Anal. Chem. 1999, 71, 319-326
Micellar Liquid Chromatography: A Worthy Technique for the Determination of β-Antagonists in Urine Samples I. Rapado Martı´nez, R. M. Villanueva Caman˜as, and M. C. Garcı´a Alvarez-Coque*
Departamento de Quı´mica Analı´tica, Facultad de Quı´mica, Universitat de Valencia, 46100 Burjassot, Valencia, Spain Several β-antagonists (acebutolol, atenolol, celiprolol, labetalol, metoprolol, nadolol, propranolol) were determined in urine samples with fluorometric detection after direct injection, in less than 15 min, with a micellar mobile phase of 0.1 M sodium dodecyl sulfate (SDS), 15% propanol, and 1% triethylamine at pH 3. The limits of detection (3s criterion) were usually between 3 and 30 ng/mL. The addition of propanol and triethylamine and the reduction of the pH of the mobile phase improved the efficiency of the chromatographic peaks that was rather low in pure micellar eluents. The selection of the composition of the mobile phase was easily performed through the use of an interpretive procedure which considered the retention times and peak shapes of the β-antagonists in six chromatograms, obtained at varying concentrations of SDS (0.05-0.15 M) and propanol (5-15% v/v). The chromatograms of urine samples from healthy volunteers, which were administered atenolol, metoprolol, and propranolol, showed only one peak for the former drug and several peaks for the other two. These peaks corresponded to the parent drug and metabolites, which indicated the partial and the extensive degradation of metoprolol and propranolol, respectively. β-Antagonists are drugs that have a high affinity and specifity on β-adrenoceptor sites and inhibit both the β-adrenergic sympathetic activity and the response to β-adrenergic agonist drugs. β-Antagonists have important pharmacological properties at different levels, such as cardiovascular, bronchial, uterine, metabolic, hormonal, renal, and neural, all of which depend on their hydrophobicity. The compounds show good absorption following oral administration, and their elimination is performed via the kidney. Hydrophobic β-antagonists (e.g., metoprolol, oxprenolol, propranolol, and timolol) are rapidly metabolized in the liver, so that the excreted amount in urine of the parent drug is very low. The most hydrophilic β-antagonists (e.g., acebutolol, atenolol, carteolol, celiprolol, and nadolol) are eliminated practically unchanged and their therapeutic action is longer because of the lower degree of their metabolization. Analytical methods for the screening and determination of β-antagonists in physiological fluids have been investigated over the years. Procedures using UV-visible,1-3 fluorescence,4 and (1) Wallace, J. E. Anal. Chem. 1968, 40, 978. (2) Dill, W. A.; Chucot, L.; Chang, T.; Glazko, A. Clin. Chem. 1971, 17, 1200. 10.1021/ac980472k CCC: $18.00 Published on Web 12/05/1998
© 1999 American Chemical Society
phosphorescence5 spectrometry, thin-layer6,7 and gas8-11 chromatography, and capillary electrophoresis12 have been reported. However, the most successful methods recommended to perform these analyses utilize high-performance liquid chromatography (HPLC) with spectrophotometric or fluorometric detection.13-17 Electrochemical detection has also been proposed.18,19 In these reports, usually only one β-antagonist was determined and occasionally some metabolites. The main problem associated with these determinations is the laborious cleanup procedure required prior to HPLC analysis. The preparation of the sample often includes liquid-liquid or solidphase extraction to isolate and preconcentrate the analytes. These operations may be automated or a column-switching technique can alternatively be used, which allows the direct injection of the physiological samples in the chromatographic system. On-line micellar chromatographic cleanup, followed by a reversed-phase analytical separation, has been reported for the determination of propranolol in urine samples.20,21 This technique, however, requires additional instrumentation and an internal standard to correct the error produced from incomplete recovery of the analyte from the matrix. On the other hand, the wide range of hydro(3) Tan, H. S. I.; Shelton, D. J. Pharm. Sci. 1974, 63, 916. (4) Schafer, M.; Geissler, H. E.; Mutschler, E. J. Chromatogr., Biomed. Appl. 1977, 143, 607. (5) Morrison, L. D.; O’Donell, C. M. Anal. Chem. 1974, 46, 1119. (6) Wesley-Hadzija, B.; Mattocks, A. M. J. Chromatogr., Biomed. Appl. 1977, 143, 307. (7) Lee, K. Y.; Nurok, D.; Zlatkis, A.; Karmen, A. J. Chromatogr. 1978, 158, 403. (8) Huffmann, D. H.; Hignite, C. E. Clin. Chem. 1976, 22, 810. (9) Maurer, H.; Pfleger, K. J. Chromatogr., Biomed. Appl. 1986, 382, 147. (10) Leloux, M. S.; Maes, R. A. A. Biomed. Environm. Mass Spectrom. 1990, 19, 137. (11) Nagasawa, M.; Kashimoto, M.; Sugawara, M.; Kimura, Y. J. Chromatogr., B 1995, 673, 294. (12) Lukkari, P.; Sire´n, H.; Panstar, M.; Riekkola, M. L. J. Chromatogr. 1993, 632, 143. (13) Buskin, J. N.; Upton, R. A.; So¨rgel, F.; Williams, R. L. J. Chromatogr., Biomed. Appl. 1982, 230, 454. (14) Bu ¨ hring, K. U.; Garbe, A. J. Chromatogr., Biomed. Appl. 1986, 382, 215. (15) Winkler, H.; Ried, W.; Lemmer, B. J. Chromatogr., Biomed. Appl. 1982, 228, 223. (16) Betnall, A. E.; Cowen, T. Anal. Proc. 1993, 30, 367. (17) Sire´n, H.; Saarinen, M.; Hainari, S.; Lukkari, P.; Riekkola, M. L. J. Chromatogr. 1993, 632, 215. (18) Wang, J.; Bonakdar, M.; Deshmukh, B. K. J. Chromatogr., Biomed. Appl. 1985, 344, 412. (19) Maguregui, M. I.; Alonso, R. M.; Jime´nez, R. M. J. Chromatogr., B 1995, 674, 85. (20) Posluszny, J. V.; Weinberger, R. Anal. Chem. 1988, 60, 1953. (21) Posluszny, J. V.; Weinberger, R. J. Chromatogr. 1990, 507, 267.
Analytical Chemistry, Vol. 71, No. 2, January 15, 1999 319
phobicity of β-antagonists makes the sample preparation step and the chromatographic separation difficult. Micellar liquid chromatography (MLC) represents an attractive alternative to conventional aqueous-organic mobile phases in clinical analysis.22,23 Micellar eluents offer both enhanced selectivity and mode of detection. The elution of both hydrophobic and hydrophilic (polar and ionic) analytes, in the same chromatogram, is possible without the need of gradient elution, and the direct injection of physiological samples is feasible owing to solubilization of proteins by the micelles and monomers of surfactant. In addition, the protein-bound drugs, as in the case of β-antagonists, are also displaced by the surfactant and released for partitioning to the stationary phase. The addition of an organic solvent to micellar mobile phases of ionic surfactants, such as propanol, solves the problem of the limited eluent strength and gives rise to efficiency enhancements.24,25 In previous work, the performance of sodium dodecyl sulfate (SDS)-propanol micellar and methanol-water mobile phases was compared in the determination of β-antagonists in pharmaceuticals.26 Micellar eluents were demonstrated to be advantageous because of the higher efficiencies (a 10-fold increase in efficiency was obtained for some drugs) and the possibility of eluting a mixture of nine β-antagonists with the same mobile phase, with retention times below 15 min. The wide range of retention of these compounds with methanol-water mobile phases compelled the use of at least two mobile phases of different eluent strength in order to elute the β-antagonists in isocratic mode in a reasonable time. In this work, it is shown how the reduction of pH and the addition of an amine lead to further improvements in the efficiencies of the β-antagonists. The amine also favors the elution of the most retained compounds. The determination of β-antagonists in urine with mobile phases containing SDS, propanol, and triethylamine represents a good example of the simplicity of the optimization of an analytical procedure in MLC. The procedure shown has clear advantages against conventional HPLC procedures and allows the direct and rapid analysis of urine samples. EXPERIMENTAL SECTION Apparatus. A Hitachi F-4500 (Tokyo, Japan) fluorometer was used to obtain the excitation and emission spectra of the drugs. The excitation and emission slits were both set at 5 nm, and a cutoff filter removed all light below 290 nm. The liquid chromatograph was from Hewlett-Packard (HP 1050) and was equipped with an isocratic pump, autosampler (HP 1100), integrator (HP 3396A), and fluorescence detector (HP 1046A) (Palo Alto, CA). The selected instrumental variables were the following: The flow rate was 1 mL/min and the injection volume, 20 µL. The excitation wavelength was 230 nm and the emission wavelength, 300 nm for atenolol, metoprolol, and nadolol, 300 or 340 nm for propranolol, and 440 nm for acebutolol, celiprolol, and labetalol. The flash frequency of the xenon lamp was 220 Hz, and the response time (time period during which (22) Westerlund, D., Chromatographia 1987, 24, 155. (23) Cline-Love, L. J.; Fett, J. J. J. Pharm. Biomed. Anal. 1991, 9, 323. (24) Dorsey, J. G.; DeEchegaray, M. T.; Landy, J. S. Anal. Chem. 1983, 55, 924. (25) Kord, A. S.; Khaledi, M. G. Anal. Chem. 1992, 64, 1894. (26) Rapado Martı´nez, I.; Garcı´a Alvarez-Coque, M. C.; Villanueva Caman ˜as, R. M. J. Chromatogr., A 1997, 765, 221.
320 Analytical Chemistry, Vol. 71, No. 2, January 15, 1999
Table 1. Structures, Protonation Constants (log K) and Octanol-Water Partition Coefficients (log P) of Some β-Antagonists
a
Reference 30; E, experimental value; C, calculated value.
data points are summed and averaged in data processing), 4000 ms. The photomultiplier gain was approximately 26, 26, and 28 for the drugs detected at 300, 340, and 440 nm, respectively. The slits were 2 mm × 2 mm before the flow cell, 4 mm × 4 mm after the flow cell, and 4 mm × 4 mm before the photomultiplier tube. A Spherisorb ODS-2 analytical column (5 µm, 120 × 4.6 mm i.d.) and a guard column of similar characteristics (35 × 4.6 mm i.d.) from Scharlau (Barcelona, Spain) were used. The dead volume was determined by measurement of the first statistically significant deviation of the baseline, following the injection of the sample. Data acquisition was made with the Peak-96 software from Hewlett-Packard (Avondale, PA). Data were treated with MICHROM, a program developed in our laboratory.27 Reagents. Stock solutions containing 100 µg/mL of the β-antagonists acebutolol chlorhydrate (AC), atenolol (AT), carteolol chlorhydrate (CA), celiprolol chlorhydrate (CE), labetalol chlorhydrate (LA), metoprolol tartrate (ME), nadolol (NA), oxprenolol chlorhydrate (OX), propranolol chlorhydrate (PR), and timolol maleate (TI), which were kindly donated by several manufacturers (Table 1), were prepared in a 0.05 M SDS (99% purity, Merck, Darmstadt, Germany) solution. The following reagents were used for the preparation of the mobile phases: SDS, 1-propanol (for analysis, Scharlau), triethylamine (99.5% purity, Fluka, Buchs, Switzerland), sodium monohydrogen phosphate (for analysis, Scharlau), and HCl (for analysis, (27) Torres Lapasio´, J. R.; Garcı´a Alvarez-Coque, M. C.; Baeza Baeza, J. J. Anal. Chim. Acta 1997, 348, 187.
Panreac, Barcelona, Spain). A solution of SDS, triethylamine, and monohydrogen phosphate was first prepared and the pH was adjusted with HCl before the addition of propanol. The column was washed weekly with 50 mL of water to eliminate the surfactant and afterward with 50 mL of methanol (HPLC grade, Scharlau). After one month of use, the column was also cleaned with chloroform (HPLC grade, Panreac) to eliminate strongly retained compounds. The column was stored with methanol. The mobile phases were filtered through Nylon membranes of 0.45 µm, 47 mm in diameter (MSI, Westboro, MA). The β-antagonist solutions were also filtered before their injection into the chromatographic system, through cellulose acetate membranes of 0.45 µm, 25 mm in diameter (MSI), which were previously conditioned by passing through 3 mL of the β-antagonist solution. All the solutions were prepared in Nanopure water (Barnstead, Sybron, Boston, MA). RESULTS AND DISCUSSION Fluorescence of the β-Antagonists. The excitation and emission spectra for several β-antagonists were obtained in micellar solutions of SDS. Atenolol, labetalol, metoprolol, nadolol, and propranolol showed intense fluorescence signals in a micellar medium of SDS containing up to 15% propanol; acebutolol and celiprolol were less fluorescent. The fluorescence of carteolol, oxprenolol, and timolol was too weak to make their detection possible; therefore, these three compounds were not included in any further studies. The excitation and emission maximum wavelengths, respectively, were as follows: acebutolol, 240 and 454 nm; atenolol, 225 and 300 nm; celiprolol, 240 and 484 nm; labetalol, 210 and 434 nm; metoprolol, 225 and 300 nm; nadolol, 220 and 300 nm; and propranolol, 225 and 338 nm. The shape and intensity of the bands of the fluorescence emission spectra were not modified in the 3-8 pH range, except for labetalol. For this compound, the fluorescence intensity decreased as pH was reduced. Only the less fluorescent drugs acebutolol and celiprolol showed an important fluorescence enhancement in factors of ×7.6 and ×6.9 in 0.1 M SDS media. This improved their quantitation. The sensitivity for the other β-antagonists was scarcely enhanced (×1.1-1.3). Effect of the Surfactant and Alcohol on the Chromatographic Retention and Efficiency of the β-Antagonists. It was observed, upon increasing the concentration of SDS, that the eluent strength of this surfactant was considerable. This behavior is typical of cationic compounds chromatographed with anionic micellar mobile phases. However, the β-antagonists were not eluted in reasonable retention times even at the larger concentrations of SDS used (the retention factor, k, was 115 for propranolol and 0.15 M SDS). For SDS concentrations greater than 0.15 M, the column pressure became a problem. Also, the chromatographic peaks displayed a high asymmetry and the efficiencies were extremely low when the mobile phase contained only SDS. The addition of alcohols, such as propanol, butanol, and pentanol to the mobile phase, decreased the retention. Among the three alcohols, propanol was checked to produce the best well-shaped peaks.26 Effect of pH and Triethylamine on the Efficiency. The cationic nature of β-antagonists produces low efficiencies in conventional HPLC with aqueous-organic mobile phases, owing to the interaction of these compounds with the free silanol groups on the alkyl-bonded reversed-phase packings. The reduction of
pH and the addition of triethylamine to protonate or bind silanol groups is a common practice to decrease peak tailing.28,29 The log protonation constants of β-antagonists in water are in the 9-10 range (Table 1); the presence of micelles probably increased these constants, owing to the stabilization of the protonated positively charged species by association with the anionic micelles. Accordingly, the pH of the micellar mobile phase did not affect the retention of the β-antagonists, which were protonated in the entire working pH range of a C18 column. Nevertheless, an interesting effect appeared: the efficiencies of the chromatographic peaks increased and their asymmetries decreased when the pH was reduced. This effect was observed previously for catecholamines.31 In MLC, two reports have been published where the micellar mobile phase contained triethylamine. In one of these reports, propranolol was determined in urine by using a Brij-35 eluent, but the efficiency still remained too low (N ) 218 for 0.08 M Brij35 and 2% triethylamine at pH 5).23 In the other report, eluents of SDS and 2-propanol were used to determine 2-(ethylhexyl)-4(dimethylamino)benzoate, 2-(ethylhexyl)-4-methoxycinnamate, and 2-hydroxy-4-methoxybenzophenone in sunscreen cosmetic products.32 In a series of experiments, the concentration of all the components of the eluent, except triethylamine, were fixed (0.12 M SDS, 15% propanol, and 0.02 M phosphate buffer at pH 7). The percentage of the amine was changed in the 0-2% (v/v) range. With these mobile phases, appropriate eluent strengths were obtained for all the β-antagonists. The addition of 0.2% triethylamine improved the efficiency and decreased the asymmetry factors of the chromatographic peaks. For larger concentrations of triethylamine, the efficiency scarcely changed, except for acebutolol and celiprolol. For these compounds, the efficiency increased as the concentration of triethylamine increased in the whole 0-2% range. Figure 1 illustrates the values of efficiency for several β-antagonists and four mobile phases of diverse composition. The reduction of pH and the addition of triethylamine increased the efficiency. For most compounds studied, the largest values were obtained in an acidic eluent containing the amine. The conditions selected to elute the β-antagonists with micellar mobile phases were pH 3 and 1% triethylamine. Effect of Triethylamine on Retention. When an anionic surfactant is used to form micelles in the mobile phase, the effects of the presence of the amine on the chromatographic system, composed of stationary phase, eluent, and analytes, are very complex. The amine will associate, in competition with the analytes, not only with the silanol groups of the bonded phase and the monomers of surfactant adsorbed on the column but also with the polar heads of the surfactant in the micellar aggregates. The retention times of the analytes will be modified, in a greater or lesser extent, depending on the strength of the diverse (28) Piotrovskii, V. K.; Belolipetskaya, V. G.; El′man, A. R.; Metelitsa, V. I. J. Chromatogr. 1983, 278, 469. (29) Hatmann, C.; Kraus, D.; Spahn, H.; Mutschler, E. J. Chromatogr. 1989, 496, 387. (30) Drayton C. J., Ed. Comprehensive Medicinal Chemistry; Pergamon: Oxford, 1990; Vol. 6. (31) Villanueva Caman ˜as, R. M.; Sanchis Mallols, J. M.; Torres Lapasio´, J. R.; Ramis Ramos, G. Analyst 1995, 120, 1767. (32) Tomasella, F. P.; Pan, Z.; Cline-Love, L. J. J. Chromatogr. 1991, 587, 325.
Analytical Chemistry, Vol. 71, No. 2, January 15, 1999
321
Figure 1. Efficiency values for the β-antagonists eluted with mobile phases of 0.12 M SDS, 15% propanol, and 0.02 M phosphate buffer. The order of elution of the compounds is the same for all the mobile phases shown.
interactions and the concentrations of surfactant, organic modifier, and amine. The elution order of the β-antagonists was almost the same with and without triethylamine for mobile phases of SDS and propanol: atenolol, nadolol, acebutolol, celiprolol, metoprolol, labetalol, and propranolol. Only the two latter compounds (the most hydrophobic) changed their order of elution when mobile phases at low concentrations of surfactant and modifier were used. Nadolol showed two peaks with mobile phases without propanol or containing a low concentration of SDS and alcohol (below 5%). However, only one peak was observed when triethylamine was added to mobile phases of SDS with a low concentration of propanol. The peaks corresponded to the two diastereoisomers of nadolol: +-/-+ (RS/SR) and ++/-- (RR/SS). The commercial drug is a mixture of approximately equal proportions of both. The addition of 0.2% triethylamine to a mobile phase of 0.12 M SDS and 15% propanol at pH 7 accelerated the elution of all compounds, especially for acebutolol (k ) 7.0 to 5.6), celiprolol (10.3 to 7.1), metoprolol (10.0 to 8.0), and propranolol (15.7 to 13.8). A larger amount of the amine produced a further but smaller reduction in the retention. A linear relationship was found when the reciprocal of the retention factor (1/k) was plotted against the concentration of the amine, in the 0.2-2% concentration range. Therefore, triethylamine behaved as a modifier, analogous to the alcohol, whose effect on the elution of the solutes in micellar mobile phases can be accurately predicted.33,34 The retention times of the β-antagonists were also obtained at pH 3 for two mobile phases of the same composition as above, in the absence and presence of triethylamine. Similarly to pH 7, the amine produced a reduction in retention times, with retention factors almost identical for the two pH values. It was thus again demonstrated that the retention did not depend on the pH of the mobile phase. Plots of 1/k vs micellar SDS concentration were found to be linear for all β-antagonists, for mobile phases at pH 3 containing (33) Torres Lapasio´, J. R.; Villanueva Caman ˜as, R. M.; Sanchis Mallols, J. M.; Medina Herna´ndez, M. J.; Garcı´a Alvarez-Coque, M. C. J. Chromatogr. 1993, 639, 87. (34) Garcı´a Alvarez-Coque, M. C.; Torres Lapasio´, J. R.; Baeza Baeza, J. J. Anal. Chim. Acta 1996, 324, 163.
322 Analytical Chemistry, Vol. 71, No. 2, January 15, 1999
Figure 2. Chromatograms of urine matrix obtained with a mobile phase of 0.1 M SDS, 15% propanol, 1% triethylamine, and 0.02 M phosphate buffer: (a) without dilution at pH 3 (1) and pH 7 (2), (b) 1:25 dilution at pH 3. The excitation and emission wavelengths were 230 and 300 nm, respectively.
5 or 15% propanol, and three or four SDS concentrations, previously and after the addition of triethylamine (r > 0.999). This indicated that the nature of the stationary phase was unchanged inside each series. It was observed that once triethylamine was utilized, mobile phases without the amine resulted in retention times being approximately 10% smaller with respect to those mobile phases employed before the addition of triethylamine to the system. This was attributed to an irreversible modification of the stationary phase, where triethylamine was associated to the remaining monomers of surfactant adsorbed on the alkyl-bonded phase or to the free silanol groups on the column. It was also checked that adequate correlations existed between log octanol-water partition coefficients (log P, Table 1) and log k for diverse mobile phases containing SDS and propanol and that the correlation coefficient slightly increased in the presence of triethylamine in the mobile phase (e.g., for a 0.15 M SDS, 15% propanol, 1% triethylamine at pH 3, r ) 0.977, and without amine r ) 0.962). Analysis of Urine Samples. Effect of pH and Sample Dilution on the Chromatogram of the Urine Matrix. Figure 2a illustrates chromatograms of the background of a urine matrix eluted with a mobile phase of 0.1 M SDS, 15% propanol, 1% triethylamine, and 0.02 M phosphate buffer, at pH 3 and 7. The urine samples were injected into the chromatograph without any
other treatment than direct filtration into the autosampler vials, through cellulose acetate membranes. The beginning of the chromatograms showed some differences. First, the intensity of the peaks appearing between 2 and 4 min was lower at pH 7 with respect to pH 3, although the appearance of these peaks depended on the selected emission wavelength. Thus, at 440 nm, this background signal was not so apparent. However, at retention times longer than 4 min, the chromatogram of urine at pH 7 presented a wide perturbation of the baseline, which was reproduced in all the urine samples analyzed. As observed, in this region of the chromatogram, a clean background was obtained for the urine matrix eluted at pH 3. Therefore, this pH was selected to perform the analysis of the β-antagonists in urine. It should be reminded that at this pH the efficiencies were greater. Although MLC permits the direct injection of physiological samples, the injection of a large number of samples can produce damage to the packing material, thus shortening the life of the column, or can force a frequent regeneration of the stationary phase with water, methanol, and chloroform. An experiment was done where relatively large volumes (20 µL) of urine sample were injected into the column without dilution (40-50 injections). The mobile phase (a volume of approximately 500 mL) was recycled during 4-5 days through the chromatographic system. Under these conditions, changes in the retention times were observed for some β-antagonists. Normal values of the retention times were, however, again obtained after cleaning the system and changing the solution used as mobile phase. It was decided to perform the analysis of urine samples after dilution with the mobile phase. A reduction of the background of the diluted urine matrix facilitated the measurement of the area of the chromatographic peaks, especially for those β-antagonists less retained. Figure 2b shows the chromatogram of the urine matrix diluted in a 1:25 factor. The background signal was very low for retention times above 2 min. The relatively high concentration of some β-antagonists found in urine, after administration of the drugs, and the sensitivity provided by the fluorometric monitoring, permitted the adequate detection of these compounds after making this dilution. Nevertheless, the direct analysis of urine samples without dilution can be convenient and feasible in some cases and is possible, as shown in Figure 3a. The chromatograms in Figures 2 and 3 show two peaks at retention times shorter than 4 min, which are due to endogenous compounds found in the physiological matrix (EN1 and EN2 ). These peaks were considered in the optimization process explained below. Selection of the Concentration of SDS and Propanol. The resolution of the peaks of a mixture of the β-antagonists acebutolol, atenolol, celiprolol, labetalol, metoprolol, nadolol, and propranolol, and the two endogenous compounds showing the most prominent peaks, was optimized against the concentration of SDS and propanol in the micellar mobile phase, by using an interpretive strategy that required knowledge of the retention behavior of the compounds. Previous work has shown the validity of this approach.33,35 The equations of retention for each solute were obtained from the retention factors in six mobile phases that contained 1% triethylamine and 0.02 M phosphate buffer at pH 3, (35) Torres Lapasio´, J. R.; Villanueva Caman ˜as, R. M.; Sanchis Mallols, J. M.; Medina Herna´ndez, M. J.; Garcı´a Alvarez-Coque, M. C. J. Chromatogr., A 1994, 677, 239.
Figure 3. (a) Chromatogram of a urine sample with 1 µg/mL of each β-antagonist injected without dilution. (b) Simulated chromatogram. Chromatograms of 1:25 diluted urine samples with (c) acebutolol (2.5 µg/mL), celiprolol (2.5 µg/mL), and labetalol (0.5 µg/mL); (d) atenolol (1.2 µg/mL), nadolol (1.3 µg/mL), metoprolol (0.2 µg/mL), and propranolol (0.06 µg/mL). EN1 and EN2 are endogenous compounds found in urine. Mobile phase: 0.1 M SDS, 15% propanol, 1% triethylamine, and 0.02 M phosphate buffer at pH 3. The excitation wavelength was always 230 nm, the emission wavelength was 300 nm for atenolol, nadolol, metoprolol, and propranolol, and 440 nm for acebutolol, celiprolol, and labetalol.
and the following concentrations of surfactant and alcohol: 0.05 M SDS/5% propanol, 0.075 M SDS/5% propanol, 0.15 M SDS/5% propanol, 0.1 M SDS/10% propanol, 0.05 M SDS/15% propanol, and 0.15 M SDS/15% propanol. The retention factors of the compounds were fitted to
1/k ) c0 + c1 µ + c2φ + c3 µφ + c4φ2
(1)
where µ is the molar concentration of total surfactant and φ the percentage concentration (v/v) of modifier. With this equation, the global fitting errors were below 1.6% and the correlation coefficients of the nonlinear fittings, r > 0.999, for all the compounds studied. A more simple model without the φ2 term yielded larger errors. An overlapping criterion was used to optimize the global resolution.35 This criterion takes into account the position and shape of the chromatographic peaks to simulate the chromatograms, for mobile phases in the variable space comprising the concentrations of SDS and propanol. Equation 1 was used to predict the position of the peaks for a given mobile phase, whereas their shape was predicted by using an asymmetrical Gaussian function where the standard deviation depended on the efficiency and asymmetry factors:36 (36) Torres Lapasio´, J. R.; Baeza Baeza, J. J.; Garcı´a Alvarez-Coque, M. C. Anal. Chem. 1997, 69, 3822.
Analytical Chemistry, Vol. 71, No. 2, January 15, 1999
323
[
h(t) ) H exp - (1/2)
(
t - tR
s0 + s1 (t - tR)
)] 2
(2)
where H and tR are the height and time at the peak maximum, respectively, s0 is the standard deviation of a symmetrical Gaussian peak describing the central region of the experimental peak, and s1 is a coefficient that quantitates its skewness. The values of efficiency and asymmetry factor were interpolated from the data obtained in the three experimental mobile phases closer to the simulated mobile phase. The simulation of chromatograms for mobile phases at varying concentrations of surfactant and modifier was facilitated with the program MICHROM.27 With this program, the changes produced in the chromatograms are observed as a graphical cursor is moved inside the rectangle representing the mobile-phase composition range. Maximum resolution was achieved in two regions of the variable space, the first in the 0.05-0.06 M SDS and 5-11% propanol concentration ranges, and the second in a wide region near 15% propanol which crosses the whole range of SDS concentrations. The chromatographic peaks of the mixture of β-antagonists were adequately resolved for the optimum mobile phase (0.05 M SDS/6.3% propanol), but the eluent strength was too low (Figure 4a). With eluents containing 15% propanol and 0.1 M SDS (Figure 3) or 0.15 M SDS (Figure 4b), which originated relative resolution maximums, the retention times were largely decreased. In both cases, good resolution was obtained, especially for the 0.1 M SDS eluent, except for labetalol and propranolol. These β-antagonists appeared partially or completely overlapped for any mobile phase inside the variable space studied, with the exception of those having a low concentration of SDS and propanol, where long retention times were obtained. The mobile phase of 0.1 M SDS, 15% propanol, 1% triethylamine, and 0.02 M phosphate buffer at pH 3 was finally selected for the determination of the β-antagonists in urine samples. This mobile phase gave an adequate separation of the drugs and endogenous compounds, with analysis times below 15 min. Chromatograms of urine samples where several β-antagonists were added are depicted in Figure 3 for the selected mobile phase. Figures of Merit. Calibration curves were constructed for each β-antagonist, using the measured areas of the chromatographic peaks at six increasing concentrations. The concentration ranges were the following: acebutolol chlorhydrate, 0.1-4.2 µg/ mL; atenolol, 0.05-2.1 µg/mL; celiprolol chlorhydrate, 0.5-4.2 µg/mL; labetalol chlorhydrate, 0.1-0.9 µg/mL; metoprolol tartrate, 0.05-0.5 µg/mL; nadolol, 0.05-2.2 µg/mL; and propranolol chlorhydrate, 0.004-0.1 µg/mL. The curves were obtained for aqueous solutions and for urine samples with added β-antagonists. All the solutions contained the same concentrations of surfactant and alcohol as those found in the mobile phase, and the dilution factor in the urine samples was 1:25. The slopes of the calibration curves in the absence and presence of urine were similar, the intercepts were statistically zero in all cases, and the regression coefficients were r > 0.99. Therefore, no matrix effect existed in the urine samples. The variations among calibration curves obtained on different days were greater than among calibration curves constructed during the same day, in the presence and absence of urine. 324 Analytical Chemistry, Vol. 71, No. 2, January 15, 1999
Figure 4. Simulated chromatograms for mobile phases of 1% triethylamine and 0.02 M phosphate buffer at pH 3; (a) 0.05 M SDS/ 6.3% propanol; (b) 0.15 M SDS/15% propanol. Table 2. Limits of Detection and Repeatabilities in Intra- and Interday Assays for β-Antagonistsa CV (%) compound
intraday
interday
LOD (ng/mL)
acebutolol atenolol celiprolol labetalol metoprolol nadolol propranolol
0.96 1.01 1.02 2.9 2.4 1.41 0.88
7.5 9.1 4.3 6.8 7.1 9.6 4.8
30 19 200 20 16 8 3
a Eluted with a 0.1 M SDS, 15% propanol, and 1% triethylamine (pH 3) mobile phase.
Repeatability values were calculated by measuring the areas of the peaks obtained by injection of series of five urine samples with an intermediate concentration of β-antagonist in the calibration range indicated above. The variation coefficient was always below 3% (Table 2). The interday repeatabilities were also obtained. The variation coefficients of the peak areas in samples injected during six different days were in the 4-10% range. The limits of detection (LODs, 3s criterion) were evaluated by injection of series of 10 solutions, containing the β-antagonists at the lower concentration of the calibration curve, in the presence and absence of urine. The LODs were similar in both cases, with values between 3 ng/mL for propranolol and 0.2 µg/mL for
Figure 5. Chromatograms of healthy volunteers which were administered (collection time with respect to dosing): (a) 50 mg of atenolol (8 h), (b) 100 mg of metoprolol tartrate (MT1 and MT2 are metabolites of metoprolol) (3 h), (c, d) 10 mg of propranolol (MT1, MT2, and MT3 are metabolites of propranolol) (2 h). The urine sample was diluted (1:25 dilution) in (a-c) and was directly injected without dilution in (d). Mobile phase: 0.1 M SDS, 15% propanol, 1% triethylamine, and 0.02 M phosphate buffer at pH 3.
celiprolol (Table 2). The low values of the LODs achieved for most β-antagonists with the method proposed in this work permitted the detection and quantitation of these drugs in urine, without the need of making a previous treatment of the physiological samples. Application of the Method to the Study of the Urinary Excretion of β-Antagonists. The possibility of detection and quantitation of β-antagonists in urine samples, collected after administration of a single oral dose at a low-therapeutic concentration, was investigated. For this study, three normal healthy volunteers were given 50 mg of atenolol, 100 mg of metoprolol tartrate, or 10 mg of propranolol chlorhydrate. Urine samples were taken during 48 h at different time intervals (during the first 6 h in 1-h intervals and, afterward, in longer intervals), and the excreted volume was measured. The samples were usually refrigerated at 4 °C before being analyzed. Samples analyzed after 12 h were frozen at -24 °C. No preservatives were added to prevent degradation and change the pH of the urine samples. Figure 5 shows chromatograms of solutions prepared by a 1:25 dilution of urine samples coming from individuals who had taken atenolol and metoprolol. The peak of atenolol appeared in the samples collected 1 h after ingestion and could still be detected in samples taken after 48 h. The peak of metoprolol was observed between 1 and 13 h after ingestion. After 24 h, the total excreted amount for atenolol was 19.2 mg, approximately 38% of the oral dose, and for metoprolol, 1.1 mg, which means that only 1% of the parent drug was excreted. In contrast, the peak of propranolol was not observed (Figure 5c). This β-antagonist could, however, be detected when the urine sample was injected into the chromatograph without any dilution (Figure 5d).
On the other hand, the chromatograms of the urine samples of the volunteers who were administered metoprolol or propranolol showed several new peaks, different from those of the β-antagonists (see Figure 5b,c), which were not observed in the samples collected just before the administration of the drugs. The new peaks increased with the first samples, were reduced, and finally disappeared in the samples taken after some hours. These peaks were attributed to diverse metabolites of the drugs.37,38 Two peaks were particularly intense for metoprolol, with k ) 2.5 and 3.3, and three for propranolol, with k ) 4.2, 4.5, and 6.2. Atenolol is scarcely metabolized in the liver, being mainly excreted unaltered via the kidneys. Accordingly, only the peak of the drug was observed in the corresponding chromatograms. The metabolites of metoprolol and propranolol were not available in our laboratory. However, a tentative assignation of the observed peaks was made. Metoprolol is metabolized according to three different mechanisms: oxidation, loss of an amine group, and aliphatic hydroxylation.37 The three processes originate compounds of low pharmacological activity. The metabolites produced have a structure showing a higher polarity than the parent drug, and therefore, their elution should occur at shorter retention times, as the peaks observed at 2.9 and 3.6 min (Figure 5b). Figure 5c,d indicates the extensive degradation of propranolol that gives rise to three metabolite peaks at an apparently high concentration, leaving behind only a small amount of the drug. Obviously, these metabolites can be better quantitated than the parent drug, but this requires the availability of the corresponding standards. For propranolol, a possible mechanism of formation of their metabolites is the loss of the isopropyl group bonded to the secondary amine, which gives rise to N-desisopropylpropranolol (DIP). This is partially further transformed into an intermediate aldehyde, which gives two products, naphthoxylactic acid (NLA) and propranolol glycol (PGL).38 Another mechanism of conversion is aromatic hydroxylation, which will mainly occur in position 4. However, 4-hydroxypropranolol is not stable in urine and its excitation and emission wavelengths (ex ) 310 nm, em ) 430 nm) are different from those of detection for propranolol and the other metabolites. Consequently, this metabolite probably will not be seen at the wavelength of detection employed for propranolol (ex ) 230 nm, em ) 340 nm). Therefore, taking into account the electrostatic and hydrophobic characteristics of the different metabolites of propranolol and the characteristics of the chromatographic technique, the peaks in the chromatogram shown in Figure 5c were assigned to the following compounds: MT1, the less retained metabolite, would be the less hydrophobic NLA; MT2, with a close retention time should have a similar structure, PGL; and finally, the most retained MT3 would be DIP, which is similar to propranolol. CONCLUSIONS Micellar liquid chromatography permits the determination of β-antagonists in urine samples in less than 15 min, using a mobile phase of SDS at pH 3 containing propanol and triethylamine. The sample preparation step is reduced to the dilution of the sample (if desired), in contrast to other chromatographic procedures using (37) Li, F.; Cooper, S. F.; Coˆte´, M. J. Chromatogr., B 1995, 668, 67. (38) Kwong, E. C.; Shen, D. D. J. Chromatogr., Biomed. Appl. 1987, 414, 365.
Analytical Chemistry, Vol. 71, No. 2, January 15, 1999
325
conventional aqueous-organic eluents that require time-consuming extraction steps or more complex chromatographic systems. The addition of triethylamine and the use of low values of pH alleviated the main drawback of micellar mobile phases: the low efficiency of the chomatographic peaks produced by slow mass transfer between stationary and mobile phases. These new modes of remediation of reduced efficiency should be added to the habitual use of alcohols and high temperature.39 The excellent correlation between log P and log k indicated that the hydrophobic interactions are the main reasons for the differences in retention of the β-antagonists. The modeling of the retention behavior is
feasible with micellar eluents, which makes extremely simple the optimization of mobile-phase composition (surfactant and modifier concentrations). The procedure developed for the detection of β-antagonists in urine samples should be considered as a good example of the usefulness of micellar eluents in drug control.
(39) Berthod, A. J. Chromatogr., A 1997, 780, 191.
AC980472K
326
Analytical Chemistry, Vol. 71, No. 2, January 15, 1999
ACKNOWLEDGMENT This work was supported by the DGES, Project PB97/1384. Received for review April 30, 1998. Accepted October 15, 1998.
