
troactive species by the immobilized tyrosinase. It was concluded
that in the selected experimental conditions the quantity of oxy-
gen was not a limiting factor and that the enzymatic process was
saturated by the substrate.
3.1.6. Michaelis–Menten curve
A typical Michaelis–Menten plot and its linearization by the
Lineweaver–Burk plot (y(A−1 ) = −7.91 × 103 × (M−1 )−8.53 × 107 ,
R2 = 1.000, RSD slope = 3.3%, RSD intercept = 7.7%, n = 3) was
obtained for l-tyrosine between 2.49 × 10−6 M and 4.41 × 10−4 M
in the above described optimal working conditions of the tyrosinase
based biosensor, namely: applied potential (Eapp ) = −100 mV, 0.1 M
phosphate buffer pH 6.5, 10 ␮L of 1.25 mg/mL tyrosinase-NMPS
suspension spiked onto the mCPE surface. The kinetic parameters
Km app and Imax were calculated as the average of 3 consecu-
tive determinations: Km app = 9.7 × 10−5 M (RSD = 3.1%, n = 3) and
Imax = −1.2 × 10−8 A (RSD = 7.5%, n = 3).
The obtained Km app values (Table 2) can be compared to
those reported in the literature for the enzyme in solution, i.e.
Km app = 6.2 × 10−4 M [23] or 3.3 × 10−4 M [41] at pH 6.5, illustrating
the fact that the applied immobilization technique maintained the
enzyme affinity for its substrate.
3.2. Inhibitors assay
The developed biosensor was designed for testing inhibitors of
the skin pigmentation process. When testing their potency, there
are many factors which contribute to the final results: type of
enzyme, quantity of immobilized enzyme, immobilization method,
type of substrate, substrate concentration, time of contact between
the enzyme, substrate and inhibitor, pH, temperature, applied
potential, rate of solution stirring. In order to obtain a correct com-
parison of the inhibition potency of the studied compounds, it was
worked under the same experimental conditions [19].
As the developed device was a novel analytical tool, in order to
test its efficiency, it was applied to the study of compounds that
have already demonstrated their activity in clinical practice or by
using other analytical methods.
Table 1
IC50 for kojic acid, benzoic acid, azelaic acid and ascorbic acid.
IC50 (␮M) Kojic acid Benzoic acid Azelaic acid Ascorbic acid
IC50 1 3.66 72.0 125 12.4
IC50 2 3.57 72.0 127 11.6
IC50 3 3.71 72.6 128 11.0
IC50 4 3.72 70.9 119 12.2
IC50 5 3.63 70.5 127 11.7
IC50average 3.66 71.6 125 11.8
The proposed biosensor was intended to test inhibitors that
have two different mechanisms of action, namely: true enzyme
inhibitors which bind to the active site of the enzyme and inhibitors
which consume the dopaquinone intermediate of the melanin syn-
thesis pathway.
It was checked that the NMPS mCPE (i.e. without tyrosinase)
gave no response under the selected experimental conditions and
in the presence of the studied compounds. It was found that l-
tyrosine, kojic acid, benzoic acid, azelaic acid and ascorbic acid gave
no amperometric signal.
Also blank experiments at the tyrosinase-NMPS-mCPE without
addition of the l-tyrosine substrate were performed. No ampero-
metric response was observed for the tested inhibitors.
3.2.1. “True” enzyme inhibitors
Kojic acid, benzoic acid and azelaic acid inhibit the melanin for-
mation by competing with tyrosinase natural substrate, l-tyrosine,
for the binding to the enzyme active site. As illustrated in Fig. 3a
typical time-dependent response was obtained at the tyrosinase-
NMPSs biosensor by l-tyrosine injection followed by stepwise
additions of kojic acid. It can be seen that the biosensor baseline
was reached within 600 s allowing the experiment to be per-
formed. Once the steady state amperometric response of l-tyrosine
(3.33 × 10−4 M) was observed, successive additions of inhibitor
produced a diminution of the reduction current. This clearly indi-
cated that kojic acid interfered with the production of dopaquinone
at the electrode surface. The inhibition percentage (In %) increased
with the inhibitor concentration. Kojic acid was tested in the con-
centration range 2.50 × 10−6 to 1.24 × 10−5 M where it inhibited
between 33% and 88% of the response to l-tyrosine. Same experi-
ments were performed for benzoic acid in the concentration range
3.98 × 10−5 to 1.96 × 10−4 M where it inhibited between 34% and
81% of the response and also for azelaic acid in the concentration
range 9.90 × 10−5 to 4.76 × 10−4 M where it inhibited between 43%
and 88% of the response.
The inhibition calibration curve of kojic acid is shown in
the insert of Fig. 3. IC50 , the concentration of inhibitor which
inhibited 50% of the l-tyrosine signal, was calculated from 5
different curves by plotting 1/In(%) versus 1/inhibitor concen-
tration (Table 1): kojic acid IC50 = 3.7 × 10−6 M, RSD = 1.6%, n = 5
(y = 4.54 × 10−8 x + 7.61 × 10−3 , R2 = 0.9999, RSD slope = 3.1%, RSD
intercept = 4.1%), benzoic acid IC50 = 7.2 × 10−5 M, RSD = 1.2%, n = 5
(y = 7.88 × 10−7 x + 8.99 × 10−3 , R2 = 0.9990, RSD slope = 5.1%, RSD
intercept = 6.3%) and azelaic acid IC50 = 1.3 × 10−4 M, RSD = 3.0%,
n = 5 (y = 1.60 × 10−6 x + 7.76 × 10−3 , R2 = 0.9996, RSD slope = 9.3%,
RSD intercept = 7.6%).
The study of the kinetics and of the mechanism of inhibition
of kojic acid, benzoic acid and azelaic acid was also carried out.
The response of the tyrosinase-NMPSs biosensor to l-tyrosine was
studied in the absence and in the presence of different concentra-
tions of inhibitor (Fig. 4). Table 2 data show that approximately
the same maximum current was obtained but different values
for the apparent Michaelis–Menten constant (Km app ), determined
following a Lineweaver–Burk plot, were calculated when various
amounts of kojic acid was added in the initial testing solution. The
same experiment was performed and the same conclusions were
Academic Press is an imprint of Elsevier
225 Wyman Street, Waltham, MA 02451, USA
525 B Street, Suite 1800, San Diego, CA 92101–4495, USA
125 London Wall, London, EC2Y 5AS, UK
The Boulevard, Langford Lane, Kidlington, Oxford OX5 1GB, UK

First edition 2015

© 2015 Elsevier Inc. All rights reserved.

No part of this publication may be reproduced or transmitted in any form or by any means,
electronic or mechanical, including photocopying, recording, or any information storage and
retrieval system, without permission in writing from the publisher. Details on how to seek
permission, further information about the Publisher’s permissions policies and our
arrangements with organizations such as the Copyright Clearance Center and the Copyright
Licensing Agency, can be found at our website: www.elsevier.com/permissions.

This book and the individual contributions contained in it are protected under copyright by
the Publisher (other than as may be noted herein).

Notices
Knowledge and best practice in this field are constantly changing. As new research and
experience broaden our understanding, changes in research methods, professional practices,
or medical treatment may become necessary.

Practitioners and researchers must always rely on their own experience and knowledge in
evaluating and using any information, methods, compounds, or experiments described
herein. In using such information or methods they should be mindful of their own safety and
the safety of others, including parties for whom they have a professional responsibility.

To the fullest extent of the law, neither the Publisher nor the authors, contributors, or editors,
assume any liability for any injury and/or damage to persons or property as a matter of
products liability, negligence or otherwise, or from any use or operation of any methods,
products, instructions, or ideas contained in the material herein.

ISBN: 978-0-12-803300-5
ISSN: 1871-5125 (Series)

For information on all Academic Press publications

visit our website at http://store.elsevier.com/
 Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 151 (2015) 945–955

Contents lists available at ScienceDirect

Spectrochimica Acta Part A: Molecular and

Biomolecular Spectroscopy
journal homepage: www.elsevier.com/locate/saa

Continuous Wavelet Transform, a powerful alternative to Derivative

Spectrophotometry in analysis of binary and ternary mixtures: A
comparative study
Eman S. Elzanfaly, Said A. Hassan ⇑, Maissa Y. Salem, Badr A. El-Zeany
Department of Analytical Chemistry, Faculty of Pharmacy, Cairo University, Kasr El-Aini Street, 11562 Cairo, Egypt

h i g h l i g h t s g r a p h i c a l a b s t r a c t

 Numerical Differentiation and

Wavelet Transform as approaches for
derivative calculation.
 Wavelet Transform is a powerful
alternative to traditional derivative
algorithms.
 Simple, selective and precise
spectrophotometric methods for
binary and ternary mixtures.
st
 The 1 spectrophotometric method
for analysis of Amlodipine, Aliskiren
and Hydrochlorothiazide.
 Methods validated as per ICH
guidelines, parameters found to be
within the limits.

a r t i c l e i n f o a b s t r a c t

Article history: A comparative study was established between two signal processing techniques showing the theoretical
Received 21 April 2015 algorithm for each method and making a comparison between them to indicate the advantages and
Received in revised form 20 June 2015 limitations. The methods under study are Numerical Differentiation (ND) and Continuous Wavelet
Accepted 25 June 2015
Transform (CWT). These methods were studied as spectrophotometric resolution tools for simultaneous
Available online 30 June 2015
analysis of binary and ternary mixtures. To present the comparison, the two methods were applied for
the resolution of Bisoprolol (BIS) and Hydrochlorothiazide (HCT) in their binary mixture and for the
Keywords:
analysis of Amlodipine (AML), Aliskiren (ALI) and Hydrochlorothiazide (HCT) as an example for ternary
Derivative Spectrophotometry
Continuous Wavelet Transform
mixtures. By comparing the results in laboratory prepared mixtures, it was proven that CWT technique
Amlodipine is more efficient and advantageous in analysis of mixtures with severe overlapped spectra than ND.
Aliskiren The CWT was applied for quantitative determination of the drugs in their pharmaceutical formulations
Bisoprolol and validated according to the ICH guidelines where accuracy, precision, repeatability and robustness
Hydrochlorothiazide were found to be within the acceptable limit.
Numerical Differentiation Ó 2015 Elsevier B.V. All rights reserved.

1. Introduction

UV–VIS absorption spectroscopy is a well-established technique

for rapid and accurate determination of analytes in a mixture form
without prior separation, if the interferences between the spectra
⇑ Corresponding author.
can be eliminated. One of the techniques used for elimination of
E-mail addresses: said.hassan@pharma.cu.edu.eg, saidmonem_84@yahoo.com
(S.A. Hassan).
interference in spectroscopy is signal processing. The most

http://dx.doi.org/10.1016/j.saa.2015.06.100
1386-1425/Ó 2015 Elsevier B.V. All rights reserved.
946 E.S. Elzanfaly et al. / Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 151 (2015) 945–955
commonly used signal processing technique is Derivative
Spectrophotometry (DS).
Rutherford [1] was the first to apply derivative methods suc-
cessfully to experimental data, he suggested that the sensitivity
of mass spectroscopy would be increased on applying the first
derivative technique. Derivative methods were first applied to
UV–VIS and IR spectroscopy in 1953 by Hammond and Price [2],
followed by Morrison [3] and in 1955 Giese and French [4]
described the theory of the derivative technique. In 1959, Martin
[5] introduced the theoretical basis for second and higher orders
of derivative, and in 1968, Morrey [6] generated first to fourth
order derivatives with a digital computer. In 1964, Savitzky and
Golay [7] introduced least-squares procedures for the smoothing
and differentiation of fluctuating data.
Derivative calculation is a powerful technique used in analytical
chemistry to resolve spectra, sharpen peaks, remove background
interference, and carry out quantitative analysis. The most impor-
tant field is its use in resolution enhancement technique to facili-
tate the detection and location of poorly resolved components in
complicated experimental signals [8]. It was applied successfully
to chromatography [9], electrochemistry [10–12], flow-injection
analysis [13] and in all fields of spectroscopy [14–16]. In the region
of UV–VIS spectroscopy, DS played an important role in resolution
of overlapped spectra [17,18]. The zero-crossing derivative spectra
method was the most common procedure for the simultaneous
determination of single components as well as binary and ternary
mixtures from their overlapping spectra [19–21]. DS was also
applied to ratio spectra for analysis of binary and ternary mixtures
[22–24].
In the spectral studies, the application of the usual DS to the
original absorption spectra possesses several disadvantages such
as: peak intensity diminishes with higher order derivation, it
required the additional smooth function as well as the scaling
factor processes. These parameters may deform the obtained
derivative spectra from the original ones. As a result, the usual
derivative method produces several errors in the quantitative anal-
ysis [25]. These drawbacks suggested using other signal processing
techniques as powerful alternatives for derivative calculation such
as Wavelet Transform [26,27].
A Wavelet Transform (WT) involves the decomposition of a sig-
nal function or vector e.g., a spectrum of a chemical species into
simpler, fixed building blocks at different scales and positions
[28]. Different algorithms were proposed to achieve this purpose.
In 1977, Coifman and Weiss [29] developed the automatic decom-
position method, based on work of Calderón and Zygmund [30],
which had a very strong influence on the development of wavelet
theory. In 1984, Grossmann and Morlet [31] proposed the
Continuous Wavelet Transform (CWT). In 1989, Mallat [32] intro-
duced the multi-resolution signal decomposition MRSD algorithm.
Daubechies [26,33,34] adopted this approach to construct families
of compactly supported wavelets and coupled it to quadrature mir-
ror filtering.
WT has been applied successfully for signal processing in differ-
ent fields of analytical chemistry. It was applied to de-noising of
data [35], compression of signals [36], analysis of electrochemical
data [37], flow-injection analysis [38], multivariate calibration
[39] and quantitative analysis of near infrared spectra [40]. In the
field of UV–VIS spectrophotometry, Continuous Wavelet
Transform (CWT) combined either with a zero-crossing technique
[41] or ratio spectra [13] was used for simultaneous determination
of chemical species in binary and ternary mixtures [18,42–45].
ConcorÒ Plus tablets contain both Bisoprolol hemifumarate
(BIS) and Hydrochlorothiazide (HCT) for the treatment of high
blood pressure [46]. There are reported methods for the determi-
nation of BIS or HCT in different dosage forms [47–50] and in their
binary mixtures [45,51–55].
AmturnideÒ is a combination of Amlodipine besylate (AML),
Aliskiren hemifumarate (ALI) and Hydrochlorothiazide (HCT) indi-
cated for the treatment of hypertension to reduce the risk of
strokes and myocardial infarctions [56]. Many reported methods
were reported for the determination of AML, ALI or HCT in different
dosage forms [57–62]. HPLC and capillary electrophoresis methods
were reported for simultaneous determination of this ternary mix-
ture [63,64]. No spectrophotometric methods were developed for
the determination of this mixture.
In this work, ND and CWT were applied for determination of
Bisoprolol (BIS) and Hydrochlorothiazide (HCT) in their binary
mixture and for determination of Amlodipine (AML), Aliskiren
(ALI) and Hydrochlorothiazide (HCT) in their ternary mixture as
examples for complex spectra. A comparative study between the
two methods was performed to compare their resolution power
in analysis of overlapped spectra. Then the CWT was used for the
analysis of the cited mixtures in pharmaceutical formulations,
namely ConcorÒ Plus tablets and AmturnideÒ tablets.
2. Theoretical background
2.1. Numerical Differentiation (ND) [65]
For a discrete spectrum xi (i = 1, . . ., n), suppose that wi
(i = 1, . . ., n) are its sampling wavelengths, the direct-difference
method can be expressed as follows:
X iþ1  X i
yi ¼
W iþ1  W i
Here yi (i = 1, . . ., n  1) represent the series of derivatives of spec-
trum xi. If the sampling wavelength is equal, this equation can be
rewritten as yi = xi+1  xi (i = 1, . . ., n  1).
2.2. Wavelet Transform (WT): [27]
Wavelet Transform (WT) involves decomposition of a signal
function or vector (spectrum) into simpler, fixed building blocks
at different scales and positions. In WT treatment, all basis func-
tions Wa;b ðxÞ can be derived from a mother wavelet WðxÞ through
the following dilation and translation processes:
 
1 xb
Wa;b ðxÞ ¼ a2 W a; b 2 R and a–0
a
where a and b are the scale and position parameters, respectively,
expressed in real number R.
The basic idea of WT is to represent any arbitrary function f ðxÞ
as a superposition of wavelets. Decomposition of x with respect to
the wavelet function series {Wj;k ðxÞ} is described by the following
formula:
X
þ1 X
þ1
f ðxÞ ¼
ðjÞ
C k Wj;k
j¼1 k¼1
Therefore, the signal is represented by a set of coefficients {C j;k } in
the wavelet domain.
The approximate first derivative of X is assigned to be equal to
the difference between two scale coefficients at the first resolution
level as
xð1Þ  C 1;D2m  C 1;D2m~ ~
m–m
where, D2m and D2m~ represent any two Daubechies wavelet
~ being any positive integer.
functions, with m and m
 E.S. Elzanfaly et al. / Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 151 (2015) 945–955
3. Experimental
3.1. Apparatus
SHIMADZU dual beam UV–visible spectrophotometer
(Kyoto/Japan), model 1650 UV-PC, with matched 1-cm quartz cells
was used. The bundled software, UV-Probe personal spectroscopy
software version 2.21 (SHIMADZU) is used. The spectral bandwidth
is 2.0 nm and wavelength scanning speed is 2800 nm/min with
0.1 nm interval.
3.2. Chemicals and reagents
3.2.1. Pure samples
- BIS, HCT and AML; kindly supplied by Al-Hekma pharmaceuti-
cal Company, Cairo, Egypt, their purity was certified to be
99.7 ± 0.8%, 100.6 ± 0.6% and 99.9 ± 0.5%, respectively.
- ALI; kindly supplied by Novartis Pharmaceuticals Corporation,
USA, its purity was certified to be 100.6 ± 0.3%.
3.2.2. Market samples
- ConcorÒ 5 Plus labeled to contain 5(BIS)/12.5(HCT) mg, batch
number 1030039 and ConcorÒ 10 Plus labeled to contain
10(BIS)/25(HCT) mg, batch number 0795049, manufactured
by Pfizer Ltd., Cairo, Egypt.
- AmturnideÒ tablet dosage forms; labeled to contain 10
(AML)/300 (ALI)/25 (HCT) mg batch number F0005, manufac-
tured by Novartis Pharmaceuticals Corporation, USA.
3.2.3. Methanol
Spectroscopy grade (El-NASR Pharmaceutical Chemicals Co.,
Abu-Zaabal, Cairo, Egypt).
3.3. Procedures
3.3.1. Standard solutions
- BIS, HCT, AML and ALI standard stock solutions; 1000 lg /mL in
methanol.
- BIS, HCT and AML standard working solutions; 50 lg/mL in
methanol.
- ALI standard working solution; 500 lg/mL in methanol.
- Preparation of laboratory prepared mixtures; aliquots of stan-
dard working solutions were transferred into two series of
10-mL measuring flasks, completed to volume with methanol
to prepare binary mixture of BIS and HCT and ternary mixture
of AML, ALI and HCT containing different ratios of the drugs.
3.3.2. Spectral characteristics of the drugs
The zero-order (D0) absorption spectra of 4 lg/mL BIS and
10 lg/mL HCT were recorded against methanol as a blank over
the range of 200–400 nm.
The zero-order (D0) absorption spectra of 4, 120 and 10 lg/mL
solutions of AML, ALI and HCT, respectively, were recorded against
methanol as a blank over the range of 200–400 nm.
3.3.3. Construction of calibration curves
Aliquots equivalent to 20–400 lg BIS, 5–250 lg HCT, 10–350 lg
AML and 150–2500 lg ALI were accurately transferred from their
respective standard working solutions into separate series of
10-mL volumetric flasks then completed to volume with methanol.
The spectra of the prepared standard solutions were scanned from
200 to 400 nm and stored in the computer.
947
3.3.3.1. Numerical Differentiation method (ND).
Binary mixture
For determination of BIS, the 3rd and 4th derivative spectra were
obtained by ND method with Dk = 8 nm and scaling factor = 1000.
The peak amplitude was measured at 283.6 and 287.5 nm,
respectively.
For determination of HCT, the first derivative spectra were
obtained by ND method with Dk = 4 nm and scaling factor = 100.
The peak amplitude was measured at 272.3 nm.
The calibration curves were constructed relating the peak
amplitudes at the specified wavelengths to the corresponding
concentrations of BIS and HCT.
Ternary mixture
For determination of AML, the first derivative spectra were
obtained by ND method with Dk = 4 nm and scaling factor = 10.
The peak amplitude was measured at 386.8 nm.
For determination of ALI, the fourth derivative spectra were
obtained by ND method with Dk = 8 nm and scaling factor = 1000.
The peak amplitude was measured at 287.5 nm.
For determination of HCT, the second derivative spectra were
obtained by ND method with Dk = 8 nm and scaling factor = 100.
The peak amplitude was measured at 332.1 nm.
The calibration curves were constructed relating the peak
amplitudes at the specified wavelengths to the corresponding
concentrations of AML, ALI and HCT.
3.3.3.2. Continuous Wavelet Transform method (CWT).
Binary mixture
The first derivative spectra were obtained by CWT method with
Morlet family (morl) and scaling = 200. The calibration curves were
constructed relating the peak amplitudes at 278.3 and 283.6 nm to
the corresponding concentrations of BIS and HCT, respectively.
Ternary mixture
For determination of AML, the first derivative spectra were
obtained by CWT method with Gaussian-1 family (Gaus-1) and
scaling = 200. The peak amplitude was measured at 387.2 nm.
For determination of ALI, the first derivative spectra were
obtained by CWT method with Gaussian-8 family (Gaus-8) and
scaling = 200. The peak amplitude was measured at 315.6 nm.
For determination of HCT, the first derivative spectra were
obtained by CWT method with Coiflet-3 family (Coif-3) and
scaling = 200. The peak amplitude was measured at 289.1 nm.
The calibration curves were constructed relating the peak
amplitudes at the specified wavelengths to the corresponding
concentrations of AML, ALI and HCT.
3.3.4. Application of the proposed methods for the determination of
laboratory-prepared mixtures
The spectra of laboratory prepared mixtures were scanned from
200 to 400 nm and stored in the computer. The same procedure
under construction of calibration curves was applied and the con-
centrations of the drugs were calculated from the corresponding
regression equations.
3.3.5. Application of the proposed methods for the determination of the
drugs in dosage forms
Ten tablets of both ConcorÒ 5 Plus, ConcorÒ 10 Plus were accu-
rately weighed, finely powdered and an amount of the powder
948 E.S. Elzanfaly et al. / Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 151 (2015) 945–955
equivalent to 10 mg BIS was weighed. Five tablets of AmturnideÒ
were accurately weighed, finely powdered and amount of the pow-
der equivalent to 300 mg ALI was weighed. The powder was dis-
solved in methanol by shaking in ultrasonic bath for about
30 min, then they were filtered and transferred quantitatively into
separate 100-ml volumetric flasks and the volume was completed
to the mark with methanol. Ten ml aliquots were transferred into
separate 50-ml volumetric flasks and the volume was completed to
the mark with methanol. Then necessary dilutions were made to
reach concentrations in the linearity range. The spectra of these
solutions were scanned from 200 to 400 nm and stored in the com-
puter. Spectra obtained were analyzed by the proposed methods.
4. Results and discussion
Signal processing techniques use different algorithms in trans-
forming the spectra into new graphs where the spectra of different
compounds are resolved. Derivative Spectrophotometry (DS) is the
most commonly used processing technique for resolution of over-
lapped spectra. Lately, Continuous Wavelet Transform algorithm
(CWT) started to be used as an alternative to Numerical
Differentiation (ND) for calculation of DS.
The aim of this work was to establish a comparative study
between the two signal processing techniques. For proper presen-
tation of the comparison, two different mixtures were chosen to
show the resolution power of the methods. The first mixture was
a combination of BIS and HCT as an example of binary mixtures.
The spectra of BIS and HCT (Fig. 1) show severe overlap as the peak
of HCT (kmax 226 nm) completely coincide with that of BIS (kmax
225 nm) showing the same features. This overlap, together with
the high absorptivity of HCT which is also found in higher level
in ConcorÒ Plus tablets, prevents direct determination of BIS. In
previous work, the transformed ratio spectra had to be combined
with zero-crossing technique for analysis of BIS in this mixture
[45]. The second mixture composed of AML, ALI and HCT was used
as an example of ternary mixtures. The absorption spectra of AML,
Absorbance
Wavelength (nm)
Fig. 1. Zero order absorption spectra of 4 lg/mL BIS (—) and 10 lg/mL HCT (- - -) using methanol as blank.
ALI and HCT show highly overlapped spectral band in the region
200–350 nm (Fig. 2). The difficulty of BIS determination in the bin-
ary mixture and the severe overlap in the ternary mixture nomi-
nated these combinations to be ideal mixtures presenting the
comparative study to show the advantages and disadvantages of
both algorithms in analysis of overlapping spectra. HCT in the bin-
ary mixture and AML in the ternary mixture can be directly deter-
mined in the zero order spectra at their kmax, 316 and 360 nm
respectively, without interference from other components, but
they were determined by ND and CWT for investigational
purposes.
For using ND method, the parameters of derivative calculation,
wavelength increment over which the derivative is obtained (Dk)
and the scaling factor, were optimized. Different Dk (2, 4 and 8)
and scaling factors of 10, 100 and 1000 were tested, and the best
parameters, in terms of spectral shapes, linearity and recovery,
are mentioned under procedure.
For analysis of chosen mixtures using CWT algorithm, different
wavelet families were tested such as Haar, Daubechies (db),
Symlets (sym), Coiflets (coif), Meyer (meyr), Gaussian (gaus),
Mexican hat (mexh) and Morlet (morl). The only parameter that
should be optimized after choosing the proper wavelet is the scal-
ing parameter. No need for scaling factor as CWT has the advantage
of signal amplification compared to ND.
4.1. Binary mixture
In this section, we will focus on determination of BIS being a
challenge in presence of highly absorptive HCT. Thus, the choice
of derivative order and optimum parameters is mainly considered
for determination of BIS.
For analysis of BIS in this mixture using ND algorithm, the 1st,
2 , 3rd and 4th order derivatives were obtained (Fig. 3). The 1st
nd
order did not show any possible zero-crossing points that can be
used for BIS determination (Fig. 3a), while the wavelengths 241.0
and 270.0 nm did not show a real zero-crossing. Possible
 E.S. Elzanfaly et al. / Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 151 (2015) 945–955
Absorbance
Wavelength (nm)
Fig. 2. Zero order absorption spectra of 4 lg/mL AML (—), 120 lg/mL ALI (....) and 10 lg/mL HCT (- - -), using methanol as blank.
zero-crossing points were noticed in higher order derivatives,
which are 230.3 and 278.5 nm in 2nd order, 233.3 and 283.6 nm
in 3rd order and 229.5, 235.3 and 287.5 nm in 4th order. In contrast
to 1st order derivative, true zero-crossing points were detected in
2nd (278.5 nm), 3rd (283.6 nm) and 4th (287.5 nm) orders
(Fig. 3b–d). The zero-crossing point in 2nd order showed poor lin-
earity and recovery of laboratory prepared mixtures, while the
points in 3rd and 4th orders resulted in good recoveries in some
laboratory-prepared mixtures, where BIS was found in higher ratio
than HCT. For determination of HCT, the 1st order derivative was
enough to get zero-crossing point (272.3 nm) and good recoveries
in all ratios of laboratory-prepared mixtures were obtained
Table 1.
For analysis of BIS using CWT algorithm, the families were
tested in different orders and with different scaling parameters
and the best results, regarding linearity and recovery% in
laboratory-prepared mixtures, were obtained with 1st derivative
calculated using morl family and scaling = 200 (Fig. 4) at
278.3 nm. Different wavelengths were detected as possible
zero-crossing points at 255.3, 265.7, 278.3, 292.2 and 306.3 nm.
The points at 255.3 and 265.7 nm were not real zero-crossing
points and the last two wavelengths showed bad linearity and sen-
sitivity compared to 278.3 nm. HCT could be determined with the
same parameters at 283.6 nm.
4.2. Ternary mixture
For analysis of this mixture using ND algorithm, the 1st order
derivative spectra were obtained where a zero-crossing point
(386.8 nm) was detected for AML determination (Fig. 5a). But in
the same order spectra, ALI and HCT were not resolved, so the
2nd order derivatives were calculated and a zero-crossing point
(332.1 nm) was noticed for determination of HCT (Fig. 5b). For
determination of ALI, the 3rd and 4th order derivatives were tried.
While no zero-crossing points were observed in the 3rd order spec-
tra (Fig. 5c), the 4th order showed a zero-crossing point (287.5 nm)
that was used for ALI determination (Fig. 5d).
949
For analysis of the ternary mixture using CWT algorithm, differ-
ent families were tried in different orders and scaling. The best
results regarding presence of zero-crossing points, linearity and
selectivity were obtained with 1st order derivatives using gaus-1,
gaus-8 and coif-3 families and scaling = 200 for determination of
AML, ALI and HCT at 387.2, 315.6 and 289.1 nm, respectively
(Fig. 6).
The selectivity of the proposed ND and CWT methods was
assessed by the analysis of laboratory prepared mixtures contain-
ing different ratios of the drugs (Tables 1 and 2). The proposed
CWT method was applied for the determination of BIS and HCT
in ConcorÒ Plus tablets and AML, ALI and HCT in AmturnideÒ
tablets and the validity was further assessed by applying the stan-
dard addition technique (Tables 3 and 4). Results obtained by CWT
for analysis of the dosage forms were statistically compared to
those obtained by the reported HPLC methods [51,63]. The results
showed no significant differences between the proposed method
and the reported ones as presented in Table 5. Validation was done
according to ICH recommendations and the validation parameters
are shown in Table 6.
4.3. CWT versus ND
ND is the simplest method of derivative calculation. The major
problem is the choice of Dk. Wide Dk values can be used to sup-
press noise and increase SNR, but they degrade resolution as well.
Finding the optimum Dk to compromise between the high resolu-
tion and good signal to noise ratio (S/N) is considered a challenge.
Also the values of S/N obtained by ND are very small, so they are
usually multiplied by scaling factor to be considerable.
The 1st order derivative fails to resolve severely overlapped
spectra, so we use higher order derivatives. This approach can help
in case the components to be determined are present in a reason-
able ratio and have comparable absorptivities. In the case of BIS in
its binary mixture with HCT, the zero order spectrum of BIS is com-
pletely enclosed with that of HCT showing very similar features
and low absorptivity, so 1st order derivative failed to present a
950 E.S. Elzanfaly et al. / Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 151 (2015) 945–955

a b

HCT BIS

c
d

BIS

BIS

Fig. 3. Different order absorption spectra of 40 lg/mL BIS (–) and 1–22 lg/mL HCT (- - -) using ND method showing zero-crossing points for determination of BIS and HCT. (a)
1st order (b) 2nd order (c) 3rd order (d) 4th order.

Table 1
Determination of BIS and HCT in laboratory prepared mixtures by ND and CWT methods.

Concentration (lg/mL) Recovery%a

BIS HCT
BIS HCT ND (3rd order) ND (4th order) CWT ND (1st order) CWT
b b
4 10 54.7 115.1 99.5 101.8 101.3
6 6 89.9b 103.9b 102.4 101.8 99.4
6 12 65.7b 107.8b 98.4 101.1 100.6
6 15 64.6b 109.7b 100.2 100.6 100.2
6 18 53.7b 92.2b 101.3 100.2 99.1
12 6 103.0 100.4 100.6 101.8 100.7
12 15 76.0b 92.6b 98.1 101.3 99.8
15 6 102.9 100.5 101.1 102.5 99.9
15 12 85.4b 99.7 99.0 101.9 100.2
15 15 77.3b 94.3b 98.5 100.9 99.3
18 6 101.6 97.3 101.6 102 99.0
Mean ± SD 102.5 ± 0.8 99.5 ± 1.5 100.1 ± 1.5 101.4 ± 0.7 100.0 ± 0.7
a
Average of three determinations.
b
Excluded according to rejection rule [67].

zero-crossing point for its determination. The 2nd, 3rd and 4th order
derivatives showed zero-crossing points for BIS determination, the
recoveries in laboratory prepared mixtures indicated that BIS can-
not be determined in some mixtures contained higher levels of
HCT. This may be attributed to the poor absorptivity of BIS, at
the wavelengths where zero-crossing points were observed, in
addition to the fact that higher order derivatives diminish the sig-
nal to noise ratio (S/N).
In contrast to ND, CWT could resolve BIS from HCT in different
ratios by 1st order derivative using morl family. Using CWT algo-
rithm offers the flexibility of using several wavelet families, which
results in transforming the spectra into 1st order derivative in dif-
ferent shapes. This fact increases the possibility of finding
zero-crossing points in the 1st order derivatives for compounds
with very similar spectra. The CWT algorithm also has the advan-
tage of signal amplification compared to ND [66]. In addition to
 E.S. Elzanfaly et al. / Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 151 (2015) 945–955 951

Amplitude

HCT

BIS

Wavelength (nm)

st
Fig. 4. 1 order absorption spectra of 40 lg/mL BIS (—) and 1–22 lg/mL HCT (- - -) using CWT method (morl) showing zero-crossing points for BIS and HCT determination.

a b
AML
HCT

c d

ALI

Fig. 5. Different order absorption spectra of 35 lg/mL AML (–), 250 lg/mL ALI (....) and 25 lg/mL HCT (- - -) using ND method showing zero-crossing points for determination
of AML, ALI and HCT. (a) 1st order (b) 2nd order (c) 3rd order (d) 4th order.

the fact that calculating higher order derivatives is not necessary as
the 1st order derivative in CWT is enough to get zero-crossing
points. Thus CWT maintains high S/N in contrast to ND calculation,
which is the reason why CWT showed higher sensitivity as shown
from LOD and LOQ values in Table 6. These facts can explain why
zero-crossing points for determination of BIS was not observed in
ND 1st order derivative, in contrast to the 1st order morl wavelet.
The higher S/N of CWT spectra allowed the determination of BIS
in mixtures where HCT is found in higher ratios, while 2nd, 3rd
and 4th order derivatives obtained by ND failed.
952 E.S. Elzanfaly et al. / Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 151 (2015) 945–955

AML

ALI

HCT

Fig. 6. First order absorption spectra of 35 lg/mL AML (–), 250 lg/mL ALI (....) and 25 lg/mL HCT (- - -), using CWT method showing zero-crossing point for determination of
AML, ALI and HCT. (a) gaus-1 (b) gaus-8 (c) coif-3.

The severe overlap between the spectra of AML, ALI and HCT in
the region 200–350 nm prevented the resolution of the drugs in 1st
order derivative in this region. So ALI and HCT were resolved only
in the 4th and 2nd order derivatives, respectively. AML was excep-
tion being determined in 1st order derivative at 386.8 nm where
ALI and HCT show no interference in their original spectra. Upon
using CWT families, gaus-1, gaus-8 and coif-3, the 1st order deriva-
tives were enough to resolve the three drugs.
The advantages of CWT over ND in calculating derivatives can
be summarized in the following:
 E.S. Elzanfaly et al. / Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 151 (2015) 945–955 953

Table 2
Determination of AML, ALI and HCT in their ternary mixture by ND and CWT methods.

Concentration (lg/mL) Recovery%a

AML ALI HCT
AML ALI HCT ND CWT ND CWT ND CWT
1 30 2.5 79.6b 102.6 99.0 98.5 102.2 99.8
2 60 5 101.0 102.0 100.0 99.2 98.5 100.9
4 120 10 101.5 102.3 99.6 99.2 99.6 101.3
5 30 25 97.6 102.3 101.7 100.0 100.6 100.6
6 180 15 98.3 102.2 99.1 100.9 98.5 101.2
12 30 5 100.2 101.8 102.0 98.9 100.7 98.1
15 15 15 100.5 100.9 94.8b 105.0b 101.0 99.4
20 20 20 99.9 100.4 95.0b 101.1 98.0 99.7
35 30 5 98.5 100.0 102.0 98.1 99.3 100.5
35 70 5 99.1 100.4 99.3 99.6 97.0 99.3
Mean ± SD 99.6 ± 1.3 101.5 ± 0.9 100.3 ± 1.3 99.5 ± 1.0 99.5 ± 1.6 100.1 ± 1.0
a
Average of three determinations.
b
Excluded according to rejection rule [67].

Table 3
Determination of BIS and HCT in ConcorÒ Plus tablets by the proposed CWT and the reported HPLC method [51] with application of standard addition technique.

Product Drug CWT Reported methoda Standard addition CWT

Takenb (lg/mL) Added (lg/mL) Found (lg/mL) Recovery%c
ConcorÒ Plus 5/12.5 BIS 99.1 ± 0.4 99.4 ± 0.9 6 5 5.04 100.8
6 5.96 99.3
7 7.06 100.9
Mean ± SD 100.3 ± 0.9
HCT 99.7 ± 0.4 100.2 ± 0.8 10 8 7.97 99.6
10 9.94 99.4
12 12.11 100.9
Mean ± SD 100.0 ± 0.8
ConcorÒ Plus 10/25 BIS 100.5 ± 0.7 100.7 ± 0.4 6 5 4.96 99.2
6 6.03 100.5
7 6.97 99.6
Mean ± SD 99.8 ± 0.7
HCT 99.5 ± 0.4 99.9 ± 0.9 10 8 7.95 99.4
10 9.98 99.8
12 12.03 100.3
Mean ± SD 99.8 ± 0.5
a
RP-HPLC using C18 and 0.1 M potassium dihydrogen phosphate buffer and acetonitrile (70:30, v/v) as mobile phase. The UV detection was carried out at 228 nm at a flow
rate of 1.0 mL/min.
b
Standard addition were done on two different dilutions of the dosage form for the drugs to be within the linearity range.
c
Average of three determinations.

Table 4
Determination of AML, ALI and HCT in AmturnideÒ tablets by the proposed CWT and the reported HPLC method [63] with application of standard addition technique.

Product Drug CWT Reported methoda Standard addition CWT

Takenb (lg/mL) Added (lg/mL) Found (lg/mL) Recovery%c
Ò
Amturnide 10/300/25 AML 100.6 ± 1.1 100.4 ± 0.9 6 4 3.98 99.5
6 6.09 101.5
8 8.03 100.4
Mean ± SD 100.5 ± 1.0
ALI 99.9 ± 0.6 100.3 ± 0.8 120 110 109.31 99.4
120 121.55 101.3
130 130.73 100.6
Mean ± SD 100.4 ± 1.0
HCT 100.4 ± 0.7 99.9 ± 0.7 10 5 5.01 100.2
10 10.11 101.1
15 14.88 99.2
Mean ± SD 100.2 ± 0.9
a
RP-HPLC using acetonitrile: methanol: 50 mM phosphate buffer (20:50:30% by volume) adjusted to pH 3 with orthophosphoric acid and flow rate of 1.0 mL/min at
239 nm.
b
Standard addition were done on two different dilutions of the dosage form for the drugs to be within the linearity range.
c
Average of three determinations.
954 E.S. Elzanfaly et al. / Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 151 (2015) 945–955

Table 5
Statistical comparison for the results obtained by the proposed CWT and the reported HPLC [51,63] methods for the analysis of ConcorÒ Plus and AmturnideÒ tablets.

Value ConcorÒ Plus tablets AmturnideÒ tablets

a
CWT Reported method CWT Reported methodb
BIS HCT BIS HCT AML ALI HCT AML ALI HCT
Mean 99.8 99.6 100.0 100.0 100.6 99.9 100.4 100.4 100.3 99.9
RSD% 0.9 0.4 1.0 0.8 1.1 0.6 0.7 0.9 0.8 0.7
n 6 6 6 6 6 6 6 6 6 6
Variance 0.832 0.140 0.903 0.582 1.329 0.347 0.496 0.755 0.617 0.499
Student’s t testc 0.474 1.116 – – 0.405 0.845 1.228 – – –
(2.228)
F value c 1.085 4.157 – – 1.76 1.777 1.007 – – –
(5.05)
a
RP-HPLC using C18 and 0.1 M potassium dihydrogen phosphate buffer and acetonitrile (70:30, v/v) as mobile phase. The UV detection was carried out at 228 nm at a flow
rate of 1.0 mL/min.
b
RP-HPLC using acetonitrile: methanol: 50 mM phosphate buffer (20:50:30% by volume) adjusted to pH 3with orthophosphoric acid and flow rate of 1.0 mL/min at
239 nm.
c
The values in the parenthesis are the corresponding theoretical values of t and F at P = 0.05.

Table 6
Assay validation sheet of ND and CWT methods for the simultaneous determination of the mixtures under study.

Parameter Binary mixture Ternary mixture

BIS HCT AML ALI HCT
ND (3rd ND (4th CWT ND CWT ND CWT ND CWT ND CWT
order) order)
Accuracya 102.0 ± 0.3 99.7 ± 2.2 100.0 ± 0.6 99.9 ± 1.6 99.8 ± 0.5 100.6 ± 1.2 100.8 ± 1.2 99.9 ± 1.0 100.3 ± 0.9 98.5 ± 0.8 99.7 ± 1.0
(Mean ± SD)
Precision
– Repeatabilityb 1.3 1.2 0.6 0.7 0.6 0.9 0.9 1.0 0.9 0.7 0.8
– Intermediate 1.8 1.6 0.9 1.2 0.8 1.2 1.3 1.2 1.1 1.0 1.1
precisionc
Robustnessd 1.5 1.5 0.8 1.4 0.6 1.1 0.8 1.3 1.3 1.6 1.6
e
LOD (lg/mL) 1.42 1.41 0.57 0.44 0.22 0.57 0.26 4.50 4.45 1.48 0.59
LOQe (lg/mL) 4.75 4.69 1.90 1.48 0.73 1.89 0.88 15.01 14.84 4.93 1.95
Linearity
– Slope 0.0291 0.0086 0.0115 0.1743 0.1062 0.0049 0.1644 0.0113 0.0392 0.0027 0.1507
– Intercept 0.0021 0.0116 0.0014 0.0049 0.0258 0.0011 0.0030 0.3070 0.0070 0.0001 0.0190
– Correlation 0.9998 0.9997 0.9999 0.9998 0.9999 0.9999 1.000 0.9999 0.9998 0.9998 0.9999
coefficient (r)
– Range (lg/mL) 5.0–40.0 5.0–40.0 2.0–40.0 2.0–22.0 1.0–22.0 2.0–35.0 1.0–35.0 15.0– 15.0– 5.0–25.0 2.0–25.0
250.0 250.0
a
The accuracy (n = 3), average of three concentrations of each drug.
b
The intraday (n = 3), average of three concentrations of each drug repeated three times within day.
c
The interday (n = 3), average of three concentrations each drug repeated three times in three days.
d
Robustness (n = 3), average of three concentrations each drug analyzed using 75% and 70% methanol.
e
LOD and LOQ were calculated using the following equations: LOD = 3Sr and LOQ = 10Sr, r is the standard deviation of residuals and S is the slope.

1. Flexibility in calculating 1st order derivatives in different shapes similar spectra without the need of higher order derivative
using different wavelet families, that allows CWT to provide calculation.
zero-crossing points for complex mixtures.
2. Signal amplification provided by the algorithm together with its
flexibility in 1st order derivatives gives CWT high sensitivity References
compared to ND. [1] E. Dymond, Math. Proc. Cambridge Philos. Soc. 22 (1924) 405–408.
[2] V. Hammond, W. Price, J. Opt. Soc. Am. 43 (1953) 924–930.
[3] J. Morrison, J. Chem. Phys. 21 (1953) 1767–1772.
[4] A.T. Giese, C.S. French, Appl. Spectrosc. 9 (1955) 78–96.
5. Conclusion [5] A. Martin, Spectrochim. Acta A 14 (1959) 97–103.
[6] J.R. Morrey, Anal. Chem. 40 (1968) 905–914.
[7] A. Savitzky, M.J. Golay, Anal. Chem. 36 (1964) 1627–1639.
Using CWT offered many advantages over traditional ND. One [8] G. Talsky, Derivative Spectrophotometry: Low and High Order, VCH, California,
advantage is that CWT algorithm maintains high S/N in contrast 1994.
to ND calculation. In addition to the fact that calculating higher [9] N. Funasaki, S. Hada, S. Neya, J. Phys. Chem. B 103 (1999) 169–172.
[10] B. Ebeshi, E. Vaikosen, D. Tuma, Int. J. Biol. Chem. Sci. 7 (2014) 2591–2599.
order derivatives is not necessary in CWT as the 1st order deriva-
[11] Y. Ni, J. Bai, Talanta 44 (1997) 105–109.
tive is enough to get zero-crossing points for severely overlapped [12] A. Molina, M. López-Tenés, C. Serna, J. Electrochem. Soc. 147 (2000) 3429–
spectra. The flexibility of CWT algorithm allowing the use of sev- 3435.
eral wavelet families in different orders, results in transforming [13] M. Blanco, J. Gene, H. Iturriaga, S. Maspoch, J. Riba, Talanta 34 (1987) 987–993.
[14] F. Singleton, G. Collier, J. Appl. Chem. 6 (1956) 495–510.
the spectra into different shapes of 1st order derivatives, which [15] W. Snelleman, T. Rains, K. Yee, H.D. Cook, O. Menis, Anal. Chem. 42 (1970)
increases the possibility of finding zero-crossing points in very 394–398.
 E.S. Elzanfaly et al. / Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 151 (2015) 945–955
[16] T.C. O’Haver, Modulation and Derivative Techniques in Luminescence
Spectroscopy, Springer, New York, 1976.
[17] A.Y. El-Sayed, N.A. El-Salem, Anal. Sci. 21 (2005) 595.
[18] C.B. Ojeda, F.S. Rojas, Microchem. J. 106 (2013) 1–16.
[19] M. Korany, A. Wahbi, M. Elsayed, S. Mandour, Farmaco Ed. Prat. 39 (1984) 243.
[20] M. Korany, A. Wahbi, M. Elsayed, S. Mandour, Anal. Lett. 17 (1984) 1373–1389.
[21] M.A. Korany, A.A.S. El-Din, N.A. Abael-salam, Anal. Lett. 17 (1984) 483–495.
[22] F. Salinas, J. Nevado, A. Mansilla, Talanta 37 (1990) 347–351.
[23] J. Berzas, C. Guiberteau, F. Salinas, Talanta 39 (1992) 547.
[24] E. Dinc, F. Onur, Anal. Chim. Acta 359 (1998) 93–106.
[25] E. Dinç, D. Baleanu, Spectrochim. Acta A 63 (2006) 631–638.
[26] I. Daubechies, Ten Lectures on Wavelets, SIAM, Pennsylvania, 1992.
[27] A.K.M. Leung, F.T. Chau, J.B. Gao, Anal. Chem. 70 (1998) 5222–5229.
[28] X.D. Dai, B. Joseph, R. Motard, Introduction to Wavelet Transform and Time-
frequency Analysis, Springer, New York, 1994.
[29] R.R. Coifman, G. Weiss, Bull. Am. Math. Soc. 83 (1977) 569–645.
[30] A. Calderón, A. Zygmund, Stud. Math. 46 (1973) 297–299.
[31] A. Grossmann, J. Morlet, SIAM J. Math. Anal. 15 (1984) 723–736.
[32] S.G. Mallat, IEEE Trans. Acoust. 37 (1989) 2091–2110.
[33] I. Daubechies, IEEE Trans. Inf. Theory 36 (1990) 961–1005.
[34] I. Daubechies, SIAM J. Math. Anal. 24 (1993) 499–519.
[35] D.L. Donoho, I.M. Johnstone, C.R. Acad. Sci. I Math. 319 (1994) 1317–1322.
[36] V. Barclay, R. Bonner, I. Hamilton, Anal. Chem. 69 (1997) 78–90.
[37] X. Zou, J. Mo, Anal. Chim. Acta 340 (1997) 115–121.
[38] M. Bos, E. Hoogendam, Anal. Chim. Acta 267 (1992) 73–80.
[39] H.W. Tan, S.D. Brown, J. Chemom. 16 (2002) 228–240.
[40] U. Depczynski, K. Jetter, K. Molt, A. Niemöller, Chemom. Intell. Lab. Syst. 47
(1999) 179–187.
[41] T. O’haver, G. Green, Anal. Chem. 48 (1976) 312–318.
[42] E. Dinç, D. Baleanu, J. Pharm. Biomed. Anal. 31 (2003) 969–978.
[43] E. Dinç, A. Özdemir, D. Baleanu, Talanta 65 (2005) 36–47.
[44] E. Dinç, M. Kanbur, D. Baleanu, Die Pharm. 60 (2005) 892–896.
[45] E.S. Elzanfaly, S.A. Hassan, M.Y. Salem, B.A. El-Zeany, Spectrochim. Acta A 140
(2015) 334–343.
[46] W.H. Frishman, B.S. Bryzinski, L.R. Coulson, V.L. DeQuattro, N.D. Vlachakis, W.J.
Mroczek, G. Dukart, J.D. Goldberg, D. Alemayehu, K. Koury, Arch. Internal Med.
154 (1994) 1461–1468.
955
[47] A.D. Gudruman, N. Bibire, G. Tantaru, M. Apostu, M. Vieriu, V. Dorneanu, Rev.
Chim. 64 (2013) 393–396.
[48] H.W. Darwish, S.A. Hassan, M.Y. Salem, B.A. El-Zeany, Spectrochim. Acta A 122
(2013) 744–750.
[49] M. Emanual, J. Chromatogr. Sep. Tech. (2012).
[50] H.W. Darwish, S.A. Hassan, M.Y. Salem, B.A. El-Zeany, Int. J. Pharma Bio Sci. 4
(2013) 345–356.
[51] S.J. Joshi, P.A. Karbhari, S.I. Bhoir, K. Bindu, C. Das, J. Pharm. Biomed. Anal. 52
(2010) 362–371.
[52] A. Singh, T. Sivakumar, D. Jain, U. Mishra, A. Yadav, T. Sharma, Int. J. Pharm.
Res. Dev. 3 (2011) 11–22.
[53] J. Shi, Y. Chen, Y.P. Long, Cent. South Pharm. 11 (2012) 000.
[54] S. Kurbanoglu, P.R. San Miguel, B. Uslu, S.A. Ozkan, Chromatographia (2013) 1–
7.
[55] P.N. Bhoya, E.M. Patelia, J. Chromatogr. Sep. Tech. 4 (2013) 163–166.
[56] Y. Lacourcière, S. Taddei, G. Konis, H. Fang, T. Severin, J. Zhang, J. Hypertens. 30
(2012) 2047–2055.
[57] F. Belal, M. Walash, N. El-Enany, S. Zayed, J. Chromatogr. B 933 (2013) 24–29.
[58] H.W. Darwish, S.A. Hassan, M.Y. Salem, B.A. El-Zeany, Spectrochim. Acta A 113
(2013) 215–223.
[59] H.W. Darwish, S.A. Hassan, M.Y. Salem, B.A. El-Zeany, Spectrochim. Acta A 104
(2013) 70–76.
[60] M.I. Ezzeldien, E. Shokry, M.A. Fouad, R.I. Elbagary, Int. J. Adv. Chem. 1 (2013)
13–20.
[61] H.W. Darwish, S.A. Hassan, M.Y. Salem, B.A. El-Zeany, Int. J. Spectrosc. 2013
(2013) 8.
[62] H.W. Darwish, S.A. Hassan, M.Y. Salem, B.A. El-Zeany, Int. J. Pharma Bio Sci. 4
(2013) 230–243.
[63] G.K. Swamy, P. Sravanthi, M.L. Surekha, J.K.J. Rao, Int. J. ChemTech Res. 4
(2012) 1666–1673.
[64] M.M. Salim, W.M. Ebeid, N. El-Enany, F. Belal, M. Walash, G. Patonay, J. Sep. Sci.
37 (2014) 1206–1213.
[65] F.T. Chau, Y.Z. Liang, J. Gao, X.G. Shao, Chemometrics: From Basics to Wavelet
Transform, John Wiley & Sons, New Jersey, 2004.
[66] E. Dinç, M. Kanbur, D. Baleanu, Spectrochim. Acta A 68 (2007) 225–230.
[67] F.J. Anscombe, Technometrics 2 (1960) 123–146.
Contributing Editor
ABDULLAH A. AL-BADR

Founding Editor
KLAUS FLOREY