artigo validação

1. Journal of Chromatography B, 1020 (2016) 43–52
Contents lists available at ScienceDirect
Journal of Chromatography B
journal homepage: www.elsevier.com/locate/chromb
A bioanalytical HPLC method for coumestrol quantification in skin
permeation tests followed by UPLC-QTOF/HDMS stability-indicating
method for identification of degradation products
Sara E. Bianchia
, Helder F. Teixeiraa
, Samuel Kaisera
, George G. Ortegaa
,
Paulo Henrique Schneiderb
, Valquiria L. Bassania,∗
a
Programa de Pós-Graduac¸ ão em Ciências Farmacêuticas, Faculdade de Farmácia, Universidade Federal do Rio Grande do Sul, Av. Ipiranga 2752, CEP
90610-000, Porto Alegre, RS, Brazil
b
Programa de Pós-Graduac¸ ão em Química, Instituto de Química—Universidade Federal do Rio Grande do Sul, Av. Bento Gonc¸ alves 9500, CEP 91501-970,
Porto Alegre, RS, Brazil
a r t i c l e i n f o
Article history:
Received 21 October 2015
Received in revised form 15 February 2016
Accepted 13 March 2016
Available online 15 March 2016
Keywords:
Bioanalytical method
Coumestrol
Stability-indicating method
Skin permeation
Validation
UPLC-QTOF/HDMS
a b s t r a c t
Coumestrol is present in several species of the Fabaceae family widely distributed in plants. The estrogenic
and antioxidant activities of this molecule show its potential as skin anti-aging agent. These character-
istics reveal the interest in developing analytical methodology for permeation studies, as well as to
know the stability of coumestrol identifying the major degradation products. Thus, the present study
was designed, first, to develop and validate a versatile liquid chromatography (HPLC) method to quan-
tify coumestrol in a hydrogel formulation in different porcine skin layers (stratum corneum, epidermis,
and dermis) in permeation tests. In the stability-indicating test coumestrol samples were exposed to
stress conditions: temperature, UVC light, oxidative, acid and alkaline media. The degradation prod-
ucts, as well as the constituents extracted from the hydrogel, adhesive tape or skin were not eluted in
the retention time of the coumestrol. Hence, the HPLC method showed to be versatile, specific, accu-
rate, precise and robust showing excellent performance for quantifying coumestrol in complex matrices
involving skin permeation studies. Coumestrol recovery from porcine ear skin was found to be in the
range of 97.07–107.28 ␮g/mL; the intra-day precision (repeatability) and intermediate precision (inter-
day precision), respectively lower than 4.71% and 2.09%. The analysis using ultra-performance liquid
chromatography coupled to a quadrupole time-of-flight high definition mass spectrometry detector
(UPLC-QTOF/HDMS) suggest the MS fragmentation patterns and the chemical structure of the main degra-
dation products. These results represent new and relevant findings for the development of coumestrol
pharmaceutical and cosmetic products.
© 2016 Elsevier B.V. All rights reserved.
1. Introduction
Coumestrol (COU) is a phytoestrogen which belongs to the
coumestan class. It is found in several species of Fabaceae family,
like Medicago sativa, Glycine max and Trifolium pratense. The antiox-
idant [1,2] and estrogenic activities [3,4] of coumestrol revealing
the potential of this molecule for topical skin anti-aging products,
especially for post-menopause women. The estrogenic activity of
coumestrol is related to its ability to be an estrogen agonist, in
∗ Corresponding author at: Faculdade de Farmácia, Universidade Federal do Rio
Grande do Sul, Av. Ipiranga 2752, CEP 90610-000, Porto Alegre, RS, Brazil.
E-mail addresses: valquiria.bassani@ufrgs.br, valqui1@gmail.com,
valquiria@pq.cnpq.br (V.L. Bassani).
other words, it exhibits high binding affinity for the estrogen recep-
tors (ER) [4]. In the human skin, two ER receptors are commonly
distributed: ER␤ and ER␣ [5]. ER␤ is widely expressed in the epider-
mis, hair follicle, blood vessels and dermal fibroblasts, while ER␣ is
especially in to the dermal papilla cells [5,6]. The binding of agonist
to the ER is able to activate them, increasing the collagen content of
the skin, thus delaying the aging and photo aging, important effect
in women during and after menopause [6]. In addition, phytoe-
strogens have demonstrated to play an important role in wound
healing, through increasing the production-level of transforming
factor growth ␤1 (TGF-␤1) by dermal fibroblasts [7].
Some chromatographic methods for quantifying coumestrol in
different matrices have been published: in serum [8], urine [9],
plasma [10] and also in drug delivery systems [11]. High perfor-
mance liquid chromatography (HPLC) has been the most employed
http://dx.doi.org/10.1016/j.jchromb.2016.03.012
1570-0232/© 2016 Elsevier B.V. All rights reserved.

2. 44 S.E. Bianchi et al. / J. Chromatogr. B 1020 (2016) 43–52
method for determining coumestrol concentration in these sam-
ples. However, these methods are not appropriated for coumestrol
quantification in different skin layers during permeation studies.
Thus the development of a HPLC method with high sensitivity
reveals to be necessary to determine coumestrol in these com-
plex matrices. Additionally, the method should present specificity
enough to separate the analyte from interferents as constituents
from the biological samples, formulation or even analyte degrada-
tion products.
In this context, the present study was designed to develop and
validate a stability-indicating HPLC method in order to quantify the
coumestrol delivered from a hydrogel in different layers of porcine
ear skin in permeation studies. To the best of our knowledge there
is no report in the literature related to the coumestrol stability test,
as well as a method with sensibility or specificity enough to be
applied to the permeation tests, what demonstrate the originality
and the applicability of the proposed method. The results of linear-
ity, specificity, accuracy, precision and robustness of the method
are reported in order to quantify coumestrol in complex matrices.
Moreover, in stability-indication UPLC-QTOF/HDMS analysis, the
MS fragmentation patterns and the chemical structure of the main
degradation products are suggested.
2. Material and methods
2.1. Chemicals
Coumestrol (95%, HPLC purity) was obtained by Sigma-Aldrich
(Germany). Acetonitrile (HPLC grade, Tedia, USA), trifluoracetic acid
(Vetec, Brazil) and purified water (Milli-QTM system, Millipore,
USA) were used in the mobile phase of the liquid chromatography
analysis. Hydroxypropylmethylcellulose (HPMC, Methocel® K4 M)
was purchased from Blanver (São Paulo, Brazil). The polypropylene
tape strip (Scotch 750, width 13 mm) was purchased from 3 M®
and propyleneglycol from Synth (São Paulo, Brazil). Porcine ear skin
samples were obtained from a local slaughterhouse (Rio Grande do
Sul, Brazil). All the other reagents used were of analytical grade.
2.2. Stock and reference solutions
Three stock solutions (20 ␮g/mL) were prepared dissolving
2.0 mg of coumestrol in methanol into a 100 mL volumetric flask.
These coumestrol stock solutions were transferred to individual
glass flask protected from light and stored under refrigeration
(2–8 ◦C). The reference solutions of coumestrol were prepared in
appropriate concentrations by dilution of the stock solutions in
mobile phase prior to the analysis.
2.3. Complex matrices
2.3.1. Porcine ear skin
Full thickness porcine ear skin with approximately 1 mm of
thickness was used as membrane in the permeation study. The
whole skin on the back of the ear was removed from the underly-
ing cartilage and subcutaneous fat using a scalpel. The circular skin
discs were stored at −20 ◦C, and used within a month. For a purpose
of validation the circular skin discs were spiked with different con-
centrations of coumestrol reference solutions. The skins were left
to dry at room temperature, then, the skin samples were cut into
pieces, put into tube-tests and sonicated with 4 mL of methanol for
30 min.
2.3.2. Formulation
The hydrogel was prepared dispersing 3.5% (w/w) of HPMC
in water 24 h prior to the incorporation of the active compound.
Coumestrol was dispersed in 1% propylene glycol (w/w) and it was
incorporated on the preformed hydrogel. The final coumestrol con-
centration was 0.1% (w/w) used in skin permeation studies. A blank
hydrogel was prepared, without the presence of coumestrol. The
hydrogel formulation was weighed (0.500 g) and spiked with dif-
ferent concentrations of coumestrol. The extraction of coumestrol
was carried out by sonication of 30 mL of methanol for 30 min.
2.3.3. Tape striping
The coumestrol retention within stratum corneum was evalu-
ated applying the hydrogel topically on the skin and the permeation
test run in Franz-typediffusion cells. Further, the corneocyteslayers
were removed with the assistance of adhesive tape. The coume-
strol content was measured in cumulative amount present in all
the skin strips [12,13]. For validation of the tape stripping step, the
skin was similarly prepared. In this case, 14 pieces of tapes, used
in tape stripping process, were spiked onto the adhesive part of
the tape with predetermined concentrations of the reference solu-
tions. After, these 14 pieces of tapes were put into a test-tube and
sonicated with 4 mL of methanol for 30 min. The supernatant was
filtered and the coumestrol concentration determined using HPLC
method.
2.4. Apparatus and procedures
2.4.1. HPLC analysis
The development and validation of the HPLC method were per-
formed using a Shimadzu HPLC-20A equipment (Kyoto, Japan)
composed of HPLC-20AT pump, a SIL-20A autosampler and a UV/vis
variable-wavelength SPD-20AV system detector. Data acquisition
and treatment were performed with Shimadzu HPLC Solution GPC
software (Shimadzu, Japan). A C18 Phenomenex Gemini column
(150 × 4.6 mm, i.d., 5 ␮m) linked to a C18 pre-column (20 × 3.9 mm
i.d.; 10 mm) (Waters, USA) were used. The mobile phase consisted
of a mixture of acetonitrile/water with 0.1% trifluoracetic acid
(40:60, v/v), filtered through a 0.45 ␮m pore size membrane filter
and degassed for 30 min, eluted in isocratic mode. The flow rate,
detection wavelength, injection volume, and column temperature
were, respectively, 0.8 mL/min, 343 nm, 20 ␮L and 30 ◦C. All sam-
ples were filtered through Millipore PVDF membrane (0.45 ␮m of
nominal pore diameter) before injection.
2.5. Validation of the analytical method
The HPLC method was validated according to the official guide-
lines in the concentration range of 0.04–10.0 ␮g/mL. For coumestrol
quantification in the hydrogel formulation it was used official ICH
guideline [14], while for biological matrices (tape strip samples,
and porcine ear skin samples) it was used the FDA recommendation
[15]. The statistical analysis was performed using Student’s t-test
and analysis of variance (ANOVA), with significance level of 0.05.
The possible interference of degradation products generated from
forced stability indicating test of coumestrol was also evaluated in
the validation test.
2.5.1. Assessment of the matrix effect
The matrix effect was evaluated as described by Watanabe et al.
[16] and Yatsu et al. [17] using the comparison of the slopes of
analytical curves of reference coumestrol dissolved in the mobile
phase or into three complex matrices (porcine ear skin, hydrogel
formulation and adhesive tape samples). Three analytical curves
were obtained, in three consecutive days, by plotting the peak area
versus the concentration of the coumestrol standard (0.04, 0.5, 2.5,
5.0, 7.5, 10.0 ␮g/mL) in acetonitrile 40% (v/v) and in the complexes
matrices solutions. Each concentration level was analyzed in five
replicates.

3. S.E. Bianchi et al. / J. Chromatogr. B 1020 (2016) 43–52 45
The matrix effect was calculated by following equation:
ME% = 100 × [1–(Sm/Ss)], where Sm is the slope of the standard
curves of coumestrol in the mobile phase and Ss are the slopes of
the standard curves of coumestrol in the complex matrices.
2.5.2. Linearity
Analytical curves were assessed using the following coumestrol
reference solution: 0.04, 0.5, 1.0, 2.5, 5.0, 7.5 and 10.0 ␮g/mL, which
were prepared by dilution of a coumestrol stock solution in ace-
tonitrile:water (40:60, v/v). Three reference curves were obtained
through the injection of these solutions in the HPLC equipment in
triplicate, in three consecutive days. The analytical curve was eval-
uated by least squares regression, by plotting the HPLC peak area
versus coumestrol concentration.
2.5.3. Determination of the limit of quantitation (LOQ) and
detection (LOD)
The limit of detection (LOD) and limit of quantification (LOQ)
of the method were calculated from the reference curves, using
the values of standard deviation of the intercept and slope. The
lower limit of quantification (LLOQ) was experimentally defined as
the lowest concentration on the analytical curve, which was estab-
lished using six replicate preparations of coumestrol solution. The
precision and accuracy of this point were assessed also by analyzing
the replicates in three different days. Relative deviation accepted
for these parameters was less than 15% [14].
2.5.4. Specificity and stability-indicating tests
The specificity of the HPLC method was first evaluated by
injection of the three complex matrices, described above: porcine
ear skin, hydrogel formulation and adhesive tape samples, in the
absence of coumestrol. These matrices were extracted (adhesive
and porcine ear skin) or diluted (hydrogel) in methanol and son-
icated for 30 min. The responses of these complex matrices were
compared with coumestrol solution response (2.5 ␮g/mL). The
coumestrol was also identified based on its retention times, in com-
paraison to that of the coumestrol reference [11,18].
For the specificity test regarding degradation products, coume-
strol stock solutions were previously submitted to a forced
degradation using the following stress conditions: heat, UVC light,
oxidative, acid and alkaline medium. All samples were compared
with a fresh coumestrol solution 2 ␮g/mL protected from light (con-
trol) and injected in three replicates.
The thermal stability was evaluated by exposure to stress con-
ditions of the coumestrol stock solution, stored in a closed glass
flask, in an oven at 100 ◦C for 120 min. The photo stability was
analyzed by conditioning the coumestrol stock solution in trans-
parent and closed glass flasks which were placed into UVC radiation
chamber (UVC Light Express LE, 245 nm, 30 W) lined with mirrors
during 54 h. For evaluating the effect of the coumestrol chemical
degradation in oxidative, acid and alkaline medium stability, an
aliquot of 5 mL of coumestrol stock solution was added to a glass
flask, and each stress medium was tested. For this, an aliquot of
5 mL from each 29% hydrogen peroxide (H2O2), 2 M hydrochloric
acid (HCl) or 2 M sodium hydroxide (NaOH) solution was added to
the different glass flask containing coumestrol stock solution. This
mixture resulted in the concentration of 14.5% H2O2, 1 M HCl and
1 M NaOH. After 120 min, the samples tested in acid and alkaline
medium were neutralized with an aliquot of 5 mL of 2 M NaOH and
2 M HCl, respectively. After the pre-determined times, all the sam-
ples were diluted to 2 ␮g/mL in acetonitrile:water mixture (60:40,
v/v) and analyzed using a PDA detector to determine the coumestrol
peak purity.
2.5.5. Precision and accuracy
The intra-day precision (repeatability) of the method was
evaluated by analyzing the coumestrol in four concentrations
(0.04 ␮g/mL (LLOQ), 0.5, 2.5, and 7.5 ␮g/mL) in presence of the
matrices, with six determinations per concentration, on the same
day. The intermediate precision (inter-day precision) was obtained
from the evaluation of the same coumestrol concentrations during
three consecutive days, under the same experimental conditions.
The deviation was expressed as relative standard deviation (RSD).
The accuracy was determined by adding the coumestrol ref-
erence solution at four concentrations levels (0.04 ␮g/mL (LLOQ),
0.5, 2.5, and 7.5 ␮g/mL) to the previously extracted matrices solu-
tions. All samples were prepared in six replicates and analyzed
by HPLC. The accuracy was evaluated as the standardized cor-
relation between the measured value and the theoretical value,
as follows: RE% = [(mean calculated concentration–theoretical
value)/theoretical value] × 100.
2.5.6. Recovery
Coumestrol reference solutions at 0.04, 0.5, 2.5 and 7.5 ␮g/mL
were added in complexes matrices (porcine ear skin, hydrogel
formulation and adhesive tape samples). Methanol was used as
solvent, and added in each matrix test tube. The samples were pre-
pared in triplicate, maintained in an ultrasound bath for 30 min,
and filtered through a 0.45 ␮m membrane. The filtrate was ana-
lyzed according to the previously described method. The recovery
was expressed as percentages, determined comparing these results
with the values obtained from the analyses of the same concentra-
tions in the absence of the matrices.
2.5.7. Robustness
The robustness was investigated with four factors in a Plackett-
Burman design, tested with eight experiments, according to
Heyden et al. [19]. The factors analyzed for each matrix, in low lev-
els (−1) and high levels (+1), were as follows: acetonitrile content in
the mobile phase (39; 41%), flow rate (0.79; 0.81 mL/min), column
oven temperature (29; 31 ◦C), and trifluoracetic acid concentra-
tion (0.08; 0.12%). The coumestrol references solutions responses
were evaluated in relation to percentages of coumestrol obtained
in the matrices. The effects were calculated for each parameter,
where the sums of the responses are related to positive and neg-
ative levels. Statistical analysis was performed using t-test, that
allows determination of a significant effect when tcalculated > tcritical.
Additional experiments were not required, since the experimental
conditions were maintained for coumestrol references solutions
and the samples. The experimental error was estimated by a half-
normal probability plot for the effects in combination with the
dummy factors, followed by identification significant effects.
2.6. Kinetic of coumestrol degradation
The kinetics of the coumestrol degradation was determined
using mathematical models of zero-, first- and second-order kinet-
ics choosing the most linear curve presenting the highest coefficient
of determination (R2) [20]. The degradation products obtained from
stress testing were analyzed using UPLC-QTOF/HDMS in order to
identify the main degradation products.
2.7. UPLC-QTOF/HDMS analysis
A coumestrol stability-indicating study was performed using
an I-Class ACQUITY ultra performance liquid chromatography
(UPLCTM) system (Waters, USA) coupled to a quadrupole time-
of-flight (QTOF) with High Definition Mass Spectrometry (HDMS)
(SYNAPT G2 Si, Waters, USA). The UPLC analysis parameters were
adjusted from HPLC conditions with modifications. An Acquity

4. 46 S.E. Bianchi et al. / J. Chromatogr. B 1020 (2016) 43–52
UPLC HSS T3 column (2.1 × 10 mm i.d., 1.8 ␮m) (Waters, USA) was
used. The mobile phase consisted of water with formic acid 0.1%
(v/v) (A), and acetonitrile with formic acid 0.1% (v/v) (B) follow-
ing a gradient elution program: 0–1 min, from 7% B; 1–8 min, from
7 to 49% B; 8–8.1 min, from 49 to 7% B; 8.1–10 min, from 7% B
isocratic; 10 min, stop time. The flow rate, injection volume and col-
umn temperature were 0.4 mL min−1, 2 ␮L and 40 ◦C, respectively.
The analysis was performed in ESI negative MSE continuum mode,
in resolution TOF mode. The source conditions were: 2 kV of cap-
illarity; 120 ◦C and 500 ◦C of source and desolvation temperatures,
respectively; 30 V and 80 V for sampling cone and source offset,
respectively; 50 L/h and 900 L/h of cone and desolvation gases
flow, respectively; and 6.5 bar of spray gas flow. The nitrogen was
employed as desolvation gas, cone gas and nebulizer gas. The argon
was employed as collision gas at trap (20 eV) and transfer (40 eV)
collision cells. All the analyses were performed with ion mobility
(IM) employing nitrogen as IM gas at 90 mL/min flow and Helium
as cell gas at 180 mL/min. A constant wave height of 40 V and a vari-
able wave velocity of 1300–600 m/s were employed. A 50–600 m/z
mass range was monitored with a scan time of 0.3 s and inter-
scan delay of 0.015 s. A leucine-enkephalin solution (554.2615 m/z
in negative mode) was employed as lock mass solution (a refer-
ence compound). The data were processed with Mass Lynx 4.1 and
Drift-scope 2.7 software (Waters, USA).
2.8. Method aplication
The determination of coumestrol in porcine ear skin, hydrogel
formulation and adhesive tape samples were analyzed as described
in Sections 2.3.1, 2.3.2 and 2.3.3, respectively. The permeation
studies for coumestrol samples were performed using modified
Franz-type diffusion cells with approximately 2.54 cm2 of area and
10 mL of acceptor volume. The porcine ear skin was used as mem-
brane with approximately 1 mm of thickness. The circular discs
of skin were placed in phosphate buffer pH 7.4 solution during
30 min, before the experiment. The Franz-type diffusion cell was
mounted using membrane of porcine ear skin with the dermal
side downward in contact with the receptor medium. The recep-
tor fluid consisted of ethanol:phosphate buffer pH 7,4 at 40:60
(v/v) mixture. The bathing solution was controlled at tempera-
ture of 32 ± 1.0 ◦C and stirring maintained at 650 rpm. In the donor
compartment were placed approximately 500 mg of hydrogel for-
mulation, containing coumestrol at 0.1%. At the end of experiment
(8 h), and an aliquot of receptor fluid was withdrawn for analysis.
The skin was removed from the cell and carefully cleaned using a
water and cotton swab.
The stratum corneum was separated from the epidermis by tape
stripping technique, using 15 adhesive tape on skin with 1 min
interval under the weighed of 1 kg. The first withdrawal tape con-
taining excess formulation was discarded, and the other adhesive
tapes were placed in test tubes. The epidermis was separated from
the dermis using a scalpel, and the remaining dermis was cut into
small pieces. This layers, epidermis and dermis, were weighed and
placed in different test tubes. The coumestrol was extracted with
2 mL of methanol added in test tubes. Then, all samples were ana-
lyzed by HPLC.
3. Results and discussion
During the development of analytical and bioanalytical method
different proposals mobile phase were preliminarly tested,
methanol/water or acetonitrile/water. A mixture of acetoni-
trile/water was selected for further studies once showed high
chromatographic performance (average peak high, good resolution
and, peak symmetry). Trifluroacetic acid (TFA) was used to acid-
ify the mobile phase and improve peak symmetry. The optimum
mobile phase composition, which yielded a sharp coumestrol peak,
consisted of a 40:60 mixture of acetonitrile/water (the aqueous
phase containing 0.1% of TFA).
3.1. HPLC method validation
3.1.1. Assessment of the matrix effect
An important issue in the development of a bioanalytical
method is the possibility occurrence of matrix effects (ME). Thus,
matrix effects can cause different compound response when ana-
lyzed in a biological matrix compared to a standard solution. The
effect can result in a suppression or enhancement of the response
of standard [21]. The measure of quantification of matrix effect
provides additional information on the validation of bioanalytical
methods and the response should not be greater than 15% [22].
The relative matrix effect was evaluated for each coumestrol
response in the complex matrices and expressed by ME%. The
matrix effects (ME%) observed for coumestrol response were 3.41,
9.78 and 2.65%, respectively, in presence of hydrogel formulation,
porcine ear skin, and adhesive tape samples. Thus, in the proposal
method, the effect of the different matrices was lower than the
European Medicine Agency recommendation [22].
3.1.2. Linearity, precision, and accuracy
The results for linearity were performed using standard ana-
lytical curves, in three different days. The HPLC method showed
good linearity between response (area) and the correspond-
ing concentration of coumestrol over a concentration range
from 0.04 to 10.0 ␮g/mL. The calibration curve obtained was
y = 135573.09x + 969.09. The correlation of determination obtained
was satisfactory (>0.999). Constant systematic error was not
observed after analysis of the confidence intervals limits calculated
from the intercept, which included zero. Furthermore, the ANOVA
test for regression residues did not reveal linearity deviation in the
analyzed concentration range, and no significant variance between
the curves was obtained in different days (p > 0.05).
The LOD and LOQ results for coumestrol were 0.008 and
0.025 ␮g/mL, respectively. The LLOQ is described by ICH as the
lowest amount of a determined analyte in a sample that can
be quantified with precision and accuracy (RSD < 15%) [14]. This
parameter is used to quantify low concentrations of drugs in bio-
logical matrices, such as in skin permeation studies. The LLOQ for
the coumestrol, determined by the HPLC analytical method, was
0.04 ␮g/mL. Taken together, they show that the method presents
good sensitivity for the analyzed solutions.
The precision and accuracy results obtained for coumestrol
in the presence of matrices were analyzed at concentrations of
0.04 ␮g/mL (LLOQ), 0.5 ␮g/mL (lowest concentration), 2.5 ␮g/mL
(medium concentration), and 7.5 ␮g/mL (highest concentration).
The relative standard deviation (RSD) value of the intra and inter-
day precision evaluated for bioanalytical assays were, respectively,
2.09% and 4.71% for the LLOQ, 0.23% and 0.24% for the lowest
concentration, 0.21% and 0.24% for the medium concentration,
and 0.11% and 0.16% for the highest concentration. The relative
standard deviation (RSD) value of the intra and inter-day accuracy
obtained for coumestrol were, respectively, −9.98% and −14.00%
for the LLOQ, 0.52% and 0.58% for the lowest concentration, 0.27%
and 0.36% for the medium concentration, and 0.13% and −0.14%
for the highest concentration. Therefore, the bioanalytical HPLC
method can be considered precise and accurate according to
official guidelines.
3.1.3. Specificity and stability indicating test
The specificity is used to evaluate the analyte in the presence
of other compounds which can be expected. Typically these

5. S.E. Bianchi et al. / J. Chromatogr. B 1020 (2016) 43–52 47
Fig. 1. Typical HPLC chromatograms obtained in the specificity assay for the different matrices: (A) porcine ear skin, (B) adhesive tape samples (C) topical blank hydrogel,
and (D) porcine ear skin, hydrogel formulation and adhesive tape samples spiked with coumestrol solution (2.5 ␮g/mL); (1) analysis of coumestrol by UV–vis absorption
spectrum.
Fig. 2. Representative HPLC chromatograms obtained for coumestrol solution submitted to different stress conditions. (A) Coumestrol reference solution (2 ␮g/mL), (B) after
UVC radiation (254 nm, 54 h), (C), after alkaline hydrolysis (0.1 M NaOH, 120 min), and (D) diluents; UV–vis absorption spectra of (D1) peak from photolytic degradation of
coumestrol (UV spectrum 254 nm and 340 nm), and (D2) peak from alkaline degradation of coumestrol (UV spectrum 224 nm and 318 nm).
might include impurities, degradation products, some matrix con-
stituents, etc. Thus, the specificity was assayed in forced stability
indicating test and verified interferences or overlaps between the
coumestrol and porcine ear skin, hydrogel formulation or adhesive
tape samples.
HPLC chromatographic conditions resulted in coumestrol reten-
tion time of 7.08 min (Fig. 1) ␭max (maximum UV absorption)
of coumestrol peak at 243/303/343. This short run time make
the chromatographic conditions suitable for the routine analy-
sis of large number of samples from permeation studies. Thus,
the proposed method was able to quantify coumestrol without
interference of the skin endogenous constituents or formulation
ingredients, demonstrating the specificity of the method against
these possible interferers, as shown in Fig. 1. Regarding the adhe-
sive tape samples, it was observed a peak corresponding to its
constituent (retention time of 13,4 min), but it does not presented
interference in the coumestrol peak (Fig. 1C). The last one presented
good resolution (purity index >0.999) against the adhesive tape
peak and the absence of co-eluting compounds was observed.
According to Li [23], stability testing can help to enlighten the
behavior of a product in quantitative and qualitative analysis, as
well as contribute for the knowledge of the degradation pathways
of the molecule. In this paper, coumestrol was tested against the fol-
lowing stress conditions: heat, UVC light, oxidative, acid or alkaline
media.
No significant changes were observed after exposure of coume-
strol in acid or oxidative media and heat conditions. In contrast, the
coumestrol degraded gradually in alkaline medium and revealed
clear sensitivity to UVC-irradiation, as is shown in Fig. 2. In alkaline
medium, besides the coumestrol peak, a new peak can be observed
at 3.21 min, which did not interfere in the coumestrol response. This
observation is in accordance with previous reports on flavonoids
degradation in alkaline medium [24–28]. Continuous UVC irradi-
ation of coumestrol, solubilized in methanol, resulted in different
peaks, but they also did not interfere in the coumestrol peak. The
PDA detector showed that the coumestrol peak presented high
purity even the presence of these degradation products (Fig. 2).
3.1.4. Kinetics of degradation and degradation products
In order to determine the degradation kinetics parameters,
it is essential to achieve 50% of degradation [23]. Thus samples
degraded in UVC light exposure or alkaline medium were ana-
lyzed using the developed HPLC method. The results showed that
coumestrol degradation against these stress conditions followed,

6. 48 S.E. Bianchi et al. / J. Chromatogr. B 1020 (2016) 43–52
Fig. 3. MS spectra for coumestrol reference solution (a), in 2D chromatogram of the contour plot from retention time vesus drift time of coumestrol solution response. Mass
spectrum [M−H]−
m/z 267 (b), in retention time 7.06 min, QTOF/HDMS products ions spectra in ESI negative MSE
(c), and fragmentation scheme proposed to coumestrol (d).
Table 1
Degradation kinetics of coumestrol in alkaline media and UVC light.
Mathematical models UVC light Alkaline media
Equation R2
Equation R2
Zero order y = −0.0264x + 2.303 0.9816 y = −0.0085x + 1.963 0.9347
First order y = −0.0176x + 0.894 0.9644 y = −0.0058x + 0.700 0.9619
Second order y = 0.0123x + 0.357 0.9251 y = 0.0042x + 0.471 0.9448
respectively, zero order (R2 0.9816) and first order (R2 0.9619)
kinetics (Table 1).
The kinetic degradation parameters for coumestrol against
UVC light exposure were: degradation rate constant (k) of
0.026 ␮g/mL h, half-life (T1/2) 43,69 h and time of 10% decompo-
sition (T90%) of 8,74 h. The parameters of the degradation kinetic
of the coumestrol in alkaline medium were: degradation rate con-
stant (k) of 0.006 ␮g/mL min, half-life of T1/2 118,71 min and time of
10% decomposition (T90%) of 18,15 min. Thus, its kinetic degrada-
tion in both stress condition, UVC light and alkaline medium were

7. S.E. Bianchi et al. / J. Chromatogr. B 1020 (2016) 43–52 49
Fig. 4. MS spectra for coumestrol degraded in alkaline medium (a), in 2D chromatogram of the contour plot from retention time vesus drift time of coumestrol solution
response. MS products ions spectra of degraded in alkaline medium with retention time 5.08 min (b), and fragmentation scheme proposed coumestrol degradation product
in alkaline medium (c) (ESI negative MSE
).
slower than that reported to other flavonoids as the isoflavones
daidzein and genistein [28].
In order to characterize the degradation products in the coume-
strol stability-indicating tests, the samples were analyzed using
UPLC-QTOF/HDMS. Fig. 3(a) shows a 2D chromatogram of the
contour plot from retention time versus drift time of coumestrol
response (2.5 ␮g/mL). Bacaloni et al. [29] and Simons et al. [30]
showed some differences in the fragmentation patterns for coume-
strol, when the analysis were carried out in positive mode. Analysis
of coumestrol, in negative mode, showed an intense molecular ion
with an m/z 267 [M−H]− in a retention time of 7.06 min (Fig. 3b).
The proposed fragmentation ions of coumestrol [M−H]− m/z 267
are shown in Fig. 3(d). The ions resulting in MS fragmentation spec-
trum are: the loss of H· forming the ion m/z 266 with a conjugated
radical anion. After, the product ion mass spectrum m/z 239 is
formed by loss of CO. Two structures have been proposed to m/z
239, in the first case due to loss of CO from the ester group [31]
and in the second case by the loss of CO in phenol group, as shown
in Fig. 3(d). The fragment [M−H CO]− m/z 239 results in a frag-
ment peak m/z 211 produced by loss of CO at the phenol group,
for both structures mentioned above. In sequence, the fragment
[M−H CO CO]−m/z 211 results in a product ion m/z 167 by loss
of CO2 from the inner ester group. This structure suggests that the
previous elimination of CO is not from ester group, in agreement
with the proposed by Bickoff et al. [31], but from phenol group. It
was also possible to observe an m/z 233 ion produced by loss of CO2
at the ester group from the precursor [M−H]− m/z 267 ion. On the
other hand, it was suggested another route fragmentation in which
the m/z 233 ion was formed from the fragmentation of m/z 239, and
after the m/z 195 ion by the loss of CO from phenol group.
Furthermore, coumestrol samples (solution) submitted to the
alkaline medium and UV light was analyzed. No interfering sig-
nals (m/z) were found in the same retention time of coumestrol.
Fig. 4(a) show a 2D chromatogram of coumestrol degraded in alka-
line medium. The retention time observed was in 5.08 min, with
the main fragments ions at m/z 285/241. The MS spectrum in neg-
ative mode is presented in Fig. 4(b) and the possible structures of
the product can be observed in Fig. 4(c). It is necessary emphasize
the formation of adducts with formic acid used in the mobile phase
for the analysis in UPLC-QTOF/HDMS. The fragment m/z 285 ion is
a formate adducts [M + CHO2]−, formed by fragment m/z 239 ion,
as well as the m/z 241 ion that is a formate adducts of fragment m/z

8. 50 S.E. Bianchi et al. / J. Chromatogr. B 1020 (2016) 43–52
Fig. 5. MS spectra for coumestrol degraded by UVC-irradiation (a), in 2D chromatogram of the contour plot from retention time vesus drift time of coumestrol solution
response. MS products ions spectra of degraded in UVC-irradiation with retention time (b) 7.17, (c) 7.56, (d) 7.72, and (e) 8.12 min. Proposed fragmentation scheme for
coumestrol degraded by UVC-irradiation (f) (ESI negative MSE
).

9. S.E. Bianchi et al. / J. Chromatogr. B 1020 (2016) 43–52 51
195 ion. Fig. 4(c) suggests the formation of two formate adducts,
in relation to fragments presented, m/z 239 and m/z 195 ion, from
Fig. 3(d).
Fig. 5(a) show a 2D chromatogram of coumestrol degraded in
UVC-irradiation. The fragments ions obtained for coumestrol can be
observed in the retention times of 7.17, 7.56, 7.72, and 8.12 min, as
shown Fig. 5(b)–(e), respectively. Some possible structures formed
with fragments ions observed at m/z 533/357/345/311/297/157
had its supposed structures proposed in Fig. 5(f). Likewise in alka-
line medium, two fragments ions with m/z at 285 and 241 were
also observed for coumestrol degraded in UVC-irradiation, both for-
mate adducts (Fig. 4c). Interestingly, the sample submitted to UVC
light degradation showed the formation of product with m/z 533,
which could correspond to a coumestrol dimer. Subsequent frag-
ments seem to be the result of degradation of this dimer formed,
and they are: m/z 357/311/297 ions. Furthermore, it was observed
another dimer formed m/z 345 from the fragmentation of coume-
strol [M−H]−m/z 267 ion, and after the degradation of this dimer
with m/z 157. It is worth mentioning that this is the first reports
on stress degradation studies of the coumestrol, as well as the elu-
cidation of the fragmentation pathway of this compound against
alkaline medium or UVC light exposure.
3.1.5. Recovery
The recovery of coumestrol from the complex matrices (porcine
ear skin, hydrogel formulation and adhesive tape samples) were
evaluated on four concentrations (0.04, 0.5, 2.5 and 7.5 ␮g/mL). The
results are demonstrated in Table 2. The recovery of coumestrol in
the different concentrations was in the range of 97.07–107.28% for
porcine ear skin samples, 98.18–103.52% for hydrogel formulation
and 98.97–101.01% for adhesive tape samples. All recovery values
were appropriate and the matrices components did not interfere
on the coumestrol quantification. The recovery, at the lower limit
of quantitation (0.04 ␮g/mL) was according the FDA recommenda-
tions for bioanalytical method validation [15].
3.1.6. Robustness
The robustness of the byoanalytical method is a test that can
examine the potential sources of variability in the analysis condi-
tions [19]. Plackett–Burman is a way to construct a multivariate
design, particularly because it allows analyzing a number of differ-
ent factors simultaneously, with a reduced number of experiments
[32].
The Fig. 6 shows the percentage of coumestrol obtained in
each complex matrices sample in relation to references solutions
responses in the Plackett-Burman design. These results indicated
that no significance differences were observed for all analyses.
The t-values calculated were lower than t-critical values (␣ = 0.05),
which indicated that no significant changes in the analysis results in
relationship to percentage of coumestrol obtained, although shifts
in retention times were observed after each analysis. Therefore,
changes made in the experimental conditions did not interfere in
chromatographic profile, demonstrating the robustness of the pro-
posed byoanalytical method.
3.2. Method application
The HPLC bioanalytical method was applied to determine the
coumestrol amount in skin permeation studies from hydrogel for-
mulation. The results indicate the precision of the method in all
samples, with a standard deviation lower than 14.99%.
Quantification of coumestrol in hydrogel formulation presented
RSD of 1.87% (range of 95.10–99.72%), indicating adequate repeata-
bility. In relation to receptor fluid no coumestrol was detected after
8 h of in vitro skin permeation studies. Conversely, the coume-
strol quantification in the stratum corneum resulted in RSD of
Fig. 6. Bar charts representing the calculated effects (%) for quantitative determi-
nation of the studied factors in Plackett-Burman experimental design (t-calculated)
and their critical effect (t-critical), represented by the vertical line, for coumestrol
in the complex matrices, where (a) is the hydrogel formulation, (b) is the porcine
ear skin, and (c) is the adhesive tape samples.
13.40% (range of 0.175–0.278 ␮g/cm2), indicating adequate preci-
sion. Finally, coumestrol detected in skin samples (epidermis and
dermis) resulted in RSD of 14.99% (range of 0.105–0.180 ␮g/cm2).
In addition, the method was able to demonstrate the quantifying of
coumestrol in skin permeation studies. These results indicate the
precision of the validated bioanalytical method.
4. Conclusion
The proposed HPLC bioanalytical method showed to be suit-
able for the coumestrol quantification in complex matrices such
as different layers of porcine ear skin, in hydrogel formulation and
adhesive tape samples involved in permeation studies. The method
revealed to be linear, specific, accurate, precise and robust accord-

10. 52 S.E. Bianchi et al. / J. Chromatogr. B 1020 (2016) 43–52
Table 2
Recovery (%) of the coumestrol in hydrogel formulation, porcine ear skin and adhesive tape samples.
Coumestrol(␮g/mL) Skin samplesa
(␮g/mL) Hydrogel formulationa
(␮g/mL) Adhesive tapea
(␮g/mL)
0.04 107.28 ± 9.57 103.52 ± 1.92 100.85 ± 3.19
0.5 97.07 ± 3.18 99.42 ± 2.33 99.02 ± 2.71
2.5 97.13 ± 1.49 103.33 ± 2.41 98.97 ± 4.40
7.5 100.51 ± 2.17 98.18 ± 1.27 101.01 ± 2.87
a
Expressed as mean ± standard deviation (s.d.).
ing to the ICH guidelines. The stability-indicating tests coupled to
UPLC-QTOF/HDMS analysis resulted in a new and relevant contri-
bution for elucidating the MS fragmentation pattern and chemical
structures of the main coumestrol degradation products.
Acknowledgements
The authors are grateful to the Brazilian Government: Coor-
denacão de Aperfeic¸ oamento de Pessoal de Nível Superior,
Conselho Nacional de Desenvolvimento Científico e Tecnológico
and Fundac¸ ão de Amparo à Pesquisa do Rio Grande do Sul for the
financial support and scholarships.
References
[1] H.Y. Jeon, D.B. Seo, H.J. Shin, S.J. Lee, Effect of aspergillus oryzae-challenged
germination on soybean isoflavone content and antioxidant activity, J. Agr.
Food Chem. 60 (2012) 2807–2814.
[2] H.I. Lee, J.H. Lee, K.H. Park, D. Sangurdekar, W.S. Chang, Effect of soybean
coumestrol on bradyrhizobium japonicum nodulation ability, biofilm
formation, and transcriptional profile, Appl. Environ. Microb. 78 (2012)
2896–2903.
[3] K. Ndebele, B. Graham, P.B. Tchounwou, Estrogenic activity of coumestrol,
DDT, and TCDD in human cervical cancer cells, Int. J. Envtron. Res. Publ.
Health 7 (2010) 2045–2056.
[4] C. Chandsawangbhuwana, M.E. Baker, 3D models of human ER alpha and ER
beta complexed with coumestrol, Steroids 80 (2014) 37–43.
[5] M.J. Thornton, A.H. Taylor, K. Mulligan, F. Al-Azzawi, C.C. Lyon, J. O’Driscoll,
A.G. Messenger, The distribution of estrogen receptor beta is distinct to that of
estrogen receptor alpha and the androgen receptor in human skin and the
pilosebaceous unit, J. Invest. Dermatol. Symp. Proc. 8 (2003) 100–103.
[6] G. Hall, T.J. Phillips, Estrogen and skin: the effects of estrogen, menopause,
and hormone replacement therapy on the skin, J. Am. Acad. Dermatol. 53
(2005) 555–568.
[7] E. Emmerson, L. Campbell, G.S. Ashcroft, M.J. Hardman, The phytoestrogen
genistein promotes wound healing by multiple independent mechanisms,
Mol. Cell. Endocrinol. 321 (2010) 184–193.
[8] J.K. Prasain, A. Arabshahi, D.R. Moore, G.A. Greendale, J.M. Wyss, S. Barnes,
Simultaneous determination of 11 phytoestrogens in human serum using a
2 min liquid chromatography/tandem mass spectrometry method, J.
Chromatogr. B 878 (2010) 994–1002.
[9] A.A. Franke, L.J. Custer, High-Performance liquid-chromatographic assay of
isoflavonoids and coumestrol from human urine, J. Chromatogr. B 662 (1994)
47–60.
[10] T.J.O. Lundh, H. Pettersson, K.H. Kiessling, Liquid-chromatographic
determination of the estrogens daidzein, formononetin, coumestrol, and
equol in bovine blood-plasma and urine, J. Assoc. Off Anal. Chem. 71 (1988)
938–941.
[11] D.F. Argenta, C. Franco, L.S. Koester, V.L. Bassani, H.F. Teixeira, LC analysis of
coumestrol incorporated into topical lipid nanoemulsions, Pharmazie 66
(2011) 929–932.
[12] J. Lademann, U. Jacobi, C. Surber, H.J. Weigmann, J.W. Fluhr, The tape stripping
procedure–evaluation of some critical parameters, Eur. J. Pharm. Biopharm.
72 (2009) 317–323.
[13] W.L. Au, M. Skinner, I. Kanfer, Comparison of tape stripping with the human
skin blanching assay for the bioequivalence assessment of topical clobetasol
propionate formulations, J. Pharm. Pharm. Sci. 13 (2010) 11–20.
[14] I.C.H. (ICH), Technical requirements for the registration of pharmaceutical for
humanuse,Validation of Analytical Procedures: Text and Methodology Q2
(R1), 2005, pp. 1–13.
[15] F.a.D.A. (FDA), Center for drug evaluation and research, Guidance for Industry:
Bioanalytical Method Validation, (2013) 1–34.
[16] E. Watanabe, Y. Kobara, K. Baba, H. Eun, Aqueous acetonitrile extraction for
pesticide residue analysis in agricultural products with HPLC-DAD, Food
Chem. 154 (2014) 7–12.
[17] F.K.J. Yatsu, L.S. Koester, I. Lula, J.J. Passos, R. Sinisterra, V.L. Bassani, Multiple
complexation of cyclodextrin with soy isoflavones present in an enriched
fraction, Carbohydr. Polym. 98 (2013) 726–735.
[18] C. Franco, L. Schwingel, I. Lula, R.D. Sinisterra, L.S. Koester, V.L. Bassani,
Studies on coumestrol/beta-cyclodextrin association: inclusion complex
characterization, Int. J. Pharm. 369 (2009) 5–11.
[19] Y. Vander Heyden, A. Nijhuis, J. Smeyers-Verbeke, B.G.M. Vandeginste, D.L.
Massart, Guidance for robustness/ruggedness tests in method validation, J.
Pharm. Biomed 24 (2001) 723–753.
[20] S. Yoshioka, V.J. Stella, Stability of Drugs and Dosage Forms, Kluwer Academic
Publishers, New York, 2002.
[21] L. Cuadros-Rodriguez, A.M. Garcia-Campana, E. Almansa-Lopez, F.J.
Egea-Gonzalez, M.L.C. Cano, A.G. Frenich, J.L. Martinez-Vidal, Correction
function on biased results due to matrix effects—application to the routine
analysis of pesticide residues, Anal. Chim. Acta 478 (2003) 281–301.
[22] E.M.A. EMEA/CHMP/EWP/192217/2009, Guideline on bioanalytical method
validation, (2011) 1–22.
[23] M. Li, Organic Chemistry of Drug Degradation, RSC Drug Discovery Series,
2012.
[24] Q.Y. Zhu, R.R. Holt, S.A. Lazarus, J.L. Ensunsa, J.F. Hammerstone, H.H. Schmitz,
C.L. Keen, Stability of the flavan-3-ols epicatechin and catechin and related
dimeric procyanidins derived from cocoa, J. Agr. Food Chem. 50 (2002)
1700–1705.
[25] Y. Zheng, I.S. Haworth, Z. Zuo, M.S.S. Chow, A.H.L. Chow, Physicochemical and
structural characterization of quercetin-beta-cyclodextrin complexes, J.
Pharm. Sci.-Us 94 (2005) 1079–1089.
[26] J.B. Zvezdanovic, J.S. Stanojevic, D.Z. Markovic, D.J. Cvetkovic, Irreversible
UV-induced quercetin and rutin degradation in solution studied by UV
spectrophotometry and HPLC chromatography, J. Serb. Chem. Soc. 77 (2012)
297–312.
[27] J.P. Carini, S. Kaiser, G.G. Ortega, V.L. Bassani, Development, optimisation and
validation of a stability-indicating HPLC method of achyrobichalcone
quantification using experimental designs, Phytochem. Anal. 24 (2013)
193–200.
[28] Y. Ungar, O.F. Osundahunsi, E. Shimoni, Thermal stability of genistein and
daidzein and its effect on their antioxidant activity, J. Agr. Food Chem. 51
(2003) 4394–4399.
[29] A. Bacaloni, C. Cavaliere, A. Faberi, P. Foglia, R. Samperi, A. Lagana,
Determination of isoflavones and coumestrol in river water and domestic
wastewater sewage treatment plants, Anal. Chim. Acta 531 (2005) 229–237.
[30] R. Simons, J.P. Vincken, M.C. Bohin, T.F.M. Kuijpers, M.A. Verbruggen, H.
Gruppen, Identification of prenylated pterocarpans and other isoflavonoids in
Rhizopus spp. elicited soya bean seedlings by electrospray ionisation mass
spectrometry, Rapid Commun. Mass Spectrom. 25 (2011) 55–65.
[31] E.M., Bickoff, R.R., Spencer, S.C., Witt, B.E. Knuckles, Studies on the Chemical
and Biological Properties of Coumestrol and Related Compounds,
Washington, 1969.
[32] M. Zeaiter, J.M. Roger, V. Bellon-Maurel, D.N. Rutledge, Robustness of models
developed by multivariate calibration. Part I: The assessment of robustness,
TrAC-Trends Anal. Chem. 23 (2004) 157–170.