simultaneous determination of atenolol, propranolol, dipyridamole and amiloride by means of...
TRANSCRIPT
Simultaneous determination of atenolol, propranolol, dipyridamoleand amiloride by means of non-linear variable-angle synchronous
¯uorescence spectrometry
Jose A. Murillo PulgarõÂn*, Aurelia AlanÄoÂn Molina, Pablo FernaÂndez LoÂpez
Department of Analytical Chemistry and Foods Technology, University Castilla La Mancha, 13071 Ciudad Real, Spain
Received 3 February 1998; received in revised form 8 April 1998; accepted 14 April 1998
Abstract
First derivative non-linear variable-angle synchronous ¯uorescence spectrometry has been developed to improve the
selectivity of ¯uorescence measurements without loss of sensitivity. It allows the rapid simultaneous determination of different
substances in a mixture from a single spectrum based on a single scan. This method was applied for the simultaneous
determination of atenolol (ATE), propranolol (PRO), amiloride (AMI) and dipyridamole (DIP) at concentrations between 10±
400, 6±200, 5.6±280 and 5±100 ng mlÿ1, respectively, by means of absolute values of ®rst derivative of non-linear variable-
angle synchronous scan at �ex/�em�228.8/300, 287.2/340, 366.4/412.8 and 288/487.2 nm for ATE, PRO, AMI and DIP,
respectively. In order to obtain maximum sensitivity and an adequate selectivity, factors affecting ¯uorescence intensity were
studied. As a result, the analyses were performed in an ethanol±water (70%(v/v)) medium at a pH 7.5, adjusted by using
trishydroxymethyl amino methane (0.08 M) as a buffer solution. Analytical parameters of the proposed method were
calculated according to the error propagation theory. The sensitivity, repeatability, reproducibility and limit of detection
achieved with the proposed method are adequate for the determination of these doping substances. # 1998 Elsevier Science
B.V. All rights reserved.
Keywords: Atenolol; Propranolol; Amiloride; Dipyridamole; Fluorimetry
1. Introduction
Atenolol (ATE), propranolol (PRO), amiloride
(AMI) and dipyridamole (DIP) are widely used in
the treatment of several diseases. It is well known that
the fraudulent consumption of this type of substance is
quite common these days, therefore they have been
included in the list of forbidden substances [1] by the
International Olympic Committee. ATE, PRO and
AMI [2] are antihypertensive substances, whereas
DIP [2] is an antithrombotic agent. ATE and PRO
are beta-adrenoceptors that are abused in sports with
little physical activity. These compounds, act on the
heart reducing the cardiac frequency, the contraction
force and the coronary ¯ow. AMI is a weak diuretic
which retains potassium, since it promotes sodium
excretion and potassium reabsorption. Further, low
doses of AMI lead to a high-volume urine excretion
producing a diminution of arterial tension since a great
quantity of electrolytes is eliminated. DIP can act as a
Analytica Chimica Acta 370 (1998) 9±18
*Corresponding author. Fax: 0034 9 26295318.
0003-2670/98/$19.00 # 1998 Elsevier Science B.V. All rights reserved.
P I I S 0 0 0 3 - 2 6 7 0 ( 9 8 ) 0 0 2 6 4 - 5
vasodilator which stimulates a rise in the blood ¯ow
through the coronary circulation. This vasodilator
agent is classi®ed in doping terms as stimulant.
Fluorescence spectrometry is widely used in quan-
titative analysis because of its great sensitivity and
selectivity as well as its relative low cost. This tech-
nique has not, however, been widely applied to the
simultaneous direct determination of several ¯uores-
cent components in mixtures, mainly because the
¯uorescence spectra of individual substances contain
broad bands which often overlap.
Several methods have been proposed to resolve
such problems without manipulation of the samples
or using time-consuming and highly expensive separa-
tion techniques. Among them, synchronous [3] and
derivative [4] ¯uorescence spectrometry are the most
popular. Synchronous spectrometry consists essen-
tially of the simultaneous scanning of both mono-
chromators, while maintaining a constant wavelength
interval (��) between them. The simpli®cation of the
spectral pro®le together with the reduction of band
width are its main characteristics [5,6]. The synchro-
nous spectrum which is a 458 section through the
contour map is of limited use.
To achieve greater selectivity, variable-angle ¯uor-
escence spectrometer was developed [7]. The vari-
able-angle ¯uorescence spectra are obtained by
scanning the monochromators (excitation and emis-
sion) at different speeds, thus the wavelength differ-
ence between them is not constant. Because the
recorded planes are not restricted to 458, any path
can be traced, depending on the sample under study.
With variable-angle ¯uorimetry, several compounds
can be determined simultaneously from a single spec-
trum.
In the non-linear variable-angle synchronous ¯uor-
escence (NLVASF) method the trajectory of the scan is
varied continuosly through the excitation and emis-
sion matrix. This technique offers several possibilities:
1. in a complex system, the maximum and minimum
emission intensities can be explored by traversing
the peaks and valleys;
2. a curved trajectory can be followed through the
emission±excitation matrix, allowing light scatter-
ing peaks to be avoided;
3. overlapping systems that cannot be resolved by
linear scanning can be resolved by this technique.
Nowadays, this technique, in its pure concept, has
not been applied ever due to the mechanical dif®culty
that entrails the synchronization of the movement
of the excitation and emission monochromators.
Therefore, several methods described in bibliography
intend to take advantage of the characteristics
described above, realizing a synchronous scan which
is a combination of several linear variable syn-
chronous scans with different angles [8±11]. The
resulting spectra represent the intensity pro®le of a
cut through the excitation±emission matrix, the tra-
jectory of which is a continuous function. It is not
possible to take the derivative because of the different
values of the limits in the point where the trajectory
angle is varied. Therefore an appropriate empirical
equation involving the effect of each compound or
a further treatment of the data by a multilinear
regression program [10], for resolving complex
mixtures may be needed. This scan is obtained by
means of the Ftotal program [12], which permits
the theoretical NLVASF scan to be obtained from
the data stored in the contour map (the curved trajec-
tory can be varied through the emission±excitation
matrix, to describe any desired path). This step helps
to optimize the route that will produce the best
NLVASF scanning spectra (highest signal value, smal-
lest band width at half-maximum intensity and inter-
ference-free bands).
The derivative ¯uorescence technique, as described
by Green and O'Haver [13], can also be used to
enhance minor spectral features.
The combination of synchronous scanning spectro-
¯uorimetry with derivative techniques is advanta-
geous in terms of sensitivity, when compared with
results obtained by differentiation of the conventional
spectrum, because the amplitude of the derivative
signal is inversely proportional to the band width of
the original spectrum. Moreover, the selectivity of
spectro¯uorimetric analysis is greatly improved by
using both techniques in conjunction, as suggested
by John and Soutar [4] and as recently reported in
[3,14±21].
In this work, a method for analyzing a mixture of
¯uorescent doping substances with broad and highly
overlapped excitation and emission spectra is
described. The application of the ®rst derivative
NLVASF to multicomponent analysis provides a clear
example of the high resolving power of this technique
10 J.A. Murillo PulgarõÂn et al. / Analytica Chimica Acta 370 (1998) 9±18
without resorting to expensive or time-consuming
procedures.
2. Experimental
2.1. Apparatus
All ¯uorimetric measurements were performed on
an Aminco Bowman Series 2 connected to the `̀ AB2''
Software which runs on the OS2 operating system.
The instrument is equipped with a continuous 150 W
Xenon lamp. Quartz glass cuvettes with a pathlength
of 1.0 � 1.0 cm have been used.
2.2. Software
The AB2 program allows the instrument control,
operation and acquisition of excitation, emission and
total ¯uorescence spectra.
The Ftotal program [12] provides the spectral char-
acterization of analytical interest of the ¯uorescence
of any luminescent compound by generating informa-
tion through the isometric representation of the three-
dimensional spectrum as a level curve. Besides, it
processes the spectral data to obtain any type of
bidimensional spectrum. For the NLVASF spectrum
Ftotal obtains the intensity values by applying Lagran-
ge's interpolation method to the excitation wave-
lengths calculated by means of the particular
function that relates emission wavelengths at 0.8 nm
intervals with the excitation wavelengths since they
are not equidistant.
The statistical analysis is totally covered by means
of a program that has a menu which includes proce-
dures such as least median of squares regression
(detection of outlier and leverage points), least squares
regression with and without replicates, weighted least
squares regression, tests of regression and correlation,
detection and determination limits (IUPAC [27,28],
Error Propagation Theory [29,30], and Clayton et al.
[31]), ellipse graph for the 95% con®dence region for
the true slope and intercept on the y-axis estimated
from the regression method, dispersion and con®dence
bands for the calibration graph and ANOVA test for
linearity and for comparison of several regression
lines.
2.3. Reagents
All experiments were performed with analytical
reagent grade chemicals, pure solvent and Milli-Q
water. The stock solution of ATE (Aldrich, 125 mg
dissolved in 500 ml water), PRO (Aldrich, 125 mg
dissolved in 500 ml water), AMI (Sigma, 125 mg
dissolved in 500 ml water) and DIP (Aldrich,
125 mg dissolved in 500 ml ethanol) were diluted
(DIP with ethanol) to prepare standard solutions by
suitable dilutions. The stock solutions were stored and
protected from light. Under these conditions, the
solutions were stable for at least three months.
Working samples of ATE, PRO, AMI and DIP were
stable for at least 2 h at room temperature.
Pharmaceutical preparations of Normopresil
(Serma S.A., Barcelona, Spain), Atenolol Leo (Byk
Elmu S.A., Madrid, Spain), Neatenolol diu (Fides-
Rottapharma, S.A., Valencia, Spain), Sumial 10,
Ameride (Du Pont Pharma, S.A., Madrid, Spain),
Kalten (Zeneca Farma, S.A., Pontevedra, Spain) and
Persantin (Boehringer Ingelheim, S.A., Barcelona,
Spain) with different nominal contents, were ran-
domly purchased from local pharmacies.
2.4. Procedure
2.4.1. Sample preparation
For the preparation of the calibration graph, place
an aliquot of ATE, PRO, AMI and DIP equivalent to
0.25±10, 0.15±5, 0.14±7 and 0.125±2.5 mg, respec-
tively, in a 25 ml volumetric ¯ask, add 2 ml of buffer
solution (pH 7.5), 17.5 ml of analytical grade ethanol,
and dilute with water to a ®nal volume of 25 ml and
shake.
Record for each sample 61 emission spectra of
384 nm width in steps of 0.8 nm, varying the excita-
tion wavelength in 6.4 nm steps. Obtain the total
luminescence spectra by means of the Ftotal program
[12]. Apply the function that passes through the
excitation and emission maxima selected for ATE,
PRO, AMI and DIP to obtain the NLVASF spectrum.
Calculate the ®rst derivative, according to the Savitzky
and Golay algorithm [22,23]. The absolute values of
the ®rst derivative of these synchronous spectra were
measured at �ex/�em�228.8/300, 287.2/340, 366.4/
412.8 and 288/487.2 nm for determination of ATE,
PRO, AMI and DIP, respectively.
J.A. Murillo PulgarõÂn et al. / Analytica Chimica Acta 370 (1998) 9±18 11
For the analysis of the formulations: Normopresil,
Atenolol Leo, Neatenolol diu, Sumial 10, Ameride,
Kalten and Persantin introduce three tablets of each
formulation in three different 1 l volumetric ¯asks.
Add water suf®ciently to dissolve them. Shake and
maintain them in an ultrasonic bath for 15 min. Dilute
with water to a ®nal volume of 1 l. Persantin, which
contains DIP, must be dissolved and diluted with
ethanol.
In both cases, the assay was completed as described
for calibration graph. The percentage recovery of
every compound is computed from regression equa-
tions for pure drugs.
3. Results and discussion
3.1. Spectral characteristics
The best characterization of the ¯uorescence of the
compounds is achieved by obtaining the three-dimen-
sional spectrum. This spectrum can be obtained and
presented as the isometric projection, where the exci-
tation spectra at stepped increments of emission wave-
length are recorded and plotted, as shown in Fig. 1(a)±
(d), where Rayleigh scattering has been removed. A
reversed projection (where emission spectra are
plotted at decreased excitation wavelength) of the
data can sometimes indicate emission peaks hidden
by the foreground. Alternatively, the three-dimen-
sional spectra can be effectively transformed to a plot
in two dimensions of the excitation and emission
wavelength by linking points of equal intensity to
form contours. In general, this contour presentation
is more useful than the isometric projection for indi-
cating the presence of hidden emission peaks and
providing the selection of the best trajectory to obtain
optimum results in the application of the synchronous
scan technique.
Fig. 2 shows the total ¯uorescence spectra of ATE
((A), solid line), PRO ((B), broken line), AMI ((C),
solid line) and DIP ((D), broken line). As can be
Fig. 1. Isometric plot of the excitation±emission matrix (backward projection) of (a) ATE, (b) PRO, (c) AMI and (d) DIP.
12 J.A. Murillo PulgarõÂn et al. / Analytica Chimica Acta 370 (1998) 9±18
observed, owing to spectral overlapping, the analysis
of mixtures of the four compounds would not be
feasible by conventional spectro¯uorimetry at their
wavelength maxima.
3.2. Influence of experimental variables
Chemical variables were studied and optimized to
obtain the best measurement conditions, maximum
¯uorescence sensitivity and an adequate selectivity.
ATE, PRO and AMI are soluble in water, but DIP is
not, and consequently, it was necessary to use an
organic solvent. DIP is highly soluble in ethanol,
therefore it was necessary to study the variation in
the ¯uorescence of ATE, PRO, AMI and DIP with
changes of the ethanol percentage. In this way the
effect of ethanol content in the medium was investi-
gated by preparing samples varying the ethanol per-
centage between 10% and 90%(v/v). In all cases, the
¯uorescence intensity increases with the percentage of
ethanol. A 70% of ethanol was selected as adequate.
The in¯uence of pH on the ¯uorescence intensity
was studied by adding different amounts of HCl and
NaOH. Hardly any variation is observed in the ¯uor-
escence intensity of ATE, PRO and AMI, while the
¯uorescence intensity of DIP is constant up to a pH
value of 4.0. It strongly increases for pH values
between 4.0 and 6.5 and the maximum ¯uorescence
intensity is reached for apparent pH values higher than
6.5. An apparent pH value of 7.7 is adequate for the
determination because the maximum intensity is
achieved. The pH selected was adjusted by adding
trishydroxymethyl amino methane buffer solution.
This buffer was soluble in the ethanol medium and
besides, the ¯uorescence intensity of ATE, PRO, AMI
and DIP was not affected by this and its concentration.
A 0.08 M concentration of the buffer was therefore
selected to get a suf®cient buffering capacity.
Another factor that affects the ¯uorescence inten-
sity is the temperature. The ¯uorescence intensity
showed a decrease when the temperature increases
from 58C to 658C. The temperature coef®cient is
obtained by performing the least squares regression
of the relative signal increment versus the tempera-
ture. The slope is the temperature coef®cient. The
results were 0.6%, 0.5%, 1.2%, and 0.2% 8Cÿ1 for
ATE, PRO, AMI and DIP, respectively. ATE, PRO and
DIP show a clear linear behavior with the temperature,
while AMI does not. Therefore, the use of a thermostat
is recommended and a measurement temperature of
208C was chosen.
The in¯uence of analytes concentration on the
¯uorescence intensity was studied under these condi-
tions. The best range for the relation ¯uorescence
intensity versus concentration was up to 400, 200,
280 and 100 ng mlÿ1 for ATE, PRO, AMI and DIP,
respectively.
3.3. Selection of optimum route
The major dif®culty encountered in the application
of the synchronous scan technique is that the best route
must be known beforehand to get optimum results.
Consequently, a detailed inspection of each of the
contour maps is necessary. The optimum route is
immediately obvious even to relatively inexperienced
operators. Moreover, the advances in analytical instru-
ments and the use of computers permits to obtain
analytical results (contour maps and its treatment to
obtain the ®rst derivative of NLVASF scan) in less than
10 min.
To determine the optimum NLVASF scan route, a
careful examination of a contour map corresponding
to a standard solution of the four compounds at
concentration levels which produce equal intensity
of ¯uorescence, has been carried out (Figs. 1 and
Fig. 2. Contour plot of ATE ((A), solid line), PRO ((B), broken
line), AMI ((C), solid line) and DIP ((D), broken line). The route is
a cubic equation that passes through the points �ex1�228.8, �em1�300 nm (ATE), �ex2�228.8, �em2�340 nm (PRO), �ex3�366.4,
�em3�412.8 nm (AMI) and �ex4�288, �em4�487.2 nm (DIP). This
route corresponds to the following function: �ex�1984.89ÿ18.4371��em�5.99749�10ÿ2��em2ÿ6.01003�10ÿ5��em3.
J.A. Murillo PulgarõÂn et al. / Analytica Chimica Acta 370 (1998) 9±18 13
2). The route is optimized to minimize the spectral
interference caused by every compound in the mixture
with no loss of sensitivity.
From the contour plots it is evident that a route
which is obtained by means of the cubic equation that
passes through the points �ex1�228.8, �em1�300 nm
(ATE), �ex2�228.8, �em2�340 nm (PRO), �ex3�366.4, �em3�412.8 nm (AMI) and �ex4�288, �em4�487.2 nm (DIP), is the optimum to pass through the
excitation and emission maximum of the four com-
pounds. This route corresponds to the following func-
tion: �ex � 1984:89ÿ 18:4371� �em � 5:99749�10ÿ2 � �2
em ÿ 6:01003� 10ÿ5 � �3em. This function
is represented in Fig. 2.
3.4. Simultaneous determination of ATE, PRO, AMI
and DIP
Fig. 3 shows NLVASF spectra for ATE, PRO, AMI
and DIP. The determination by measuring the max-
imum intensity produces errors in the quanti®cation of
PRO by the presence of ATE and in DIP by the
presence of AMI. Owing to this overlap of the spectra,
the determination of these doping substances only by
NLVASF was not feasible. This overlap has been
resolved by taking the ®rst derivative of the spectrum.
The ®rst derivative of NLVASF spectra for ATE,
PRO, AMI and DIP are shown in Fig. 4. Although the
®rst derivative does not separate the bands, satisfac-
tory results could be obtained by applying the zero-
crossing technique. We selected 238.6/307.2 (A),
301.4/349.6 (B), 354.6/393.6 (C) and 312.4/
477.6 nm/nm (D) as optimal �ex/�em for the determi-
nation of ATE, PRO, AMI and DIP, respectively.
The number of points through which the derivative
is obtained was optimized, and it was concluded that
derived spectra with a suitable signal-to-noise ration
are obtained with 25 points.
3.5. Statistical analysis results
In order to test the mutual independence of the
analytical signal of ATE, PRO, AMI and DIP, i.e., to
show that zero-crossing wavelengths selected are
independent of the compounds present, three calibra-
tion graphs were constructed from ®rst derivative
signals for each one in the absence and in the presence
of different concentration of the rest of the com-
pounds. The concentration interval of the calibration
graphs and the concentration of the rest of the com-
pounds are given in the second and third column of
Tables 1 and 2. The concentration levels of the cali-
bration graphs were selected according to maximum
Fig. 3. NLVASF spectra of (A) ATE, (B) PRO, (C) AMI and (D)
DIP. (ATE)�400 ng mlÿ1; (PRO)�200 ng mlÿ1; (AMI)�280 ng mlÿ1; (DIP)�100 ng mlÿ1.
Fig. 4. First derivative of NLVASF spectra of (A) ATE, (B) PRO,
(C) AMI and (D) DIP. (ATE)�400 ng mlÿ1; (PRO)�200 ng mlÿ1;
(AMI)�280 ng mlÿ1; (DIP)�100 ng mlÿ1.
14 J.A. Murillo PulgarõÂn et al. / Analytica Chimica Acta 370 (1998) 9±18
sensitivity for the four compounds in the instrumental
conditions. First, the voltage of the photomultiplier
was optimized at the wavelength selected for the
determination for each compounds at same concen-
tration levels (100 ng mlÿ1). The minimum adjusted
voltage of the photomultiplier was selected, which
was in the case of DIP, then, the upper limit of the
calibration graph concentration for the other sub-
stances (ATE, PRO and AMI) was optimized accord-
ing to this voltage.
In order to test the presence of outlier the least
median of squares regression (LMS) [24] was applied.
As LMS is a robust regression method, it is able to
detect outlier points. These outliers cause errors in the
true line when experimental data are ®tted according
to the least squares regression. No outliers were
detected. Table 1 shows the outstanding results of
the statistical analysis.
The proposed method was evaluated by a statistical
analysis of experimental data by ®tting the least-
squares line according to y�a�bx [25,26]. Table 2
shows the outstanding results of the statistical analysis
of data.
To obtain the most representative calibration graph,
a classic overall least squares regression was devel-
oped, including all the data pairs. In order to test the
validity of the overall regression line an ANOVA test
comparing the three lines proposed for every analyte
was performed [26]. The F1 value compares the total
deviations from the overall linear region with the
deviation within each set from the set line testing
the suitability of the single overall regression line. If
the experimental value F1 is less than the theoretical
value F1, the departure of the individual sets from the
overall regression line is not signi®cant. The F2 value
compares the differences among the regression coef-
®cients, i.e., the slopes with the deviation within each
set from the set line testing the differences between the
regression coef®cients. If the experimental value F2 is
less than the theoretical value F2, then there are no
signi®cant differences between the individual slopes
and the overall regression slope may be taken as the
representative one. In all cases the validity of the
overall regression lines is proven by the values of
experimental F1 and F2 values which are less than the
theoretical ones at 95% con®dence level.
The residual values about the regression lines pro-
posed show a uniform variance (the errors of measure-
ments are independent with the concentration of the
drugs), therefore the regression lines meet the require-
ment of homocedasticity [25,26].
The signi®cance of the intercept on the y-axis was
established by testing the 95% joint con®dence inter-
vals for the parameters of the linear regression model
Table 1
Application of least median of squares regression (y�a�bx) to the data of the calibration sets obtained by means of first derivative of non-
linear variable-angle synchronous fluorescence
Compound
determined
Concentration
interval (ng mlÿ1)
Compound
present (ng mlÿ1)
Intercept (a) Slope (b) Determination
coefficient (r2)
SD of
estimation
ATE 10±400 Absence of the others 8.805�10ÿ4 6.041�10ÿ4 0.9999 4.1�10ÿ4
PRO 100, AMI 140, DIP 80 4.622�10ÿ3 5.628�10ÿ4 0.9999 1.9�10ÿ4
PRO 18, AMI 16.8, DIP 20 ÿ3.108�10ÿ4 5.808�10ÿ4 0.9999 1.8�10ÿ4
PRO 6±200 Absence of the others ÿ4.410�10ÿ4 1.450�10ÿ3 0.9999 9.5�10ÿ4
ATE 300, AMI 140, DIP 80 1.510�10ÿ3 1.418�10ÿ3 0.9999 6.5�10ÿ4
ATE 40, AMI 16.8, DIP 20 ÿ4.660�10ÿ4 1.440�10ÿ3 0.9999 7.5�10ÿ4
AMI 5±280 Absence of the others ÿ1.409�10ÿ4 1.108�10ÿ3 0.9999 2.9�10ÿ4
ATE 300, PRO 100, DIP 80 ÿ7.756�10ÿ6 1.074�10ÿ3 0.9999 3.6�10ÿ4
ATE 40, PRO 18, DIP 20 7.458�10ÿ4 1.028�10ÿ3 0.9999 1.5�10ÿ4
DIP 5±100 Absence of the others ÿ2.713�10ÿ4 3.952�10ÿ3 1.0000 1.3�10ÿ4
ATE 300, PRO 100, AMI 140 ÿ1.293�10ÿ3 4.061�10ÿ3 1.0000 5.1�10ÿ4
ATE 40, PRO 18, AMI 16.8 1.487�10ÿ4 4.107�10ÿ3 1.0000 3.0�10ÿ4
ATE (atenolol), PRO (propranolol), AMI (amiloride), DIP (dipyridamole).
J.A. Murillo PulgarõÂn et al. / Analytica Chimica Acta 370 (1998) 9±18 15
[26]. If the intercept on the y-axis `̀ a'' is plotted
against the slope `̀ b'' for repeated random samples
the points will fall elliptically around the true center
(�, �), and conversely, any con®dence interval for the
true combination of � and � will take the form of an
elliptical region around the best estimates (a,b) as
center [19]. If the null intercept on the y-axis falls
within the joint con®dence region, the intercept is not
signi®cantly different from zero. Therefore a propor-
tional relation between analytical signal and concen-
tration can be assumed. All intercepts found are not
signi®cantly different from zero.
By applying the IUPAC [27,28] de®nition, based
only on three times the standard deviation of the blank
a detection limit of 5.2, 1.9, 2.1 and 0.1 ng mlÿ1 was
obtained for ATE, PRO, AMI and DIP, respectively.
The propagation of errors will give a detection limit
consistent with the reliability of the blank measure-
ments (IUPAC) and besides the signal measurements
of the standards [29,30]. In this case a detection limit
of 5.9, 2.0, 2.8 and 1.0 ng mlÿ1, respectively, was
obtained for each. Moreover, Clayton [31] considers
the probability of positive false and negative false, the
detection limit being 6.7, 2.3, 3.2 and 1.2 ng mlÿ1 for
ATE, PRO, AMI and DIP, respectively. Because of the
limit of detection according to Clayton includes all the
possible errors described above, it is suitable to
assume this type of detection limit.
Table 2
Application of least squares regression (y�a�bx) to the data of the calibration sets obtained by means of first derivative of non-linear variable-
angle synchronous fluorescence (comparison of several regression lines by means of ANOVA test (F1 and F2))
Compound
determined
Concentration
interval
(ng mlÿ1)
Compound
present
(ng mlÿ1)
Intercept (a) Slope (b) Determination
coefficient (r2)
SD of
estimation
F1 F2
ATE 10±400 Absence of the others 2.849�10ÿ3 5.796�10ÿ4 0.9985 3.4�10ÿ3
PRO 100, AMI 140,
DIP 80
2.381�10ÿ3 5.729�10ÿ4 0.9994 2.1�10ÿ3 0.78 0.77
PRO 18, AMI 16.8,
DIP 20
4.585�10ÿ4 5.852�10ÿ4 0.9995 2.0�10ÿ3
Overall 1.896�10ÿ3 5.792�10ÿ4 0.9991 2.5�10ÿ3
PRO 6±200 Absence of the others 3.073�10ÿ4 1.446�10ÿ3 0.9997 1.9�10ÿ3
ATE 300, AMI 140,
DIP 80
2.105�10ÿ3 1.422�10ÿ3 0.9998 1.5�10ÿ3 2.18 1.92
ATE 40, AMI 16.8,
DIP 20
2.854�10ÿ4 1.416�10ÿ3 0.9994 2.5�10ÿ3
Overall 8.993�10ÿ4 1.428�10ÿ3 0.9995 2.2�10ÿ3
AMI 5±280 Absence of the others 1.569�10ÿ3 1.059�10ÿ3 0.9992 3.0�10ÿ3
ATE 300, PRO 100,
DIP 80
5.021�10ÿ5 1.074�10ÿ3 0.9999 6.4�10ÿ4 1.71 2.72
ATE 40, PRO 18,
DIP 20
1.311�10ÿ3 1.046�10ÿ3 0.9996 2.1�10ÿ3
Overall 9.766�10ÿ4 1.060�10ÿ3 0.9995 2.3�10ÿ3
DIP 5±100 Absence of the others ÿ9.366�10ÿ4 4.023�10ÿ3 0.9996 2.8�10-3
ATE 300, PRO 100,
AMI 140
ÿ2.029�10ÿ3 4.104�10ÿ3 0.9999 1.3�10ÿ3 2.85 2.47
ATE 40, PRO 18,
AMI 16.8
1.383�10ÿ3 4.040�10ÿ3 0.9996 2.8�10ÿ3
Overall ÿ5.276�10ÿ4 4.056�10ÿ3 0.9996 2.9�10ÿ3
ATE (atenolol), PRO (propranolol), AMI (amiloride), DIP (dipyridamole). Theoretical F1 (4, 15, 0.05)�3.06. Theoretical F2 (2, 15,
0.05)�3.68.
16 J.A. Murillo PulgarõÂn et al. / Analytica Chimica Acta 370 (1998) 9±18
Tab
le3
Res
ult
from
the
assa
ys
of
ph
arm
aceu
tica
lp
rep
arat
ions
by
firs
tder
ivat
ive
of
non-l
inea
rvar
iable
-angle
synch
ronous
fluore
scen
cete
chniq
ue
Ph
arm
aceu
tica
l
pre
par
atio
n
No
min
alco
nte
nt
(mg
)A
mount
added
(mg)
Am
ount
found
(mg)
Rec
over
y(%
)
AT
EP
RO
AM
ID
IPA
TE
PR
OA
MI
DIP
AT
EP
RO
AM
ID
IPA
TE
PR
OA
MI
DIP
Kal
ten
50
.0Ð
2.5
ÐÐ
13.3
Ð8.3
48.4
14.2
2.6
58.2
96.8
106.9
106.0
99.3
Nea
ten
olo
ld
iu1
00
.0Ð
ÐÐ
Ð50.0
43.8
31.2
99.3
53
45.4
31.4
99.3
106.1
103.6
100.6
Ate
no
lol
leo
10
1.3
ÐÐ
ÐÐ
50.0
43.8
31.2
105.5
53.6
45.8
32.5
104.1
107.2
104.5
104.3
No
rmo
pre
sil
10
0.0
ÐÐ
ÐÐ
50.0
43.8
31.2
110.0
54.0
45.8
31.6
110.0
108.1
104.5
101.4
Su
mia
l1
0Ð
10
.0Ð
Ð25.0
Ð11.7
8.3
26.9
10.9
12.1
8.6
107.5
109.1
103.4
103.2
Per
san
tin
ÐÐ
Ð1
01.5
312.5
166.7
145.8
Ð332.2
166.7
132.8
96.1
106.3
100.0
91.1
94.7
AT
E(a
ten
olo
l),
PR
O(p
rop
ran
olo
l),
AM
I(a
mil
ori
de)
DIP
(dip
yri
dam
ole
).
J.A. Murillo PulgarõÂn et al. / Analytica Chimica Acta 370 (1998) 9±18 17
In order to study the precision of the method, a
series of ten solutions was prepared containing 80, 24,
22.4 and 40 ng mlÿ1 of ATE, PRO, AMI and DIP,
respectively, and they were measured 15 times in a day
(repeatability study) and once a day for 10 days
(reproducibility study). The standard deviation, and
coef®cients of variation of repeatability at each of
these levels were 2.0, 0.35, 0.19, and 0.32 ng mlÿ1 and
2.5%, 1.5%, 0.85% and 0.80%, respectively. The
standard errors and coef®cients of variation of repro-
ducibility at each of these levels were 9.2, 0.99, 0.19,
and 1.43 ng mlÿ1 and 11.5%, 4.1%, 0.85% and 3.6%,
respectively (95% con®dence level).
In order to check the usefulness of the proposed
method, and because there are no pharmaceutical
dosage forms commercially available containing
ATE, PRO, AMI and DIP simultaneously, the pro-
posed method has been applied to their determination
in six different pharmaceutical preparations which
contains at least one of the four drugs. These prepara-
tions were conveniently spiked with different amounts
of the other compounds and the mixture analysis was
carried out. The assay results expressed as a percen-
tage of the nominal contents (R%) resulting from the
average of the determination of three different tablets
are summarized in Table 3. The recoveries (from
96.8% up to 110.0%) agree well enough with the
nominal content and the precision is quite satisfactory.
Acknowledgements
The authors gratefully acknowledge ®nancial sup-
port from the `̀ DireccioÂn General de InvestigacioÂn
Cientõ®ca y TeÂcnica'' (Project no. PB 94-0743).
References
[1] C. RodrõÂguez Bueno, Dopaje, 1st ed., Mc-Graw Hill, Madrid,
1992.
[2] A.C. Moffat, Clarke's Isolation and Identification of Drugs,
2nd ed., The Pharmaceutical Press, London, 1986, pp. 339,
362±363, 562±563 and 1037±1038.
[3] T. Vo-Dinh, Anal. Chem. 50 (1978) 396.
[4] P. John, I. Soutar, Anal. Chem. 48 (1976) 520.
[5] J.B.F. Lloyd, Nature Phys. Sci. 231 (1971) 64.
[6] J.B.F. Lloyd, Forensic Sci. Soc. 11 (1971) 83.
[7] J.N. Miller, Analyst 109 (1984) 191.
[8] B.J. Clark, A.F. Fell, K.T. Milne, D.M.G. Pattie, H. Williams,
Anal. Chim. Acta 170 (1985) 35.
[9] F. GarcõÂa SaÂnchez, A.L. Ramos Rubio, V. CerdaÂ, M.T. Oms,
Anal. Chim. Acta 228 (1990) 293.
[10] M.T. Oms, R. Forteza, V. CerdaÂ, S. Maspoch, J. Coello, M.
Blanco, Anal. Chim. Acta 233 (1990) 159.
[11] F. GarcõÂa SaÂnchez, A. FernaÂndez GutieÂrrez, C. Cruces Blanco,
Anal. Chim. Acta 306 (1995) 313.
[12] J.A. Murillo, A. AlanÄoÂn, Computer Chem. 17 (1993) 341.
[13] G.L. Green, T.C. O`Haver, Anal. Chem. 46 (1974) 2191.
[14] A. Petidier, S. Rubio, A. GoÂmez-HeÂns, M. ValcaÂrcel, Anal.
Biochem. 157 (1986) 212.
[15] J.A. Murillo PulgarõÂn, A. AlanÄoÂn Melina, Talanta 41 (1993)
21.
[16] A. MunÄoz de la PenÄa, F. Salinas, I. DuraÂn MeraÂs, Anal. Chem.
60 (1988) 2493.
[17] J.A. Murillo PulgarõÂn, A. AlanÄoÂn Molina, Anal. Chim. Acta
319 (1996) 361.
[18] F. Salinas, A. MunÄoz de la PenÄa, I. DuraÂn MeraÂs, M.S. DuraÂn,
Analyst 115 (1990) 1007.
[19] J.A. Murillo, A. AlanÄoÂn, Analyst 119 (1994) 1915.
[20] S. Rubio, A. GoÂmez-Hens, M. ValcaÂrcel, Talanta 46 (1974)
2194.
[21] M. Mariaud, P. Dubois, P. Levillain, Analyst 113 (1988)
929.
[22] A. Savitzky, M.J.E. Golay, Anal. Chem. 36 (1964) 1627.
[23] J. Steinier, Y. Termonia, J. Deltour, Anal. Chem. 44 (1972)
1906.
[24] P.J. Rousseeuw, A.M. Leroy, Robust Regression and Outlier
Detection, 1st ed., Wiley, New York, 1987.
[25] P.D. Lark, B.R. Craven, R.C.L. Bosworth, The Handling of
Chemical Data, 1st ed., Chapter 4, Pergamon Press, Exeter,
1968.
[26] D.L. Massart, B.G.M. Vandeginste, S.N. Deming, L. Kauf-
man, in: B.G.M. Vandeginste, L. Kaufman (Eds.), Chemo-
metrics: A Textbook, 1st ed., Oxford University Press,
Oxford, 1988.
[27] Nomenclature, symbols, units and their usage in spectro-
chemical analysis, II, Spectrochim. Acta B 33B (1978)
242.
[28] Guidelines for data acquisition and data quality evaluation in
environmental chemistry, Anal. Chem. 52 (1980) 2242.
[29] J.N. Miller, Analyst 116 (1991) 3.
[30] G.L. Long, J.D. Winefordner, Anal. Chem. 55 (1983) 712.
[31] C.A. Clayton, J.W. Hines, P.D. Elkins, Anal. Chem. 59 (1987)
2506.
18 J.A. Murillo PulgarõÂn et al. / Analytica Chimica Acta 370 (1998) 9±18