SciELO - Scientific Electronic Library Online

Home Pagelista alfabética de revistas  

Servicios Personalizados




Links relacionados


Journal of the Chilean Chemical Society

versión On-line ISSN 0717-9707

J. Chil. Chem. Soc. v.54 n.2 Concepción jun. 2009 

J. Chil. Chem. Soc, 54, N° 2 (2009)






a Department of Analytical and Inorganic Chemistry, Faculty of Chemical Sciences, University of Concepción, Concepción, Chile.

b Department of Chemistry, Faculty of Sciences, University of Chile, Santiago, Chile.

c Renewable Resources Laboratory, Biotechnology Centre and Faculty of Chemical Sciences, University of Concepción, Concepción, Chile.

*e-mail adress:


This paper analyses several mathematical approaches for simultaneous determination of dibucaine (D) and chlorphenamine maléate (CM). The different calibration methods utilised were: first derivative spectrophotometry (FDS), classical least squares (CLS), regression of partial least squares (PLS), and principal components regression (PCR). For the multivariated methods, the limits of detection (LOD) for multivariate calibrations were determined creating a surrogate signal variable (SSV).

The recoveries were selected to confirm the validity of the models. In general, the CLS model presents the poorest recovery. In order to compare the exactitude of the different calibration methods, a one-way ANOVA test was performed. In these analyses, only the CLS calibration method produced a significantly different prediction.

The lowest LODs were achieved by CLS and FDS, although the former method presented lower precision and poor prediction. Therefore, the FDS method is proposed as a pure binary calibration method for D-CM mixed samples, including the prepared according to a medical prescription and plasma samples simulating an event of intoxication.

Keywords: Spectrophotometry derivative, CLS, PCR, PLS, dibucaine, chlorphenamine maléate



The dibucaine (D) is a topical that acts as local anesthetic or analgesic in smaller burns or hemorrhoids is found in products ranging from teething gels to hemorrhoid creams to ear preparations. These compounds are available as both prescription and non-prescription preparations. Commonly used and readily available, these medications may be dispensed or sold without safety closures, making them more accessible to adult and children. Often, toxic exposures are due to inadvertent but overzealous administration to children by parents. The effects of such exposures can be shocking. In 2003, there were 8576 exposures to local/topical anesthetics reported to the American Association of Poison Control Centers, with 67% of cases in the age group younger than 6 years old1. This report reviews the available literature involving topical anesthetic exposures in children younger than 6 years old, additionally; there were seven deaths in this age range from topical anesthetics2.

On the other hand, chlorphenamine maléate (CM) has therapeutic and an-tihistaminic activities, is massively used and commonly prescribed in cases of allergic rhinitis or allergic conjunctivitis and this drug also is within reach of peoples.

Due to the frequent antihistaminics consumption for ambulatory patients and domestic accident with anesthetics, it increases the opportunity of intoxication with this types of drugs and consequently it is possible the presence of this drugs in the blood (3,4). In this context it is very important to develop method in order to protecting the human health that every day takes more relevance.

Several methods have been described to quantitatively determine CM and D separately or in combination with other drugs in pharmaceutical formulations. In the case of D, hydrophobic interaction electrokinetic chromatography (HIEKC)5, atomic absorption spectrometry and spectrophotometry (AAS) 6, electrochemical methods 7, gas chromatography (GC) 8, capillary isotacho-phoresis ', and liquid chromatography (LC) 10 have been described. The CM determinations are realized by chromatographic methods (HPLC, LC, GC)11-13, capillary electrophoresis 14,15, electrochemical methods 16,17, proton NMR spectroscopy18, atomic absorption spectrometry19, and spectrophotometric methods 20. The simultaneous determination of both drugs has been described by the thermochromism of ion associates 21.

The separate determination of D and CM has been performed in biological fluids (serum, plasma, blood, etc.) together with other drugs. In the cases that include D, GC 9,22, and LC 23 methods are used; in the case that contain CM, LC-TSM 24 and HPLC is also used25,26.

In general, derivative spectrophotometry has been directly used for the simultaneous determination of organic and inorganic compounds27-30. The present work studies this technique and describes solvent selection and spectral

variable optimisation in order to ensure precise procedures and accurate results in the application of the proposed method. For all the tested solutions, the first derivative spectra were recorded on the range of 400 to 190 nm against me-thanol.

Multivariate methods offer an alternative for data analyses that include many variables. These methods consider all the variables at the same time, presenting several advantages over univariate methods, especially with respect to interaction considerations and the number of experiment required to obtain a significant response 31. Multivariate calibration methods have been widely applied to the simultaneous quantification of analytes in mixtures. These methods are related with the establishment of an association between matricial algorithms and calibration data 31,32. Afterwards, an unknown data set can be predicted using the validated models.

Among multivariate calibration methods, principal component regression (PCR) and partial least-squares regression (PLS) have been successfully adopted in many quantitative assays of pharmaceutical formulations 32-34. The theoretical base of these methods has been fully described by several authors 35. CLS analysis is one of the simplest multivariate methods and is easy to perform, although its results are not very accurate in the quantification of mixtures when the analyse spectra have significant overlapping 36. On the other hand, PCR and PLS regression have been used more successfully in quantification of those types of samples, although its use presents more complications than CLS and FDS.

The aim of this work is to develop methods by FDS and three multivariate calibration methodologies: CLS, PLS and PCR. Also included to apply the methods in plasma spiked with CM and D, in order to simulate an intoxication condition, as well as the comparison of the develops methods.®



A Perkin-Elmer spectrophotometer Lambda 12®, with 10-mm quartz cells was used to measure the absorbance and derivative absorption spectra. For all tested solutions, the first derivative spectra were recorded on the range of 400 to 190 nm against methanol. The spectral data were processed by the software UV-WinLab 1.1 version.

Reagents and solutions

All reagents were of analytical grade. Stock solutions 1.Ox1O-3 mol L-1 D (Laboratory Chile) and CM (Sigma-Aldrich®) were prepared by dissolving 16.7 and 22.7 mg of each compound in a 50 niL volumetric flask in different solvents, respectively. Other concentration ranges were prepared by appropriate dilution with the respective solvent and in suitable containers to minimize solvent evaporation.

Spectrophotometry derivative

Calibration curves

D and CM aliquots stock solutions were simultaneously diluted in metha-nol to obtain a concentration between of l.OxlO-5 and 6.0xl0-5 mol L-1 for each drag. Calibration curves were determined for each compound in presence of 3.0xl0-5mol L-1 of the other. In all cases, the corresponding absolute values of the first-derivative spectra at 312 and 274 nm for D and CM respectively were obtained, and the values were plotted against the corresponding concentration.

Mixtures of D and CM in different concentration ratios D and CM aliquots stock solutions were simultaneously diluted in metha-nol to obtain D and CM standard solution mixtures in different concentration ratios (D:CM = 1:1, 2:3, 1:2, 1:3, 2:1, 3:2 y 3:1). The relation 3:2 (CM:D) corresponds to a medical prescription.

In all cases, the corresponding absolute values of the first derivative spectra at 312 and 374 nm for D and CM, respectively, were obtained and evaluated with linear regression equations (Table 1).

Preparation of plasma sample

A 500 uL aliquot of plasma sample previously was enriched with D 1.0 x 10-3 mol L-1 between 10 and 25 uL and CM of 1.0 x 10-3 mol L-1 between 25 and 30 uL. To each sample, 50 uL methanokwater (50:50, v/v) and 0.1 mL of 1.0 mol I71 NaOH were added. The sample was briefly mixed and 3 mL of ethyl acetate were added. The mixture was vortex-mixed for approximate 1 min, then shaken on a mechanical shaker for 15 min. After centrifugation at 3,000 rpm for 10 min, the upper organic layer was removed and evaporated to dryness at 40 °C under a gentle stream of nitrogen. The dry residues were reconstituted in 5 ml of methanol and then the solutions were evaluated by FDS and multivariate methods.

Multivariate calibration methods


CLS, PCR and PLS analysis were performed by the software Pirouette version 3.11 (Infometrix) and the Spectra transformations by Spectra version 1.60 (Perkin Elmer).

Data pre-treatment

Twelve independent standards were used for the calibration data set, ranging from 1.0 x 10-5 to 6.0 x 10-5 mol L-1 for each compound.

The recorded spectra data were mean-centered. The wavelength region between 220 nm and 400 nm was selected for multivariate calculation in order to minimise the influence of variables that do not contribute to model and could contribute to the error component36. The spectra resolution was 1 nm.

Outlier diagnose

The outlier diagnose for CLS was performed considering the Leverage, Sample Residual and Mahalanobis distance. The outlier diagnose for PCR and PLS was carried out through the cross validation performed, considering the Studentised residual together with Leverage with thresholds based on a 95%. The unusual samples were eliminated, and then the models were reconstructed, being optimised, validated and utilised for all the calculations 37.

Validation and Optimisation

All the multivariate models utilised cross validation, removing in turn each standard from the training set and then predicting the excluded sample with the calibration, similar to procedures utilised in the literature 38. The optimal baseline order for CLS was determined by minimising the model error sum of squares (ESS).


where e is the error matrix

The Prediction Residual Error Sum of (PRESS), Standard Error of Cross-Validation (SECV) as well as the correlation coefficient (rVal) for the predicted Y values versus known Y values, considering the cross validation 37. In the PCR and PLS calibration methods, these parameter were calculated according

where f the prediction residual matrix


Limit of detection

The limits of detection (LOD) for multivariate calibrations were determined creating a surrogate signal variable (SSV) 35,4°] in order to create a pseudo-univariated model. The selected SSV were the model predicted values. Thus, the LOD were calculated in a manner similar to FDS.

Recovery statistical analysis

Recovery data for all the calibration methodologies were compared using One-way ANOVA, with a threshold level of a = 0.05. These procedures were performed using Microsoft Excel v XP.


Spectrophotometry Derivative Method

Solvent effect on spectral behaviour of D and CM

The structure of both drags are different (Figure 1), consequently is possible to wait a difference spectral behaviour in relation to solvent. This effect was studied for D and CM separately. Methanol, ethanol, and acetonitrile (ACN) were found to be the best solvents because the analytes produced higher and defined spectral bands. Dimethylformamide (DMF) and n-Hexane were discarded because the signals presented increased noise (DMF) and due to solubility problems (n-Hexane). In the Figure 2, the classical spectra of D and CM on the 400 to 190 nm wavelength range are presented. When evaluated directly against solvent, D dissolved in methanol presented two maximum absorption broad peaks at 328 and 275 nm. Under similar conditions, the CM spectrum presented one broad absorption band between 255 and 270 nm.

When using the digital derivative spectrophotometry proposed by Savitzky and Golay41 in the classical spectra of the drags (Figure 2), it can be observed that D can be determined in presence of CM, although CM cannot be determined in the presence of D.



Selection of Derivative Order

Different derivative orders of the spectra were digitally obtained from the classical spectra. Fig. 3 shows that the first derivative could be used for the simultaneous determination of D and CM. The derivative can present characteristic zones for each compound, which can be used for analytical purposes. Taking into account that when the derivative order increases, the sensitivity decreases while spectra resolution improves. In this context, the first derivative was selected in the simultaneous determination of these compounds because the signal/noise ratio is higher.


Selection of Smoothing and Scale Factor on the derivative process

The objective of Smoothing Factor on (SF) value optimisation in analyte determination is to obtain higher sensibility, resolution and signal/noise ratio values. When increasing the SF, the calibration graphs slopes decrease (Figure 4 I (b) and II (b)) and noise also decrease. On the other hand, in the Figure 4 I (a) and II (a), it can be observed that there is no mutual interference between the drugs. Several SF values were tested, and SF = 5 was selected as the optimum for the both drugs. The first derivatives, using a value 5 as SF and; 1000 and 10000 as scale factor were recorded, the last value was selected. It was possible to use this parameter to the improve the lecture of analytical signal without effect distortions.

The selection of analytical wavelength

The first derivative spectrum of D was evaluated directly against the solvent, presenting two zero-crossings at 274 nm and 298 nm; CM does not absorb at either 300 nm or at 312 nm, while D presents a high analytic signal (Figure 3). In this context, the wavelengths 312 and 274 nm were selected for simultaneous determination of D and CM, respectively, because satisfactory regression coefficients were obtained in all cases. When using these wavelength values, better sensitivities with higher precision were obtained.


Different standard mixtures containing D and CM were stored cooled. After one month, these were analysed using the proposed method. In all the cases, the analytical signals were not altered, indicating there is no evidence of decomposition.

Analytical Features

The calibration graphs were established by measuring at appropriated wavelengths in the first derivative spectra. All analytical features are shown in Table 1. In all cases the corresponding absolute values of the first derivative spectra at 274 nm for CM and 312 nm for D were obtained and plotted against the corresponding concentrations.

Validation and optimisation of multivariate calibrations


The criteria used to adjust the optimal baseline for CLS was to reach the best ESS, PRESS and rVal values considering the cross validation (Table 2). Consequently, the optimal selected baseline was a constant. Although ESS and PRESS are not the minimal values, the rVal values are optimal for both D and CM, increasing model accuracy in the prediction of recovery samples.


(a) Effect of CM concentration on the signal of D at 3.0xl0-5 mol L-1, obtained at 312 nm, by FSD at different SF values

(b)  Calibrations graphs prepared from first derivative spectra of CM in the presence of D at 3.0xl0-5 mol L-1 at different SF values, at 274 nm.


(a) Effect of D concentration on the signal of CM at 3.0xl0-5 mol L-1, obtained at 274 nm, by FSD at different SF values.

(b)  Calibrations graphs prepared from first derivative spectra of D in the presence of CM at 3.0xl0-5 mol L-1 at different SF values, at 312 nm.




For PCR and PLS, the optimal number of PCs were estimated considering validation based criteria. In order to not over fit the models, two PCs for PCR and PLS calibration were selected according to data in Table 3. Even though, the PRESS and SECV values are non minimal values at this PC number, these have acceptable values close to minimal values while the rVal is close enough to maximal value.






Recoveries and Comparison between methods

The model-predicted concentrations were very close to the added concentrations, confirming model validity (Fig. 5). In general, the CLS model had the poorest recovery.




In order to compare the exactitude of the different calibration methods, a contrasts method was performed. Since the standard deviation (SD) of the CLS recoveries was different from the other calibration methodologies, a One-way ANOVA test was performed. This test indicated that the CLS quantification of both D and MC was significantly different (P = 0.002) from the other calibration methods assayed. The PCR, PLS and FDS methods were compared using a oneway ANOVA test since its SD are similar. This test does not indicate any significant difference (P = 0.9516) between these calibration methods (Table 4).

The LODs for the multivariate methods are presented in the Table 4. The highest LOD was achieved by PCR calibration, followed by PLS method. This difference lies in the slightly higher sensitivity of PLS in comparison with PCR in the case of an extensive spectral overlapping 38. The best LODs were achieved by CLS; although it has lower precision (Fig. 5.) and poor prediction in mixtures where the spectra are extensively overlapping 33. Considering this analysis, we conclude that the lowest LOD was achieved by FDS (Tables 1 and 5).


The propose methods were applied in the determination of D and CM in plasma by simulating an event of intoxication, with satisfactory results. These are described in Table 6.


For the simultaneous determination of binary drug mixtures whose UV-Vis spectra overlap in, FDS and recently multivariate calibration methods have been used. The recovery of D and CM from the synthetic samples including the prepared according to a prescription, were close to 100%. FDS and CLS methods has been applied successfully for the quantitative analysis of D and CM in plasma by simulating an event of intoxication.

The present study found no significant differences between the recoveries of calibration methods using PCs (PCR and PLS) and FDS.

The LOD and LOQ for PLS and PCR were slightly higher than for FDS, because FDS is zero order (from a chemometric view) and does not incorporate noise into the measurement, while the multivariate calibration methods consider the spectral ranges that do not have only analytic information since the noise has not been totally eliminated by the introduction of PCs.

FDS is the preferred method when the matrix effect is constant and the analyte mixture spectrum can be resolved. Additionally, FDS is an easy application that does not require highly trained technicians, increasing its feasible use in control laboratories. Still, first-order calibrations should be used when the FDS method does not achieve reliable results. In this case, its use is conditioned by specialised technicians since it is based on complex algorithms.


The authors are grateful to the Dirección de Investigación, Universidad de Concepción, projects DIUC N° 204.021.021-1.0 and 206.021.024-1.0. and FONDECYT project N° 1070905, for the financial support.



(1) AAPCC TESS Report. American Association of Poison Control Centers, 2005        [ Links ]

(2) W. A. Watson; T. L. Litovitz; W. Klein-Schwartz; G. C. Rodgers; J. Youniss; N. Reid; W. G. Rouse; R. S. Rembert; D. Borys, Am J Emerg Med, 22, 335, (2004).        [ Links ]

(3) T. Saito; S. Morita; I. Kishiyama; S. Miyazaki; A. Nakamoto; M. Nishida; A. Namera; M. Nagao; S. Inokuchi, J ChromatogrB, 872, 186, (2008).        [ Links ]

(4) A. Sen; A. Akin; K. J. Craft; D. V. Canfield; A. K. Chaturvedi; Aerospace Medical Assoc: 2007, p 514.        [ Links ]

(5) H. Okamoto; A. Uetake; R. Tamaya; T. Nakajima; K. Sagara; Y. Ito, J ChromatogrA, 929, 133, (2001).        [ Links ]

(6) N. T. Abdel-Ghani; A. F. A. Youssef; M. A. Awady, IlFarmaco, 60, 419, (2005).        [ Links ]

(7) M. S. M. Fernandez-Marcóte; M. C. Mochon; J. C. J. Sanchez; A. G. Perez, Analyst, 121, 681, (1996).        [ Links ]

(8) M. Culea; N. Palibroda; Z. Moldovan; A. D. Abraham; P. T. Frangopol, Chromatographia, 28, 24, (1989).        [ Links ]

(9) K. Igarashi; F. Kasuya; E. Mori; M. Fukui, J Chromatogr B Biomed Sci Appl, 415, 407, (1987).        [ Links ]

(10)  S. Izumoto; Y. Machida; H. Nishi; K. Nakamura; H. Nakai; T. Sato, J Pharm Biomed Anal, 15, 1457, (1997).        [ Links ]

(11)  S. F. Sun; G. R. Liu; Y. H. Wang, Chromatographia, 64, 719, (2006).        [ Links ]

(12)  I. M. Palabiyik; F. Onur, Chromatographia, 66, S93, (2007).        [ Links ]

(13)  U. R. Cieri, JAOACInt, 89, 53, (2006).        [ Links ]

(14)  X. Liu; L. Liu; H. Chen; X. Chen, J Pharm Biomed Anal, 43, 1700, (2007).        [ Links ]

(15)  M. Yururi; H. Nakanishi; K. Taniguchi, Bunseki Kagaku, 43, 575, (1994).        [ Links ]

(16)  H. J. Song; Z. J. Zhang; F. Wang, Electroanalysis, 18, 1838, (2006).        [ Links ]

(17)  M. El-Tohamy; M. El-Maamly; A. Shalaby, Asian J Chem, 19, 3035, (2007).        [ Links ]

(18)  G. M. Hanna; C. A. Lau-Cam, J Pharm Biomed Anal, 11, 855, (1993).        [ Links ]

(19)  Y. M. Issa; A. S. Amin, Analusis, 24, 139, (1996).        [ Links ]

(20)  N. B. Pappano; Y. C. De Micalizzi; N. B. Debattista; F. H. Ferretti, Talan-ta, 44, 633, (1997).        [ Links ]

(21)  T. Sakai, Analyst, 107, 640, (1982).        [ Links ]

(22)  H. Hattori; S. Yamamoto; T. Yamada; O. Suzuki, Journal of Chromato-graphy: Biomedical Applications, 564, 278, (1991).        [ Links ]

(23)  S.-I. Kawano; H. Murakita; E. Yamamoto; N. Asakawa, J Chromatogr B, 792, 49, (2003).        [ Links ]

(24)  X. Y. Chen; Y. Zhang; D. F. Zhong, Biomed Chromatogr, 18, 248, (2004).        [ Links ]

(25)  J. A. Holeman; N. D. Danielson, J Chromatogr A, 679, 277, (1994).        [ Links ]

(26)  M. Yamaguchi; H. Monji; K. Yamashita; I. Aoki; T. Yashiki, J Chroma-togr B Biomed Sci Appl, 661, 168, (1994).        [ Links ]

(27)  M. I. Toral; P. Richter; N. Lara; M. T. Escudero; C. Soto, Anal Lett, 33, 93, (2000).        [ Links ]

(28)  M. I. Toral; P. Richter; N. Lara; P. Jaque; C. Soto; M. Saavedra, Int J Pharm, 189, 67, (1999).        [ Links ]

(29)  M. I. Toral; A. Tassara; C. Soto; P. Richter, J AOAC Int, 86, 241, (2003).        [ Links ]

(30)  M. I. Toral; J. Rivas; M. Saldias; C. Soto; S. Orellana, J Chil Chem Soc, 53, 1543,(2008).        [ Links ]

(31)  D. Contreras; J. Freer; J. Rodriguez, Int Biodeterior Biodegrad, 57, 63, (2006).        [ Links ]

(32)  J. H. Kalivas, Anal Lett, 38, 2259, (2005).        [ Links ]

(33)  H. C. Goicoechea; A. C. Olivieri, Talanta, 47, 103, (1998).        [ Links ]

(34)  J. M. Lemus Gallego; J. Perez Arroyo, Anal ChimActa, 460, 85, (2002).        [ Links ]

(35)  A. C. Olivieri; N. K. M. Faber; J. Ferre; R. Boque; J. H. Kalivas; H. Mark, Pure Appl Chem, 78, 633, (2006).        [ Links ]

(36)  S. J. Haswell; A. D. Walmsley, Anal Chim Acta, 400, 399, (1999).        [ Links ]

(37)  I. T. i. A. C. Inc; 3.11 ed. Bothell, WA, 2003.        [ Links ]

(38)  G. Ragno; G. Ioele; A. Risoli, Anal Chim Acta, 512, 173, (2004).        [ Links ]

(39)  A. Afkhami; N. Sarlak; A. R. Zarei, Talanta, 71, 893, (2007).        [ Links ]

(40)  J. Saurina; C. Leal; R. Compano; M. Granados; M. D. Prat; R. Tauler, Anal ChimActa, 432, 241, (2001).        [ Links ]

(41)  A. Savitzky; M. J. E. Golay, Anal Chem, 36, 1627, (1964).        [ Links ]

(Received: June 5, 2008 - Accepted: March 19, 2009)

Creative Commons License Todo el contenido de esta revista, excepto dónde está identificado, está bajo una Licencia Creative Commons