Simultaneous determination of rifampicinand isoniazid in urineand pharmaceutical formulations by multivariate visible spectrophotometry

jbchs

ARTICLE

Simultaneous determination of rifampicinand isoniazid in urineand pharmaceutical formulations by multivariate visible spectrophotometry

Sandra Stets^I; Talita M. Tavares^I; Patricio G. Peralta-Zamora^I; Christiana A. Pessoa^II; Noemi Nagata^{I, *}

^IDepartamento de Química, Universidade Federal do Paraná, 81531-990 Curitiba-PR, Brazil

^IIDepartamento de Química, Universidade Estadual de Ponta Grossa, 84030-900 Ponta Grossa-PR, Brazil

ABSTRACT

Partial least squares regression (PLSR) was used for the simultaneous quantification of rifampicin (RIF)and isoniazid (INH) by visible spectrophotometry using a simple derivatization reaction. In the presence of neocuproine, copper(II) is easily reduced by INH to a Cu(I)-neocuproine complex that shows an absorption maximum at 455 nm. Under these conditions, RIF shows an absorption maximum at 449 nm. The calibration set was established between 8and 57 mg L^-1 for RIFand 1.5and 7 mg L^-1 for INH. The method was applied to the determination of the drugs in urine samples (recoveries between 92and 119%)and in pharmaceutical formulations (relative error lower than 5%).

Keywords: antibiotics, PLSR, urine, derivatization reaction

RESUMO

O método de regressão por mínimos quadrados parciais (PLSR) foi utilizado para a quantificação simultânea de dois tuberculostáticos rifampicina (RIF) e isoniazida (INH) por espectrofotometria na região do visível utilizando uma reação de derivatização simples. Na presença de neocuproína, íons cobre(II) foram facilmente reduzidos por INH para um complexo Cu(I)-neocuproína de absorção máxima a 455 nm. Sob essas mesmas condições, RIF mostrou uma absorção máxima a 449 nm. O conjunto de calibração foi estabelecido entre 8 e 57 mg L^-1 de RIF e 1,5 e 7 mg L^-1 de INH. O método foi aplicado para a determinação das drogas em amostras de urina (recuperações entre 92 e 119%) e em formulações farmacêuticas (erro relativo inferior a 5%).

Introduction

The World Health Organization (WHO) classifies tuberculosis (TB) as a neglected disease that affects thousands of people but does not present an attractive opportunity for economic investmentand development of pharmaceuticals, particularly for reaching people in developing countries.¹ According to the WHO, Brazil, the Russian Federation, India, Chinaand South Africa report 48% of the world TB cases.²

One of the most effective antituberculosis treatments used in many countries is based on a fixed dose combination (FDC) of two or more tuberculostatic agents in a single pharmaceutical formulation. In general, the use of FDC increases the treatment continuityand reduces the risk of resistance or relapses, treatment costsand errors in drug administrationand distribution.^3,4 The combination of drugs has therapeutic advantages; however, the combination of drugs brings new challenges to the pharmaceutical industry with respect to stability studies of combined drugsand their simultaneous analysis.⁵

The analysis of antituberculosis drugs (e.g., rifampicin, isoniazid, pyrazinamidand daptomycin) has been performed for pharmaceutical formulations and/or biological fluids. The analytical methods include chromatographic techniques such as high-performance thin-layer chromatography (HPTLC),⁶ high-performance liquid chromatography (HPLC),^7-12 ultra-performance liquid chromatography (UPLC),¹³ micellar electrokinetic capillary chromatography (MEKC),¹⁴ and, less frequently, spectrophotometric analysis combined with multivariate regression,^15-17 derivative spectrophotometry¹⁸and voltammetric methods.^19-21

In the present study, a multivariate visible spectrophotometric determination is reported for the simultaneous analysis of rifampicin (RIF) and isoniazid (INH) in pharmaceutical formulationsand urine samples. Because isoniazid does not show an absorption band in the visible region, the procedure is based on a simple chemical derivatization involving copper(II), isoniazidand neocuproine (NC). In the presence of neocuproine, copper(II) is reduced by isoniazid to a Cu(I)-neocuproine complex, which shows an absorption maximum at 455 nm.²² The proposed procedure allows the determination of both analytes in the visible region using partial least squares regression (PLSR), multivariate calibration tool that allows the simultaneous determination of chemical species, even in the presence of strong spectral overlap.

Experimental

Instruments

The visible absorption spectra were recorded on a Shimadzu UV-2401PC spectrophotometer (Kyoto, Japan) using a glass cuvette with a path length of 1 cm.

Analysis by high-performance liquid chromatography was performed according to United States Pharmacopeia (USP) recommendations²³ using a Varian 920-LC chromatograph (Mulgrave, Australia) equipped with an autosampler, a quaternary gradient pumpand a diode arraydetector (DAD, 238 nm). Routine chromatographic separations were performed by gradient elution on a C₁₈ column (Microsorb, 25mm×4.6mm × 5 µm), using injection volume of 20 µL.

The mobile phase consisted of acetonitrile (A)and sodium acetate buffer solution at pH 6.8 (B), using a gradient elution of 96-55% A in 0-6 minand 55% B in 10-15 min. Finally, the gradient was reverted to original conditions within next 5 min.

Chemicalsand standard solutions

Rifampicinand isoniazid were kindly supplied by the Farmanguinhos Laboratory (Fiocruz of Brazil, Rio de Janeiro, RJ, Brazil); the purity of RIF was 99.7%and that of INH was 100.8%. Chromatographic determinations involve the use of HPLC-grade solvents (JT Baker or similar)and ultra-pure water (Milli-Q, Millipore, Bedford, MA, USA). All other chemicals were of analytical grade.

Isoniazid stock solutions (137 mg L^-1) were prepared daily by dissolving 13.7 mg of isoniazid in 100 mL of deionized water. Rifampicin stock solutions (411 mg L^-1) were prepared daily by dissolving 20.5 mg of rifampicin in 10 mL of methanoland diluting to 50 mL with deionized water. A Cu(II) solution (242 mg L^-1) was prepared by dissolving 121 mg of Cu(NO₃)₂·3H₂O in 50 mL of deionized water. The neocuproine solution (218 mg L^-1) was prepared by dissolving 109 mg of neocuproine in 10mL of methanoland diluting to 50 mL with deionized water. The acetate buffer solution (pH 5) was prepared by dissolving 4.115 g of sodium acetate in 1.6 mL of acetic acidand diluting to 500 mL with deionized water. Further dilutions were made with this acetate buffer solution.

Factorial designand response surface

The colorimetric reaction for the indirect determination of isoniazid was optimized using a factorial design. A two-level factorial design was selected,and the quantitative factors evaluated were pH (4and 6), concentration of the Cu(II) solution (12and 24.2 mg L^-1)and concentration of the neocuproine solution (43.4and 217.3 mg L^-1). The two significant factors to increase the sensitivity of the spectrophotometric method were further studied by a central composite design (Table 1), in which fixed concentrations of INH (5.3 mg L^-1)and neocuproine (54.3mg L^-1) were maintained. In both designs, the response that was monitored was the absorbance signal at 455 nm.

Analytical procedure

Forty six synthetic mixtures were prepared by mixing known amounts of RIFand INH standard solutions, 1.0mL of NCand 1.0 mL of Cu(II) stock solutions in a 10 mL volumetric flaskand diluting with acetate buffer solution (pH 5). The final concentration of these solutions ranged between 8and 60 mg L^-1 of RIFand between 1.5and 7mgL^-1 of INH.

The absorption spectra were recorded between 350and 800 nm using a spectral resolution of 1 nm. The precision (reported as the relative standard deviation, RSD (%)), linearity (evaluated by regression analysis)and accuracy of the method for the determination of the drugs in pharmaceutical formulations were validated by considering the results obtained by the application of the chromatographic standard procedure.²³

Multivariate calibration

PLSR models were developed from twenty five synthetic mixtures containing 8 to 57 mg L^-1 of RIFand 1.5to 7mgL^-1 of INH (Figure 1). In addition, seven synthetic mixtures were prepared in triplicateand reserved as an external validation set. Two of these samples (26and 27) show similar concentrations to that shown by the analyzed drugs, while the remaining samples were randomly selected.

The mixtures were treated according to the previously described general procedure. The absorbance data were processed using PLS-Toolbox 3.0 (Eigenvector Research, Inc., Wenatchee, USA) software operated in MATLAB version6.5 (Mathworks, Natick, USA) software.

Analysis of the pharmaceutical formulations

Four pharmaceutical formulations (capsulesand tablets) were kindly supplied by the Health Secretary of the Paraná State (Brazil); the samples contained 300mgper200mgand 150 mg per 100 mg of RIFand INH, respectively, per capsule or tablet. These pharmaceutical formulations contain a large number of excipients, including aerosol, explocel, talcand magnesium stearate (tablet form), magnesium stearate, sodium starch glycolateand microcrystalline cellulose (capsules form).

Ten tablets were individually weighed to obtain their representative average weightsand were then finely powderedand mixed. In the case of capsules, the contents of ten capsules were completely removed from their shells.

Each of the pharmaceutical formulations was accurately weighed (40 mg)and transferred to a 100 mL volumetric flask. Approximately 10 mL of methanol were added to dissolve the drugs,and deionized water was used for dilution. An aliquot of 440 µL of this solution was prepared according to the previously described procedure in order to obtain a final concentration within of the calibration concentration range.

General procedure for the analysis of urine samples

Urine samples were obtained from eight volunteers (maleand female healthy donors) of two different age ranges. An aliquot of 2.0 mL of urine in a 10 mL volumetric flask was spiked to achieve a final concentration of approximately 20.5 mg L^-1 for RIFand 1.5 mg L^-1 for INH. Afterwards, the samples were submitted to the previously described general procedure. This concentration range was selected on the basis of previous literature, which suggests typical concentrations between 0.2-3.0 mg L^-1 for INH²⁴and 0.3-100 mg L^-1 for RIF²⁵ in urine samples of patient with active pulmonary tuberculosis.

Resultsand Discussion

Derivatization reaction of isoniazid

The proposed spectrophotometric method is based on the reducing capacity of INH toward the Cu(II)-neocuproine complex, with formation of a colored Cu(I)-neocuproine complex that absorbs at 455 nm.^22,25-27

A full factorial design (2³) was initially performed to study the influence of relevant variables (i.e., pH, concentration of the Cu(II) solutionand concentration of neocuproine solution) on the colorimetric reaction (results not shown). The most significant effect that improved the reaction sensitivity (evaluated by the evolution of the spectral signal at 455 nm) was caused by pH (+0.1589). The effect of the concentration of the Cu(II) solution (+0.1187)and its interaction factor with the pH (+0.0715) was also significant. These results show that the concentration of neocuproine (studied in the concentration range) does not interfere with the reaction sensitivity, most likely because the reactant is present in excess.

A central composite design with two levelsand two factors (pHand concentration of the Cu(II) solution, Table 2) was used to optimizeand model the reaction. A quadratic model was determinedand evaluated by analysis of variance (ANOVA) (Table 3).

The quadratic model showed good agreement between the percentage of explained variance (99.10%)and the maximum percentage of explainable variance (99.80%). The value of the mean square ratio MS_reg/MS_res was statistically significant (F-value >>> F-crit_95%). Moreover, the value of the mean square ratio MS_lof/MS_pe was not statistically significant (F-value <<<< F-crit_95%), which indicated no evidence of lack-of-fit for the quadratic model.

Figure 2 shows the response surface of the quadratic model that describes the reaction sensitivity as a function of the coded factors. The experimental conditions for maximum sensitivity are a pH of 6.0and Cu(II) concentration of 26.6 mg L^-1. However, in view of the high sensitivity also observed at the pH of the central point (pH5.0), this condition was selected for further assays, mainly to avoid the hydrolysis of Cu(II).

The influence of pH on the reduction of Cu(II) by INH has been described previously.^22,28 At pH 5.0, INH has only a single positive charge (pK_a values: 1.8 for the nitrogen of pyridine, 3.5 for the hydrazine group -NHand 10.8 for the hydrazine group -NH₂),²⁹ which is a favorable condition for the oxidation of the hydrazine group³⁰ by formation of an acyl radical, resulting in the formation of the corresponding carboxy lic acid.^28,31

The spectral profiles shown in Figure 3 confirm that INHand the Cu(II)-neocuproine complex do not absorb in the monitored spectral region, whereas RIF shows an intense signal centered at 470 nm (Figure 3). Under the selected experimental conditions, the characteristic band of RIF is changed significantly, which produces an absorption profile that is compatible with RIF quinine.³² In the presence of INH, the characteristic signal of the Cu(I)-neocuproine complex becames visible as a broad band centered at 450nm (Figure 3).

After derivatization (Figure 3, curves d and e), the spectra of both analytes are similar,and a strong spectral overlap can be observed (Figure 3, curve f). For this reason, PLSR model was used for the simultaneous determination of INH and RIF.

PLSR models: calibration and validation

The multivariate models were elaborated from 25 synthetic mixtures containing RIF and INH (Figure 1) submitted to the derivatization system. Several models were developed by PLSR using different pre-processing systemsand several latent variables (LVs).

Seven synthetic mixtures in triplicate were used as an external validation set.

The performance of the regression models was evaluated by analysis of the calibration model root-mean square error of calibration (RMSEC)and of validation (RMSEP) as well by the observed correlation (R) between the predictedand experimental values. In our case, smoothed spectral data and 3 factors were found to be optimum for the PLS-1 method. The loading data (Figure 4) indicate that these three latent variables enclose relevant analytical information without adding noise to the model. LV1 explains much of the spectral information of the derivatization product of RIF (Figure 3, curve c),and LV2 is responsible for capturing much of the spectral information from the derivatized product of INH (Figure 3, curve b). The lowest prediction errorand RMSEP (Table 4) were obtained using three latent variablesand smoothed spectral data. For this model, all prediction errors for the external validation set were lower than 5% (most prediction errors were lower than 2%).

Considering the limiting values of ± 2.5 for studentized residues (95% confidence) and 3(LV)/n for leverage (0.36), anomalies were not observed in the calibration set (Figure 5). The high leverage value of sample 25 implies an important influence on the developed model, not because it represents an anomaly, but due to the higher relative concentration of both study drugs (see Figure 1).

A one-way ANOVA test was conducted to compare the estimated concentration of INH and RIF in aqueous solutionsand the reference concentrations on both calibrationand validation sets (Table 5). In this procedure, Snedecor's F-values were computedand compared with the tabulated F-value (p = 0.05). The same computation process was repeated for both drugs. The value of the mean square ratio MS_reg/MS_res was much greater than the critical F-value, which implies a statistically significance of the regression at a 95% confidence level. Likewise, the value of the mean square ratio MS_lof/MS_pe proved to be below the critical F-value, revealing no evidence of lack-of-fit for the model. Thus, the numerical values of all statistic parameters indicated that our methods are suitable for the simultaneous determination of both drugs in aqueous solutions.

Limits of detection (LOD) of 0.06and 0.04 mg L^-1and limits of quantification (LOQ) of 0.19and 0.13 mg L^-1 were established for RIFand INH, respectively, according to procedures described by Valderrama et al..³³

Analysis of real samples

Different oral pharmaceutical formulations were analyzed using the proposed method,and the results are shown in Table 6, together with the results obtained using the official HPLC method. The statistical significance of the difference between the methods for the determination of RIFand INH was obtained using paired t-tests (ν_RIF=4and ν_INH = 4). No significant differences were observed at the 95% confidence level (t_RIF = 1.37and t_INH = 2.76, both of which are less than the critical value of 3.182). These results indicated a good agreement between the proposed multivariateand the official chromatographic method. The simplicity of the multivariate spectrophotometric method allows a high analytical throughput, allowing approximately one assay per min.

Determination of rifampicinand isoniazid in urine

Because of the good performance of the proposed method for simultaneous determination of RIFand INH in pharmaceutical preparations, the method was evaluated for the analysis of a more complex matrix (urine). Figure 6 shows that the spectral profile of spiked urine samples was very similar to that observed for the synthetic RIF/INH mixtures, which demonstrates the practical absence of spectral interferences caused by urine matrix. Moreover, the slight difference observed between samples from people under medical treatment with other drugsand samples of people who do not make use of any drug is an argument that suggests robustness of the proposed method.

The results obtained in the analysis of eight spiked urine samples are shown in Table 7. Taking into account the excellent observed recoveries, the efficiency of the proposed method for the analysis of complex matrices was demonstrated.

Conclusions

The proposed method avoids matrix interferences with a simple derivatization reaction, while spectral interferences can be overcome by using PLSR. The derivatization reaction was optimized by a response surface to provide a best sensitivity for the determination of INH (minor component in pharmaceutical formulations),and the best conditions were pH 5.0and higher concentrations of Cu(II) solutions.

PLSR models of high predictive capability were obtained using smoothed spectral dataand 3 LVs. Under these conditions, relative mean errors of approximately 1% were observed in the external validation process.

In the analysis of commercial drugs (capsulesand tablets), prediction errors lower than 3% for RIFand 5% for INH were observed for results obtained by applying the standard chromatographic method. This method is simple, inexpensiveand fast (less than 50 s per assay). The method can be applied to the analysis of urine samples without pretreatment (only a dilution is necessary). Interference of the urine matrix is not observed in the simultaneous determination of RIF and INH with adequate recoveries.

Finally, it is important to emphasize that, in view of the fact that the determination is based on spectral signals located in the visible region, interferences from the drug excipientsand from the several other components of the urine matrix do not significantly interfere.

Acknowledgments

The authors are grateful to CAPES (Coordenação de Aperfeiçoamento de Pessoal de Nível Superior) for the scholarship grantedand to CNPq (Conselho Nacional de Desenvolvimento Científico e Tecnológico) for financial support.

Submitted: January 29, 2013

Published online: June 25, 2013

• 1. Morel, M. C.; Cadernos de Saúde Pública 2006,22,1522.
• 2. World Health Organization (WHO), Tuberculosis (TB), 2013, http://www.who.int/tb/en/, accessed in April 2013.
• 3. World Health Organization (WHO), Global Tuberculosis Report 2012, http://apps.who.int/iris/bitstream/10665/75938/1/9789241564502_eng.pdf accessed in April 2013.
• 4. Agrawal, S. K.; Kaur, J.; Singh, I.; Bhade, S. R.; Kaul, C. L.; Panchangula, R.; Int. J. Pharm. 2002,233,169.
• 5. Bhutani, H.; Singh, S.; Jindal, K. C.; Chakraborti, A. K.; J.Pharm. Biomed. Anal. 2005,39,892.
• 6. Argekar, A. P.; Kunjir, S. S.; Purandare K. S.; J. Pharm. Biomed. Anal. 1996,14,1645.
• 7. Gaitonde, C. D.; Pathak, P. V.; Drug Dev. Ind. Pharm. 1991,17,1201.
• 8. Panchagnula, R.; Sood, A.; Sharda, N.; Kaur, K.; Kaul, C. L.; J. Pharm. Biomed. Anal. 1999,18,1013.
• 9. Calleri, E.; Lorenzi, E.; Furlanetto, S.; Massolini, B. G., Caccialanza, G.; J. Pharm. Biomed. Anal. 2002,29,1089.
• 10. Espinosa-Mansilla, A.; Acedo-Valenzuela, M. I.; Peña, A. M.; Cañada, F. C.; Madan, J.; Dwivedi, A. K.; Singh, S.; Talanta 2002,58,273.
• 11. Khuhawar, M. Y.; Rind, F. M. A.; J. Chromatogr., B: Analyt. Technol. Biomed. Life Sci. 2002,766,357.
• 12. Fang, P.; Deli, H.; Huazhu, R.; Youtan, Q.; Xu, W. P.; Pingliu, Y.; Yuanzhang, W.; Changchen, Y.; Zhang, F.; J. Chromatogr., B: Analyt. Technol. Biomed. Life Sci. 2010,878,2286.
• 13. Gikas, E.; Bazoti, F. N.; Nourgiakis, P.; Perivolioti, E.; Roussidis, A.; Skoutelis, A.; Tsarbopoulos, A.; J. Pharm. Biomed. Anal. 2010,51,901.
• 14. Acedo-Valenzuela, M. I.; Espinosa-Mansilla, A.; La-Pena, A.M.; Canada-Canada, F.; Anal. Bioanal. Chem. 2002,374,432.
• 15. Mahalanabis, K. K.; Basu, D.; Royt, B.; Analyst 1989,114,1311.
• 16. Espinosa-Mansilla, A.; Valenzuela, M. I. A.; La-Pena, A. M.; Salinas, F.; Canada, F. C.; Anal. Chim. Acta. 2001,427,129.
• 17. Li, B.; He, Y.; Lv, J.; Zhang, Z.; Anal. Bioanal. Chem. 2005,383,817.
• 18. Benetton, S. A.; Kedor-Hackmann, E. R. M.; Santoro, M.I.R.M.; Borges, V; Talanta 1998,47,639.
• 19. Lomillo, M. A. A.; Renedo, O. D.; Martinez, M. J. A.; Anal. Chim. Acta 2001,449,167.
• 20. Hammam, E.; Beltagi, A. M.; Ghoneim, M. M.; Microchem. J. 2004,77,53.
• 21. Leandro, K. C.; Carvalho, J. M.; Giovanelli, L. F.; Moreira, J.C.; Braz. J. Pharm. Sci. 2009,45,332.
• 22. Safavi, A.; Karimi, M. A.; Nezhad, M. R. H.; Kamali, R.; Saghir, N.; Spectrochim. Acta, Part A 2004,60,765.
• 23. US Pharmacopeial Convention, US Pharmacopeia National Formulary, 30^th ed., Rockville, MD, USA, 2007.
• 24. Venkataramann, P.; Menon K.; Nair, G. K.; Radhakrishnac, S.; Andrar, H.; Tripathy, S. P; Tubercle 1972,53,84.
• 25. Mitchison, D. A.; Allen, W.; Miller, A. B.; Tubercle 1970,51,300.
• 26. Guçlu, K.; Sozgen, K.; Tutem, E.; Ozyurek, M.; Apak, R.; Talanta 2005,65,1226.
• 27. Gouda, A. A.; Amin, A. S.; Arab. Chem. J. 2010,3,159.
• 28. Lee, G.; Rossi, M. V.; Coichev, N.; Moya, H. D.; Food Chem. 2011,126,679.
• 28. Zhang, H.; Zhang, Y.; Xianjun, W.; Xinzhen, D.; Quanmin, L.; Chin. J. Chem. 2009,27,518.
• 29. Wheate, N. J.; Vora, V.; Anthony, N. G.; McInnes, F. J.; J. Inclusion Phenom. Macrocyclic Chem. 2010,68,359.
• 30. Greenwood, N. N.; Earnshaw, A.; Chemistry of the Elements, 2^nd ed; Butterworth-Heinemann: Oxford, UK, 1997.
• 31. Amos, R. I. J.; Yates, B. F.; Gourlay, B. S.; Schiesser, C. H.; Smith, J. A.; Chem. Commun. 2008,1695.
• 32. Reisbig, R. R.; Woody, A.; Young, M.; Woody, R. W.; Biochemistry 1982,21,196.
• 33. Valderrama, P.; Braga, J. W. B.; Poppi, R. J.; Quim. Nova 2009,32,1278.

Publication Dates

• Publication in this collection
  31 July 2013
• Date of issue
  July 2013

History

• Received
  29 Jan 2013
• Accepted
  25 June 2013