Quantification of Etoposide and Etoposide Phosphate in Human

Quantification of Etoposide and Etoposide. Phosphate in Human Plasma by Micellar. Electrokinetic Chromatography and Near-Field. Thermal Lens Detection...
0 downloads 0 Views 95KB Size
Anal. Chem. 2004, 76, 3804-3809

Quantification of Etoposide and Etoposide Phosphate in Human Plasma by Micellar Electrokinetic Chromatography and Near-Field Thermal Lens Detection Natalia Y. Ragozina,† Michael Pu 1 tz,† Stefan Heissler,‡ Werner Faubel,‡ and Ute Pyell*,†

Department of Chemistry, Philipps University Marburg, Hans-Meerwein-Strasse, D-35032 Marburg, Germany, and Institute for Instrumental Analysis, Research Center Karlsruhe, P.O. Box 3640, D-76021 Karlsruhe, Germany

A method employing micellar electrokinetic chromatography in combination with near-field thermal lens detection (NF-TLD) was developed for the rapid simultaneous determination of etoposide phosphate and etoposide in human blood plasma, taking teniposide as internal standard. The method developed allows the baseline separation of the solutes of interest from each other and from potential interfering matrix constituents within 4 min. The NF-TLD device employed permits detection of solutes absorbing electromagnetic radiation at λ ) 257 nm in fused-silica capillaries with 75-µm i.d. via the near-field thermal lens effect with LODs of 100 µg L-1 for etoposide phosphate and 170 µg L-1 for etoposide. Comparison of the performance of this detector to the performance of a commercial absorption spectrometric detector working at λ ) 257 nm showed a substantial improvement in detection limits (up to 60-fold improvement) for the near-field thermal lens detector. Etoposide is a podophyllotoxin derivative with antineoplastic activity (topoisomerase II inhibitor). It has clinical value in the treatment of leukemia, lymphoma, germ cell tumors, and cell lung tumors. Etoposide (Figure 1) is poorly soluble in water and has to be formulated in polysorbate 80, poly(ethylene glycol), and ethanol. Etoposide phosphate (Figure 1) is an etoposide prodrug approved for intravenous use by the U.S. Food and Drug Administration in 1996. It is soluble in water at concentrations up to 20 g L-1. Several studies have shown that etoposide phosphate is rapidly (within 15 min) and completely converted into etoposide by the action of alkaline phosphatases in blood and that this prodrug is pharmacokinetically equivalent to etoposide.1 Analytical methods used so far for the determination of etoposide phosphate and etoposide are high-perfomance liquid chromatography with fluorometric,2-4 photometric,5,6 or ampero* To whom correspondence should be addressed. Tel: +49 6421 28 22192. Fax: +49 6421 28 28917. E-mail: [email protected]. † Philipps University Marburg. ‡ Research Center Karlsruhe. (1) Hande, K. R. Eur. J. Cancer 1998, 37, 1514-1521. (2) Thompson, D. S.; Greco, F. A.; Miller, A. A.; Srinivas, N. R.; Igwemezie, L. N.; Hainsworth, J. D.; Schacter, L. P.; Kaul, S.; Barbhaiya, R. H.; Garrow, G. C.; Hande, K. H. Clin. Pharmacol. Ther. 1995, 57, 499-507.

3804 Analytical Chemistry, Vol. 76, No. 13, July 1, 2004

Figure 1. Structures of etoposide phosphate, etoposide, and teniposide.

metric7,8 detection and capillary electrophoresis (CE) with photometric detection and z-shaped detection cell9 or fluorometric detection.10 Soetebeer et al.10 reported in 2001 the first method for the simultaneous determination of etoposide and etoposide phosphate in plasma samples. This method is based on the capillary electrophoretic separation of these two compounds by complexation with borate employing a separation buffer containing 150 mmol L-1 disodiumtetraborate and laser-induced native fluorescence detection at 257 nm. Due to the separation conditions selected, run times of ∼12 min and relatively broad solute zones have been observed. Soetebeer et al.10 used an in-house constructed detection unit. Fluorescence was excited with a frequency-doubled argon ion (3) Robieux, I.; Aita, P.; Sorio, R.; Toffoli, G.; Boiocchi M. J. Chromatogr., B 1996, 686, 35-41. (4) Manouilov, K. K.; McGuire, T. R.; Gordon, B. G.; Gwilt, P. R. J. Chromatogr., B 1998, 707, 342-346. (5) Chabot, G. G.; Armand, J. P.; de Forni, T. M.; Abigerges, D.; Winograd, B.; Igwemezie, L.; Schacter, L.; Kaul, S.; Ropers, J.; Bonnay, M. J. Clin. Oncol. 1996, 14, 2020-2030. (6) Soni, N.; Meropol, N. J.; Pendyala, L.; Noel, D.; Schacter, L. P.; Gunton, K. E.; Creaven, P. J. Clin. Oncol. 1997, 15, 766-772. (7) Kaul, S.; Igwemezie, L. N.; Stewart, D. J.; Fields, S. Z.; Kosty, M.; Levithan, N.; Bukowski, R.; Gandara, D.; Goss, G.; O’Dwyer, P.; Schacter, L. P.; Barbhaiya, R. H. J. Clin. Oncol. 1995, 13, 2835-2841. (8) Kreis, W.; Budman, D. R.; Vinciguerra, V.; Hock, K.; Baer, J.; Ingram, R.; Schacter, L. P.; Fields, S. Z. Cancer Chemother. Pharmacol. 1996, 38, 378384. (9) Mrestani, Y.; Neubert, R. Electrophoresis 1998, 19, 3022-3025. (10) Soetebeer, U. B.; Schierenberg, M. O.; Schulz, H.; Hempel, G.; Andresen, P.; Blaschke, G. Anal. Chem. 2001, 73, 2178-2182. 10.1021/ac0304222 CCC: $27.50

© 2004 American Chemical Society Published on Web 05/15/2004

laser operating at 257 nm with a power of 200 mW. The capillary is illuminated at a length of 1.5 mm and a height of 50 µm with help of a cylindrical quartz lens. The emitted fluorescence radiation is imaged onto a spectrograph with an attached intensified CCD camera. According to the authors, this system permits a significant enhancement of the sensitivity of the detection system compared to systems focusing the excitation radiation into a small volume of less than 100 µm in diameter. The detection limits (S/N ) 3) obtained with this method are 100 µg L-1 for etoposide and 30 µg L-1 for etoposide phosphate. It has to be emphasized that the fluorescence quantum yield of these two compounds is relatively low and that fluorescence is quenched by the presence of anionic micelles.11 Mrestani and Neuberger9 obtained 1 order of magnitude higher detection limits (0.2 mg L-1) for etoposide phosphate in plasma after electrophoretic separation and photometric detection with a z-shaped detection cell at 200 nm. In a previous paper,12 we showed that near-field thermal lens detection (NF-TLD) at 257 nm employing an in-house-constructed detection device makes it possible to detect solutes absorbing at this wavelength with a substantial improvement (up to 30-fold improvement) in detection limits compared to photometric detection at the identical wavelength. The detector was realized by using a frequency-doubled argon ion laser operating at a wavelength of 257 nm and a power of ∼100 mW as pump laser. NF-TLD is a photothermal detection technique. Photothermal detection methods have been proven to offer higher sensitivity and lower detection limits than absorption spectrometric detection for small-volume detection cells.13,14 The successful use of photothermal detection in combination with capillary electromigration separations has been demonstrated by several workers. In 1992, Waldron and Dovichi15 reported excellent detection limits for phenylthiohydantoin amino acids separated by micellar electrokinetic chromatography (MEKC) and detected with a laboratorybuilt photothermal detector. Also, the use of photothermal detection in combination with CEC16 and in combination with nonaqueous CE17 has been reported. Thermal lens detection is based on nonradiative relaxation processes converting the absorbed radiative energy into heat that is released into the surrounding medium creating a local gradient in temperature corresponding to a local gradient in the refractive index.18,19 This effect can be monitored with a second laser beam (probe beam with a wavelength not absorbed by the sample) as a local change in the beam intensity monitored with a photodiode placed behind a pinhole. The development of a thermal lens detector working in the near-field mode is based on the theoretical (11) Ragozina, N.; Pu ¨ tz, M.; Pyell, U., unpublished results. (12) Ragozina, N.; Heissler, S.; Faubel, W.; Pyell, U. Anal. Chem. 2002, 74, 44804487. (13) Nolan, T. G.; Weimer, W. A.; Dovichi, N. J. Anal. Chem. 1984, 56, 17041707. (14) Bornhop, D. J.; Dovichi, N. J. Anal. Chem. 1989, 59, 1632-1636. (15) Waldron, K. C.; Dovichi, N. J., Anal. Chem. 1992, 64, 1396-1399. (16) Qi, M.; Li, X. F.; Stathakis, C.; Dovichi, N. J. J. Chromatogr., A 1999, 853, 131-140. (17) Li, X. F.; Liu, C. S.; Roos, P.; Hanssen, E. B.; Cerniglia, C. E.; Dovichi, N. J. Electrophoresis 1998, 19, 3178-3182. (18) Harris, J. M.; Dovichi, N. J Anal. Chem. 1980, 52, 695-706. (19) Bialkowski, S. E. Photothermal spectroscopy methods for chemical analysis; Wiley: New York, 1996.

work of Wu and Dovichi,20 of Power,21 and of Li et al.22 It was first experimentally realized by Seidel et al.23 employing a modemismatched crossed-beam configuration of the pump (argon ion laser) and probe laser (HeNe laser) beams. The thermal lens effect was observed via a photodiode placed behind a pinhole that is placed only 4 mm apart from the separation capillary. In fluorescence detection, the fraction of absorbed energy that is released as radiation generates the monitored signal, while in NF-TLD, nonradiative relaxation processes are responsible for the signal generation. Hence, photothermal detection can be regarded to be complementary to fluorescence detection. The photothermal signal is directly proportional to the heat conversion yield, which is optimum for low fluorescence yield. For solutes with low fluorescence yield, the photothermal signal is not affected by fluorescence quenching. Consequently, in case of solutes with very low fluorescence yield and significant fluorescence quenching via constituents of the separation electrolyte, photothermal detection can be expected to provide lower limits of detection than laser-induced fluorescence detection when the same laser is employed for fluorescence excitation or thermal lens pumping. In the present paper, we investigate whether MEKC combined with NF-TLD at a detection wavelength of 257 nm can be used for the rapid simultaneous quantification of etoposide and etoposide phosphate in human plasma. We investigate whether the detection limits for the solutes of interest can be improved (compared to absorption spectrometric detection with a commercial detector) by photothermal detection with the detection unit developed in our laboratories. The dependence of the signal height on several instrumental parameters of the detector is studied. Accuracy and precision of the MEKC-NF-TLD method developed was determined with spiked plasma samples. EXPERIMENTAL SECTION Chemicals. Etoposide, etoposide phosphate, and teniposide were obtained fromh Bristol-Myers Squibb (Munich, Germany) and used as received. Separation buffers were prepared from disodium tetraborate (p.A., Fluka, Seelze, Germany), boric acid (Suprapure, Merck, Darmstadt, Germany), and sodium dodecyl sulfate (SDS) (p.A., Fluka). Standard solutions were prepared in acetonitrile and diluted with separation buffer to the final concentration. Acetonitrile (LiChrosolv, gradient grade) was from Merck. MEKC. The SpectraPhoresis 100 CE system was (ThermoQuest) equipped either with the home-built photothermal detector (see section Thermal Lens Detection) or (for comparison) with a conventional UV absorbance detector (Linear UVIS 200, Spectra Physics) equipped with a unit for on-column detection in capillaries. Fused-silica capillaries (Supelco and CS Chromatography Service) of 75-µm inner diameter, 715-mm total length (390 mm to the detector), and 360-µm outer diameter were used. The polyimide coating was removed by heating at the detection window and washed with acetone. Samples were injected in the hydro(20) Wu, S.; N. J. Dovichi, J. Appl. Phys. 1990, 67, 1170-1182. (21) Power, J. F. Appl. Opt. 1990, 29, 52-63. (22) Li, B. C.; Zhang, S. Y.; Fang, J. W.; Shui, X. J. Rev. Sci. Instrum. 1997, 68, 2741-2749. (23) Seidel, B. S.; Faubel, W.; Ache, H. J. J. Biomed. Opt. 1997, 2, 326-331.

Analytical Chemistry, Vol. 76, No. 13, July 1, 2004

3805

dynamic mode, applying vacuum for 0.7 s. Thiourea was used as a marker of the hold-up time. Separations were performed using an aqueous buffer with c(SDS) ) 25 mmol L-1, c(Na2B4O7) ) 5 mmol L-1, and c(H3BO3) ) 25 mmol L-1 (pH ) 8.9) as separation electrolyte and a separation voltage of 30 kV. Sample Preparation. Blank plasma from a female healthy volunteer was used to prepare spiked plasma samples. In procedure 1, 400 µL of plasma was spiked with 10 µL each of standard solutions of etoposide and etoposide phosphate in acetonitrile (c ) 5 mg mL-1). The protein fraction was pecipitated by adding 750 µL of acetonitrile and centrifuged for 10 min. A 250-µL sample of the supernatant was transferred into a clean tube. Water (740 µL) and internal standard (teniposide, dissolved in acetonitrile, 11 µL, c ) 1 g L-1) were added. The resulting volume fraction of acetonitrile in this solution is 16%. Sample preparation procedure 2 was according to the procedure described by Soetebeer et al.10 In this procedure, after centrifugation, the supernatant was dried by evaporating the solvent in a gentle stream of nitrogen. The sample was then reconstituted in 990 µL of water and 10 µL of the internal standard teniposide in acetonitrile (c ) 0,25 g L-1). Thermal Lens Detection. The laboratory-developed photothermal detection device for capillary electrophoresis was described in our previous paper.12 It contains a frequency-doubled Ar+ laser (Lexel 95 SHG), which is used as a pump beam source. This laser system delivers up to 200 mW cw at λ ) 257 nm. The intensity of the pump beam is modulated with an ultraminiature chopping head (model 360 H, Scitec Instruments). The chopper gives simultaneously the reference signal of the lock-in amplifier to permit phase-sensitive detection. A second laser diode emitting in the visible range, intensity stabilized, and guided by a monomode optical fiber (Schaefter and Kirchhoff) is used for the detection of the change in the refractive index. The thermal lens signal is detected with a photodiode behind a pinhole (d ) 200 µm), which is located ∼1 cm behind the capillary. RESULTS AND DISCUSSION Near-Field Thermal Lens Detector. If the concentration of absorbing species is not reduced due to processes induced by the high-energy density in the detection volume (e.g., photoinduced decomposition), the thermal lens signal is expected to be directly proportional to the power of the pump laser beam. To this end, we filled the separation capillary completely with a solution of the solutes etoposide phosphate or teniposide in separation buffer (c ) 6.25 mg L-1). A voltage of 30 kV was applied, and the signal was recorded dependent on the pump laser power (see Figure 2). In a range from 10 to 150 mW, no deviation from linearity was observed. It can therefore be concluded that the solutes of interest are sufficiently photostable and the signal height is not being reduced by photobleaching. In quantitative measurements, a high laser power of 105 mW was selected. With a capillary filled completely with a solution of one of the solutes in separation buffer (c ) 50 mg L-1), the amplified thermal lens signal is strongly dependent on the chopper frequency. In a range from 120 to 30 Hz, the signal for etoposide phosphate at static conditions (no voltage applied) dropped from 90 to 53 relative units. In further measurements, a chopper frequency of 30 Hz was selected. 3806 Analytical Chemistry, Vol. 76, No. 13, July 1, 2004

Figure 2. Dependence of the thermal lens signal on pump laser power. Capillary completely filled with a solution of analyte (c ) 6.25 mg L-1) in separation buffer (c(SDS) ) 25 mmol L-1, c(Na2B4O7) ) 5 mmol L-1, c(H3BO3) ) 25 mmol L-1 (pH ) 8.9)). b, etoposide phosphate; 9, teniposide; chopper frequency, 33 Hz; voltage, 30 kV. Capillary: 715 (390) mm, 75-µm i.d.

Figure 3. Dependence of the thermal lens signal on the velocity of the electroosmotic flow. Capillary completely filled with a solution of analyte (c ) 50 mg L-1) in separation buffer (c(SDS) ) 25 mmol L-1, c(Na2B4O7) ) 5 mmol L-1, c(H3BO3) ) 25 mmol L-1 (pH ) 8.9)). b, etoposide; 2, teniposide; 9, etoposide phosphate; chopper frequency, 30 Hz; voltage, 30 kV; pump laser power, 53 mW. Capillary: 715 (390) mm, 75-µm i.d.

In Figure 3, the signal generated by the lock-in amplifier is plotted for different velocities of the electroosmotic flow resulting from varied voltages applied (constant electroosmotic mobility). It has to be emphasized that in all these measurements the capillary is completely filled with the solution of interest, so that the effects observed cannot be attributed to different velocities of the solute zone passing the detector or to peak-broadening processes dependent on varying resident times of the solute zone in the capillary. In all cases, there is strong increase of the signal with the migration velocity at lower velocity, while at higher velocity the signal approaches a saturation limit. For each measuring point shown in Figure 3, the alignment of the two laser beams has been manually optimized. These results are in accordance with those presented in our previous paper12 and with measurements performed with a similar type of NF-TL detector.24 Obviously, the accelerated heat flow due to the convective heat

Table 1. Limits of Detection Determined under the Conditions of Figure 4

Figure 4. Separation of standards (c(solute) ) 25 mg L-1): (a) UV detection at λ ) 257 nm; (b) thermal lens detection (pump laser power, 105 mW; chopper frequency, 30 Hz) by MEKC (separation buffer: c(SDS) ) 25 mmol L-1, c(Na2B4O7) ) 5 mmol L-1, c(H3BO3) ) 25 mmol L-1 (pH ) 8.9)). Peak assignment: 1, thiourea; 2, etoposide phosphate; 3, etoposide; 4, teniposide. Capillary: 715 (390) mm, 75-µm i.d. Voltage, 30 kV. Injection, vacuum 0.7 s.

transport increases at low flow velocity the periodically induced temperature difference, while at high flow velocity the convective heat transport removes the heated liquid segment out of the probe beam region within the time scale of the formation of the diffracting element. Consequently, we can expect that the thermal lens signal will be substantially decreased at very high electroosmotic velocity.25,26 Both effects described are counteracting, explaining the saturation limit observed. Because of the limited output of the HV generator used, no measurements with higher electroosmotic velocity have been possible. It should be noted that the pluglike electroosmotic flow in first approximation will not distort (merely transport) the formed diffraction element. Consequently, in further experiments, maximum separation voltage was used. Limits of Detection, Linearity of Calibration Function, and Precision. Etoposide phosphate, etoposide, and teniposide can be easily separated from each other under standard MEKC conditions with very short run times (Figure 4a). Photometric detection is possible. However, even at a detection wavelength (24) Bendrysheva, S.; Ragozina, N.; Heissler, S.; Pyell, U.; Faubel, W.; Proskurnin, M. A., unpublished results. (25) Weimer, W. A.; Dovichi, N. J. Appl. Spectrosc. 1985, 39, 1009-1013. (26) Nickolaisen, S. L.; Bialkowski, S. E. Anal. Chem. 1986, 58, 205-220.

compound

TLD (257 nm) (µg L-1)

UVD (257 nm) (µg L-1)

UVD (214 nm) (µg L-1)

etoposide phosphate etoposide teniposide

100 170 120

6000 8300 4000

580 740 710

near the absorbance maximum of these compounds (200 nm) and with a z-shaped detection cell, detection limits ∼1 order of magnitude higher9 than those with laser-induced fluorescence detection excited at 257 nm10 have been obtained. For the method presented in Figure 4, in Table 1 limits of detection (signal is 3σ of the background noise, mean of five consecutive measurements, calculated from 1200 data points of the original data file (baseline) and values for the peak heights) are given for UV detection at the absorbance maximum (214 nm) and at a wavelength of 257 nm. The differences in the LOD reflect the differences in the absorbance coefficients of the solutes detected. To evaluate the performance of NF-TLD for the determination of these solutes having only a low fluorescence quantum yield, the UV detector was replaced with the laboratory-made NF-TL detector. In Figure 4, electropherograms obtained with the same sample (c(solute) ) 25 mg L-1) under identical sample injection and separation conditions recorded by UV detection (λ ) 257 nm) and by NF-TLD (emission wavelength of pump laser, 257 nm) are compared to each other. There is a very large improvement in signal-to-noise ratios. A comparison of the elctropherograms shown in Figure 4 also suggests that there is no extracolumn band broadening induced by replacing the commercial UV detector with the NF-TL detector. The mean number of theoretical plates (five consecutive runs) for the etoposide phosphate peak is 153 000, for the etoposide peak 55 000, and for the teniposide peak 53 000. Additional band broadening would be expected, if the effective detection volume or the signal rise time had been increased. The results are in accordance with results presented in our previous paper. It has to be emphasized that in a crossed-beam configuration a very high spatial resolution can be obtained. The probed volume is defined by the overlap between the formed diffraction element (cylindrical lens) and the probe laser beam. In Table 1, limits of detection (calculated as already described) for the solutes of interest are given for UV detection (214 and 257 nm) and NF-TLD (257 nm). There is a large improvement of the LOD by replacing the UV detector with the NF-TL detector when comparing data obtained at identical wavelength. The improvement for etoposide phosphate is 60-fold, and for etoposide ∼50-fold. The accuracy of the calculated LODs has been verified by preparing samples with a concentration of solute corresponding to the LOD. Figure 5 shows the electropherogram recorded by NF-TLD for etoposide phosphate with a sample containing 100 µg L-1 of this compound. The LODs obtained with NF-TLD for etoposide phosphate and etoposide are sufficient for pharmacokinetic studies. Feisel27 reported a concentration range of etoposide in plasma samples of normal dose patients of 2-100 mg L-1 (observation time, 2 min to 8 h after administration). (27) Feisel, C. Ph.D. Thesis, University of Marburg, Germany, 2001.

Analytical Chemistry, Vol. 76, No. 13, July 1, 2004

3807

Table 2. Repeatability of Migration Times (tm) and Peak Areas Determined from Five Consecutive Runs (Solution of Standards)a compound

tm (min)

RSDb (%)

RSDNc (%)

peak aread

RSD (%)

RSDN (%)

etoposide phosphate etoposide teniposide

2.21 3.29 3.60

1.0 1.2 0.85

0.25 0.16

3.58 3.12 4.78

7.0 2.9 7.0

5.8 2.9

a For experimental conditions, see Figure 4b. b RSD, relative standard deviation. c RSDN, relative standard deviation of magnitudes normalized on the corresponding magnitude for teniposide as internal standard. d In arbitrary units.

Figure 5. Determination of etoposide phosphate (c ) 100 µg L-1) with MEKC-NF-TLD. For remaining experimental parameters, refer to Figure 4b.

Figure 6. Calibration function (thermal lens detection) for (9) teniposide and (b) etoposide phosphate in the range of 0.5-25 mg L-1. For experimental conditions, see Figure 4b.

However, 257 nm does not correspond to the maximum of the absorbance band for the solutes of interest. If the LODs obtained with the NF-TL detector described are compared to the LODs obtained with a commercial UV detector at a wavelength close to the maximum of the absorbance band (214 nm), the improvement is significantly lower, although even under these conditions, the NF-TL detector offers a significant improvement in limits of detections (4-6-fold). The LODs for etoposide phosphate and etoposide obtained with NF-TLD are of the same order of magnitude as those that have been reported for CE with optimized laser-induced fluorescence detection (LIFD) with a powerful frequency-doubled argon ion laser (P ) 200 mW, λex ) 257 nm) as excitation source10 and a laboratory-built noncommercial detection unit: 30 µg L-1 for etoposide phosphate and 100 µg L-1 for etoposide. With the NF-TL detector, calibration functions (six data points, peak area (mean of three consecutive runs) versus concentration of solute in the sample) were measured under the conditions of Figure 4 for the compounds etoposide phosphate and teniposide in the range of 0.5-25 mg L-1. Linear calibration functions have been obtained for the two compounds (see Figure 6). The regression coefficients for the two lines (data points are the mean of three consecutive runs) are 0.9998 and 0.9990. The procedure 3808 Analytical Chemistry, Vol. 76, No. 13, July 1, 2004

according to Mandel (F test) does not show any significant deviation from linearity at a significance level of 95%.28 Under the conditions of Figure 4 employing the NF-TL detector, the repeatability for the migration times and the peak areas has been determined for five consecutive runs. Teniposide was taken as internal standard. The data are given in Table 2. The relative standard deviations for the migration times can be reduced significantly by normalizing to the migration time of the internal standard. However, normalizing to the peak area of the internal standard does not improve the relative standard deviations for the peak areas to the same extent. Obviously, precision of the peak area is not mainly determined by the precision of the injection procedure (in that case, we would expect a large improvement in the relative standard deviation due to normalization of the data on the data of an internal standard). Noise of the thermal lens signal (which is proportional to the signal height24) can be assumed to be mainly determining the peak area precision. Analysis of Spiked Human Blood Plasma. Soetebeer et al.10 emphasized that the main advantage of their method (CE-LIFD) is the capability for the simultaneous quantification of etoposide and etoposide phosphate in plasma samples. After precipitation of proteins by addition of acetonitrile, no coelution and interference of matrix constituents with the solutes of interest was observed. It has to be emphasized that under the separation conditions of Soetebeer et al.10 the run time was ∼14 min. To verify whether the MEKC-NF-TLD method developed in this paper has the same advantages, a blank plasma sample and a spiked plasma sample were pretreated according to ref 10 (procedure 2; see Experimental Section) and analyzed by MEKC-NF-TLD. The overlay of the traces for the spiked and the blank serum (see Figure 7) clearly shows that there is no coelution and interference of matrix constituents with the solutes of interest. The run time needed is very short: less than 4 min. This result is corroborated by determinations of the recovery for the solutes of interest. For a spiked plasma sample containing etoposide phosphate and etoposide at a concentration of 2.13 mg L-1 after sample pretreatment (procedure 1), 2.2 mg L-1 has been determined for etoposide phosphate and 2.5 mg L-1 for etoposide. Calibration was done with standard solutions containing the solute of interest and teniposide as internal standard. The ratio peak area (solute)/peak area (internal standard) was taken for quantification. This normalization takes imprecision of the injected volume and (28) Funk, W.; Damman, V.; Donnevert, G. Qualita ¨tsssicherung in der Analytischen Chemie; Verlag Chemie: Weinheim, Germany, 1997.

consecutive measurements were determined. The data are presented in Table 3. Also in this case, teniposide was taken as internal standard. There is some increase in the normalized relative standard deviations compared to the results obtained with a sample prepared by dissolving pure standard compounds in a solvent mixture (comparison with data in Table 2). However, the data in Table 2 have been obtained with a sample containing a 10-fold higher concentration of solutes. The data in Table 3 confirm that the method developed provides repeatabilities of the quantitative data in the presence of matrix constituents lower than 10% relative standard deviation even at low solute concentration.

Figure 7. Electropherograms obtained for blank plasma (solid trace) compared to a spiked plasma sample (dotted trace) (c(analyte) in injected solution ) 2.5 mg L-1). For remaining experimental parameters, refer to Figure 4b. Table 3. Repeatability of Migration Times (tm) and Peak Areas Determined from Five Consecutive Runs (Spiked Plasma Sample)a compound

tm (min)

RSDb (%)

RSDNc (%)

peak aread

RSD (%)

RSDN (%)

etoposide phosphate etoposide teniposide

2.80 4.33 4.67

0.7 0.8 1.2

0.31 0.69

1.82 1.93 3.78

7.6 7.9 6.4

4.5 7.8

a For experimental conditions, refer to Figure 7. b RSD, relative standard deviation. c RSDN, relative standard deviation of magnitudes normalized on the corresponding magnitude for teniposide as internal standard. d In arbitrary units.

shifts in the electroosmotic velocity due to matrix constituents into account. For the sample described in Figure 7, the repeatability of the migration times and the repeatability of the peak area for five

CONCLUSIONS It has been shown that NF-TLD in combination with MEKC allows for the precise and accurate simultaneous determination of etoposide phosphate and etoposide in human blood plasma with very short run times (4 min). Taking the determination of etoposide phosphate and etoposide as an example, it was shown that for solutes with very low fluorescence quantum yield LIFD is not necessarily the method of choice regarding sensitivity and selectivity. NF-TLD is by nature complementary to LIFD, offers limits of detection 1-2 orders of magnitude lower than those obtained by photometric detection at the wavelength of the pump laser beam and is (for compounds with low fluorescence quantum yield) not influenced by quenching phenomena. ACKNOWLEDGMENT We thank Bristol-Myers Squibb for donation of standards. Financial support from the German Science Foundation and from the Fonds der Chemischen Industrie is gratefully acknowledged.

Received for review December 30, 2003. Accepted April 8, 2004. AC0304222

Analytical Chemistry, Vol. 76, No. 13, July 1, 2004

3809