Application of Rhodamine 800 for Reversed Phase Liquid

A column heater set at 40 °C was used to maintain the guard and analytical .... The dynamic reserve was calculated using a method described elsewhere...
0 downloads 0 Views 182KB Size
Anal. Chem. 1996, 68, 3763-3768

Application of Rhodamine 800 for Reversed Phase Liquid Chromatographic Detection Using Visible Diode Laser-Induced Fluorescence Sadayappan V. Rahavendran and H. Thomas Karnes*

Department of Pharmacy and Pharmaceutics, Medical College of Virginia, Richmond, Virginia 23298-0533

This study reports the application of rhodamine 800, a far-red dye, suitable for excitation using visible diode laser-induced fluorescence (VDLIF) detection. A reagent synthesized from rhodamine 800 was evaluated as a precolumn reagent for derivatization with amino-containing analytes. The derivative of this reagent with primary amine analytes showed a loss of fluorescence. Rhodamine 800 was then applied as a mobile phase additive in the indirect mode for quantitation of valproic acid in plasma using reversed phase HPLC in combination with VDLIF detection. A visible diode laser (output power 8.50 mW) temperature-tuned to oscillate at 674.70 nm was used as a light source for a laboratory constructed HPLC fluorescence detector. A liquid/liquid extraction procedure was applied to human blank plasma. The selectivity of this method was validated by demonstration of a lack of interfering peaks in extracts of plasma (n ) 3 sources). A calibration curve for valproic acid between 40 and 200 µg/mL was shown to be linear (r ) 0.9932). The recoveries of valproic acid at concentrations of 50 and 100 µg/mL were evaluated and determined to be 73 and 72%, respectively. The precision and accuracy (n ) 5) of the assay was within 7.0% RSD and 8.0% difference from the spiked concentration, respectively. The limits of detection (S/N ) 3) for extracted and unextracted valproic acid were 15.0 and 11.54 µg/mL, respectively. The theoretical (Clim) and practical (Cdet) limits of detection in the detector flow cell for unextracted valproic acid at a S/N ) 1 were found to be within 15%. High-performance liquid chromatography (HPLC) in combination with visible diode laser-induced fluorescence detection (VDLIF) has generated a great deal of interest among researchers involved in the measurement of low levels of analytes in biological matrices. This has resulted in the publication of a number of review articles.1-5 The spectral region of excitation and fluorescence and the type of excitation source used are responsible for the superior sensitivity and selectivity of the technique. Selectivity is enhanced because biological matrices demonstrate minimal blank fluorescence in the far-red regions (>620 nm) of the spectrum. Problems associated with sample degradation are also reduced considerably at long wavelengths and the intensity of (1) Rahavendran, S. V.; Karnes H. T. Pharm. Res. 1993, 10 (3), 328-334. (2) Imasaka, T.; Ishibashi, N. Am. Biotechnol. Lab. 1988, 6, 34-35. (3) Mank, A. J. G.; Lingeman, H.; Gooijer, C. Trends. Anal. Chem. 1992, 11 (6), 210-217. (4) Imasaka, T. Anal. Sci. 1993, 9, 329-344. (5) Imasaka, T.; Ishibashi, N. Anal. Chem. 1990, 62, 363A-371A. S0003-2700(96)00548-3 CCC: $12.00

© 1996 American Chemical Society

raman scatter is reduced by a factor of λ-4.1 Diode lasers are excellent light sources because of their long lifetimes (>80 000 h), minimal flicker noise, minimal power consumption, stability, compact size, low cost, and excellent spectral characteristics when compared to gas discharge lasers (e.g., argon ion, helium, cadmium, etc.).4 Outputs of diode lasers are optimum in this wavelength region, and analytical methods may therefore be provided that are instrument rather than matrix limited. Although, HPLC-VDLIF is an attractive technique, derivatization of the analytes with fluorescent reagents in the pre-postcolumn mode is necessary for measurement of most analytes. The applicability of derivatization reagents from the oxazine, thiazine, cyanine, and squarine classes of dyes have been reported.6-8 An alternative approach to chemical derivatization of analytes is the application of indirect fluorescence detection. This technique provides a universal response when a fluorescent moiety (probe) possessing detectable properties and affinity for the column stationary phase is included in the mobile phase system. Analytes that are injected onto the column may provide a detector response owing to their disruption of the equilibrium distribution of the probe between the mobile and the stationary phases.9-20 Lehotay et al. and Kawazumi et al. have reported the analysis of alcohols using diode laser-based indirect fluorescence detection in combination with HPLC.9,20 Fuchigami and Imasaka first reported rhodamine 800 (λex ) 680 nm; λem ) 700 nm;  ) 90 000 M cm-1; Φ ) 10%) and other fluorescent labels possessing reactive functionalities suitable for labeling in the far-red region.21 An analytical calibration curve for rhodamine 800 in ethanol was constructed using a VDLIF fluorometer, and a detection limit of 4 × 10-12 M was reported.8 Imasaka and colleagues reported a detection limit of 1 × 10-7 M for the VDLIF assay of sodium lauryl sulfate by ion-pair solvent (6) Rahavendran, S. V.; Karnes, H. T. J. Pharm. Biomed. Anal., in press. (7) Karnes, H. T.; Rahavendran, S. V.; Gui, M. Proc. SPIE 1995, 2388, 4. (8) Imasaka, T.; Tsukamoto, A.; Ishibashi, N. Anal. Chem. 1989, 61, 22852288. (9) Lehotay, S. J.; Pless, A. M.; Winefordner, J. D. Anal. Sci. 1991, 7, 863871. (10) Berthod, A.; Glick, M.; Winefordner, J. D. J. Chromatogr. 1990, 502, 305315. (11) Takeuchi, T.; Ishii, D. J. Chromatogr. 1987, 396, 149-155. (12) Sun-II Mho.; Yeung, E. S. Anal. Chem. 1985, 57, 2253-2256. (13) Pfeffer, W. D.; Yeung, E. S. J. Chromatogr. 1990, 506, 401-408. (14) Takeuchi, T.;. Murase, K.; Ishii, D. J. Chromatogr. 1988, 445, 139-144. (15) Hackzell, L.; Rydberg, T.; Schill, G. J. Chromatogr. 1983, 282, 179-191. (16) Takeuchi, T.; Ishii, D. J. Chromatogr. 1987, 393, 419-425. (17) Schill, G.; Arvidsson, E. J. Chromatogr. 1989, 492, 299-318. (18) Vigh, G.; Leitold, A. J. Chromatogr. 1984, 312, 345-356. (19) Takeuchi, T.; Yeung, E. S. J. Chromatogr. 1986, 366, 145-152. (20) Kawazumi, H.; Nishimura, H.; Ogawa, T. J. Liq. Chromatogr. 1992, 15, 2233. (21) Fuchigami, T.; Imasaka, T. Anal. Chim. Acta 1993, 282, 209-213.

Analytical Chemistry, Vol. 68, No. 21, November 1, 1996 3763

extraction with rhodamine 800.22 Fuchigami and Imsaka reported labeling of amino acids with 9-cyano-N,N,N′-triethyl-N′-(5′-succinimidyloxycarbonylpentyl)pyronin chloride, which contained a chromophore similar to rhodamine 800 using capillary electrophoresis (CE)-VDLIF and reported attomole detection limits.23 Higashijima et al.24 and Kaneta and Imsaka25 in separate works evaluated the application of rhodamine 800, oxazine 750, and methylene blue as far-red probes suitable for indirect VDLIF detection using CE and micellar electrokinetic chromatography (MEKC). Valproic acid is routinely monitored in clinical laboratories to assess compliance in epileptic patients and to optimize therapy. Plasma therapeutic levels in patients range from 50 to 150 µg/ mL.26 Valproic acid lacks a chromophoric group, and current methods for its analysis include precolumn derivatization-HPLC,27 gas chromatography,28 or immunoassay.29 This paper reports the application of a reagent synthesized from rhodamine 800 (carboxyl derivative) as a precolumn derivatization reagent for amino-containing analytes. In addition, a detailed report of unmodified rhodamine 800 applied as a mobile phase additive in reversed phase (RP)-HPLC for the quantitation of valproic acid in plasma using indirect VDLIF detection has been described. EXPERIMENTAL SECTION Chemicals. Rhodamine 800 (Catalog No. LD 800) was purchased from Exciton Laser Dyes (Dayton, OH). 2-Propylpentanoic acid sodium salt (valproic acid sodium salt), 2-chloro-1methylpyridinium iodide, and anhydrous potassium carbonate were obtained from Sigma Chemical Co. (St. Louis, MO). Sodium phosphate monobasic and ethylene glycol were purchased from Mallinkrodt Inc. (Paris, KY). Phosphoric acid was purchased from Baker Chemical Co. (Phillipsburg, NJ). Methanol and methylene chloride were of HPLC grade and obtained from Baxter, Burdick & Jackson (Muskegon, MI). Triethylamine has been purchased from Pierce (Rockford, IL). Spectral Characterization. Absorbance measurements were performed using either a LKB Ultrospec Model II spectrometer or a Perkin-Elmer lambda 2S spectrometer. Fluorescence spectral measurements were obtained using a Perkin-Elmer Model LS-50 scanning luminescence spectrometer equipped with a pulsed xenon excitation source (Perkin-Elmer Corp., Rockville, MD) and a red-sensitive photomultiplier tube (PMT) Model R 928. Quantities (2.0 mg) of rhodamine 800 and the reagent synthesized from rhodamine 800 (carboxyl derivative) were separately placed in a mortar and ground to a fine powder. A small portion of each sample was then dissolved in methylene chloride and placed separately on sodium chloride plates for IR analysis using a Nicolet Model 5D ZX FT-IR spectrometer. (a) Thin-Layer Chromatographic System and Conditions. Normal phase silica gel GHLF plates (NP-TLC) (10 cm × 20 cm; (22) Imasaka, T.; Nakagawa, H.; Ishibashi, N. Anal. Sci. 1990, 6, 775-776. (23) Fuchigama, T.; Imasaka, T. Anal. Chim. Acta 1994, 291, 183-188. (24) Hijgashijima, T.; Fuchigami, T.; Imasaka, T.; Ishibashi, N. Anal. Chem. 1992, 64, 711-714. (25) Kaneta, T.; Imasaka, T. Anal. Chem. 1995, 67, 829-834. (26) Clark’s Isolation and Identification of Drugs, 2nd ed.; Pharmaceutical Press: London, 1986. (27) Moody, J. P.; Allan, S. M. Clin. Chim. Acta 1983, 127, 263-269. (28) Gupta, R. N.; Eng, F.; Gupta, M. L. Clin. Chem. 1977, 25, 1303-1306. (29) Liu, H.; Montoya, J. L.; Forman, J.; Eggers, C. M.; Barham, C. F.; Delgado, M. Ther. Drug. Monit. 1992, 14 (6), 513-522.

3764

Analytical Chemistry, Vol. 68, No. 21, November 1, 1996

250 µm particle size; Analtech Inc., Newark, DE) with a solvent system consisting of 80% methylene chloride and 20% methanol were used for the determination of retardation factors (rf) of rhodamine 800 and the reagent synthesized from rhodamine 800 (carboxyl derivative), respectively. Visual detection was performed using a Spectroline Model ENF-280C light source (Spectronics Corp., Westbury, NY). (b) Liquid Chromatographic System and Conditions. The chromatographic system for the indirect mode consisted of a Shimadzu Model LC 6A pump (Shimadzu Instruments, Columbia, MD), a Rheodyne Model 7125 manual injector equipped with a 50 or 100 µL sample loop (Cotati, CA), Vydac Model 201TP, C18 RP HPLC column (150 mm × 4.6 mm i.d.; 5 µm particle size), and Vydac Model 201GCC guard column (20 mm × 4.6 mm i.d.; 5 µm particle size) packed with Vydac Model 201TP packing (The Separations Group, Mojave, CA). A column heater set at 40 °C was used to maintain the guard and analytical columns at constant temperature. Absorbance detection was performed using an isco Model V4 HPLC detector (Lincoln, NE) set at 680 nm and equipped with both a deuterium and a tungsten lamp. Indirect visible diode laser-induced fluorescence detection at 700 nm was performed with diode laser excitation at 674.70 nm. The mobile phase was continously degassed with helium. Indirect Visible Diode Laser-Induced Fluorescence Detection. Instrumental Design. The VDLIF detector used for measuring indirect fluorescence in this study was similar to the one reported previously except for the following modifications.6 A visible solid state diode laser, Toshiba Model 9215 S (Melles Griot, Boulder, CO) was tuned to 674.70 nm (output power 8.50 mW). Two band-pass filters placed in series, Melles Griot Models 03FIB 016 (λ ) 700 nm; fwhm ) 80 + 16, -20; peak transmittance >65%) and 03FIV 024 (λ ) 700 nm, fwhm ) 10 ( 2 nm; peak transmittance 50%; Melles Griot), were used to discriminate the fluorescence from the scatter. An Oriel Model 62030 iris diaphragm (providing a pinhole with 3 mm diameter) was used to attenuate the signal before being directed onto a side-on Hamamatsu Model R636 HA red-sensitive PMT (Hamamatsu Corp., Bridgeport, NJ). The PMT was adjusted to 1000 V with an Oriel Model 70705 high-voltage power supply. The PMT was attached to the optical base plate through two Oriel Model 16021 precision translators placed one on top of the other (to provide precise movement of the PMT in both the X and Y directions). The cuvette holder was held with a Fisher Scientific Model 05769-1 castaloy fixed-angle clamp (Fisher Scientific, Pittsburgh, PA), which was attached to the optical base plate with an Oriel Model 12153 magnetic base via an Oriel Model 12320 rod. Precolumn Derivatization. Rhodamine 800 was evaluated for use as a Model reagent for precolumn derivatization with amino-containing analytes, based on its excellent spectral characteristics and the availability of a diode laser oscillating close to its maximum excitation wavelength. Rhodamine 800 as such does not possess a functional group that would be suitable for conjugation to an amine containing analyte. A suitable reagent was therefore synthesized by converting the nitrile functionality on rhodamine 800 (Figure 1) to a carboxylic acid through base hydrolysis.7 The reagent synthesized from rhodamine 800 (carboxyl derivative) was then reacted with propylamine in the presence of 2-chloro-1-methylpyridinium iodide and base (triethylamine or potassium carbonate) for 1 h at room temperature.30

Clim ) Cm/(DR × TR)9

Figure 1. Structure of rhodamine 800.

Indirect VDLIF Detection. (a) Rhodamine 800 Column Loading. This was evaluated by first removing the column from the chromatographic system after equilibration with the mobile phase (50% methanol/50% deionized water pH 2.40; 15 µg/mL rhodamine 800). The column was determined to be equilibrated when injections of the mobile phase provided no measurable response9 and also when the absorbance of the mobile phase at 680 nm was approximately the same before and after the column. Rhodamine 800 was eluted from the column with 100% methanol. The eluent was collected, and its absorbance measured at 680 nm using a Perkin-Elmer Lambda 2S spectrometer. The concentration of rhodamine 800 was determined from a calibration curve of rhodamine 800 concentration vs absorbance in methanol. The molar ratio of rhodamine 800 bound to the C18 stationary was calculated as described elsewhere.10 (b) RP-HPLC Chromatographic Studies. The peak obtained as a result of the injection of valproic acid onto the equilibrated chromatographic system was optimized for maximum resolution and sensitivity through modification of pH, organic modifier concentration, and rhodamine 800 concentration, respectively. (c) pH Optimization. Mobile phases containing 50% methanol/ 50% water (5 mM sodium phosphate monobasic; pH adjusted to 2.47, 5.0, and 6.50) with 15 µg/mL rhodamine 800 as the mobile phase additive were delivered to the column for equilibration. Each of the three mobile phases was used to optimize resolution of the response due to valproic acid. A 50 µL aliquot of valproic acid (50, 100, and 200 µg/mL) prepared in mobile phase was injected onto the column. The response was measured at 680 nm using absorbance detection. Rhodamine 800 Concentration. The sensitivity of the response due to the injection of valproic acid onto the equilibrated column was evaluated by varying the content of rhodamine 800 (between 5 and 5000 ng/mL) in the mobile phase (50% methanol/ 50% deionized water pH 2.40). Calibration curves for valproic acid (12.12-129.30 µg/mL) were constructed at each concentration of rhodamine 800 present in the mobile phase by injection of the solution of valproic acid onto the column. Valproic Acid Solution Preparation. In all the experiments, valproic acid was reconstituted in the mobile phase used for equilibration of the column (i.e., mobile phase containing rhodamine 800). Limits of Detection for Unextracted Valproic Acid. The limit of detection (LOD) for valproic acid was calculated as the concentration that provided a signal 3 times the mean peak-topeak noise (3Sp-p). The peak-to-peak noise was determined across the elution window of the intended peak. In addition, the LOD was calculated as 3Sb/m, where Sb is the standard deviation of blank signal and m is the slope of calibration curve. The theoretical limit of detection (Clim) in the detector flow cell at a S/N ) 1 for unextracted valproic acid was calculated as (30) Saigo, K.; Usui, M.; Kikuchi,K.; Shimada, E.; Mukaiyama, T. Bull. Chem. Soc. Jpn. 1977, 50, 1863-1866.

Cm is defined as the concentration of rhodamine 800 in the mobile phase (1.0 µg/mL); DR (dynamic reserve) is defined as the magnitude of the background signal to noise level. The dynamic reserve was calculated using a method described elsewhere.9 TR (transfer ratio) is described as the number of rhodamine 800 molecules displaced per valproic acid molecule. TR was determined from a peak height and a peak volume.11

TR ) change in [rhodamine 800]/[valproic acid]

The change in rhodamine 800 concentration was determined by the difference of the background signal (µA) due to rhodamine 800 in the mobile phase from the signal (peak height) obtained as a result of the injection of valproic acid. The difference obtained was then divided by the background signal due to rhodamine 800. This result was then multiplied by the concentration of rhodamine 800 added to the mobile phase. The concentration of valproic acid was calculated using the equation described by Takeuchi and Yeung.19

[VPA] ) (8/π)0.5 (C0Vi/Vw)

C0 is the concentration of valproic acid injected onto the column, Vi is the injection volume and Vw is the peak volume. Lehotay and co-workers9 defined the experimental limit of detection (Cdet) in the detector flow cell at a S/N ) 1 as

Cdet ) LOD/(3 × dilution factor)

where LOD is 3Sp-p and the dilution factor was calculated as C0/ [VPA]. Validation of Method for Valproic Acid in Plasma. (a) Linearity. Valproic acid was prepared in deionized water at 2000, 1500, 1000, 750, and 400 µg/mL through serial dilution. A calibration curve of valproic acid in plasma at 40, 75, 100, 150, and 200 µg/mL was therefore constructed by the addition of 20 µL aliquots of each of the valproic acid concentrations to 180 µL aliquots of blank plasma (n ) 3). (b) Extraction Procedure. To a 1.50 mL microcentrifuge tube containing 0.20 mL of blank/spiked plasma, 0.05 mL of 4 M HCl and 1.0 mL of methylene chloride were added. The vial was centrifuged at 20 000 rpm, and triethylamine (0.05 mL) was added to the organic layer following separation from the aqueous layer. The organic layer was dried under nitrogen and reconstituted with 0.20 mL of mobile phase; a 0.10 mL aliquot was injected onto the column. (c) Selectivity. The method was studied by injection of extracts of blank plasma (n ) 3 sources) onto the optimized chromatographic system and observing for interfering peaks over the elution window of the peak due to valproic acid. (d) Recovery. Absolute recovery using a direct comparison method31 was performed at 50 and 100 µg/mL of valproic acid (n ) 3). (31) Karnes, H. T.; March, C. Pharm. Res. 1993, 10, 1420-1426.

Analytical Chemistry, Vol. 68, No. 21, November 1, 1996

3765

% recovery )

extracted response × 100 unextracted response

(e) Precision and Accuracy. The precision and accuracy (n ) 5) of the assay was determined using peak heights at three concentrations (50, 125, and 175 µg/mL) which represented the low, mid, and upper ranges of the calibration curve (40-200 µg/ mL). Precision was calculated as % RSD. Accuracy was calculated as percent difference from the spiked concentration (% DFS).32

% DFS ) [(obsd mean conc - spiked conc)/spiked conc] × 100 RESULTS AND DISCUSSION Precolumn Derivatization. The rf values for rhodamine 800 and the reagent synthesized from rhodamine 800 (carboxyl derivative) were 0.76 and 0.60, respectively. The infrared spectrum of the reagent synthesized from rhodamine 800 (carboxyl derivative) demonstrated peaks at 1700 and at 3300 cm-1, which suggested the presence of a carbonyl stretch and a hydroxyl stretch. The fluorescence spectra of the reagent synthesized from rhodamine 800 (carboxyl derivative) demonstrated a blue shift of 100 nm (λex ) 575 nm; λem ) 593 nm) in excitation and emission relative to that of rhodamine 800 (λex ) 680 nm; λem ) 700 nm). The blue shift in the spectrum on conversion from a cyano to a carboxyl functionality may be due to a decrease in the planarity of the rhodamine 800 molecule.33 The nitrile functionality probably promotes the delocalization of π electrons over the conjugated ring system due to its linear orientation. The conversion to a carboxyl functionality may disrupt the delocalization of π electrons because of the change in hybridization from a sp to a sp2 state. This change in hybridization may have decreased the extent of delocalization of π electrons and thus resulted in the observed blue shift. The blue shift of the carboxyl derivative may also be due to its weaker electron-acceptor properties compared to the nitrile of rhodamine 800. Derivatization of propylamine with the reagent synthesized from rhodamine 800 (carboxyl derivative) resulted in a complete loss of fluorescence for the resultant derivative. Indirect VDLIF Detection. Rhodamine 800 Column Loading. A 76.31 µg (154 nmol) sample of rhodamine 800 was calculated to be adsorbed to the stationary phase of the reversed phase column on equilibration by the mobile phase. The molar ratio of rhodamine 800 bound to the stationary phase was determined to be 3.1 × 10-4 and is similar to that reported elsewhere.10 The carbon loading for this type of stationary phase is reported to be about 9.0%.34 Rhodamine 800 was easily and completely eluted from the column by washing with 100% methanol. Gnanasambandan and Freiser35 and Berthod et al.10 in separate works used methylene blue, a thiazine dye, as the mobile phase additive and reported the presence of the dye in the column effluent even after extended rinsing with a chloroform/ methanol mixture. Therefore, this observation suggests that the choice of the mobile phase additive becomes important if the column is to be reused for other separations. (32) Karnes, H. T.; Shiu, G.; Shah, V. P. Pharm. Res. 1991, 8 (4), 421-426. (33) Ingle, J. D.; Crouch, S. R. Spectrochemical Analysis; Prentice Hall, Inc.: Englewood Cliffs, NJ, 1988; Chapter 12. (34) The separations Group handbook of columns; Mojave, CA, 1995. (35) Gnanasambandan, T.; Freiser, H. Anal. Chem. 1982, 54, 1282-1285.

3766 Analytical Chemistry, Vol. 68, No. 21, November 1, 1996

Figure 2. (a) Mobile phase pH adjusted to 5.0. Peaks due to valproic acid (VPA) can be observed at 3.30 and 4.0 min, respectively. (b) Mobile phase pH adjusted to 6.50. Negative peak due to VPA can be observed at 2.6 min. (c) Mobile phase pH adjusted to 2.47. Positive peak due to VPA can be observed at 9.0 min. Refer to text for details.

RP-HPLC Chromatographic Studies. (a) pH Optimization. Figure 2 indicates the responses due to valproic acid at mobile phase pH’s of 5.0, 6.50, and 2.47 respectively. The presence of coeluting positive and negative peaks at pH 5.0 suggests coelution of a system peak with the analyte (Figure 2a). To test this, rhodamine 800 was injected onto the column (the mobile phase composition did not include rhodamine 800) and a peak was not observed at the retention times of the two coeluting peaks. Separation of the two coeluting peaks was attempted by varying the organic modifier composition from 50% methanol to 45% methanol. This resulted in broadening of both of the coeluting peaks with no improvement in resolution. Valproic acid exhibits a pKa of 4.50.26 At a mobile phase pH of 5.0, valproic acid molecules would exist in both the ionized and unionized states. It appeared that the coeluting peaks were the result of a mixed equilibrium. The transfer mechanism responsible for RPHPLC is often based on a partition mechanism. Therefore, when valproic acid was injected onto the equilibrated column, the unionized valproate molecules would have a greater affinity for the stationary phase as compared to the mobile phase, which would result in partitioning of the rhodamine 800 molecules from the stationary phase into the mobile phase. The increased concentration of rhodamine 800 displaced into the mobile phase results in

Figure 3. Effect of rhodamine 800 concentration on the sensitivity of the response due to injection of VPA. 2, 1000 ng/mL; 9, 500 ng/ mL; 0, 50 ng/mL; 0 with ×, 5000 ng/mL; ×, 5 ng/mL.

a positive peak with a longer retention time than the peak due to ionized valproate molecules. Likewise, when ionized valproate molecules are injected onto the column, there would be a greater affinity of the molecules for the mobile phase as compared to the stationary phase. This would result in the partitioning of rhodamine 800 molecules from the mobile phase into the stationary phase. A negative peak with a shorter retention time would be expected because there would be a lower concentration of rhodamine 800 in this band of the mobile phase compared to that in the surrounding mobile phase. This mechanism may be occurring because valproic acid is 99% ionized at pH 6.50, and the chromatogram shows that the response is a totally negative peak with a shorter retention time(Figure 2b). At a pH of 2.47, valproic acid would exist as 99% un-ionized, a positive peak with a longer retention time would be expected, and this was observed experimentally (Figure 2c). (b) Rhodamine 800 Concentration. Figure 3 shows that as the concentration of rhodamine 800 in the mobile phase was increased from 5.0 ng/mL to 1.0 µg/mL an increase in the sensitivity of the response was observed. However, when the rhodamine 800 concentration was further increased to 5.0 µg/ mL, a decrease was observed. This decrease in sensitivity is due to PMT signal saturation from the background signal of rhodamine 800 in the mobile phase. The sensitivity of the response due to the injection of valproic acid onto the equilibrated column was highest at a rhodamine 800 mobile phase concentration of 1.0 µg/ mL. An optimized mobile phase consisting of 50% methanol/50% deionized water (adjusted to pH 2.40 with phosphoric acid) and containing 1.0 µg/mL rhodamine 800 was used for all subsequent measurements. Limits of Detection for Unextracted Valproic Acid. The LOD at a S/N ) 3 for unextracted valproic acid was determined to be 11.54 µg/mL. The LOD calculated using 3Sb/m was determined to be 1.79 µg/mL. When this concentration of valproic acid was injected onto the chromatographic system, no response was observed. However, 11.54 µg/mL unextracted valproic acid provided a peak with a S/N ) 3. This method of calculating LOD (i.e., 3Sp-p) is therefore conservative and of more practical value. The theoretical (Clim) and experimental LODs (Cdet) in the detector flow cell at a S/N ) 1 for unextracted valproic acid are 0.518 and 0.440 µg/mL, respectively. The difference between Clim and Cdet

Figure 4. Evaluation of selectivity: (I) chromatogram of extracted blank plasma. (II) Chromatogram of extracted spiked plasma. The peak at 10.80 min represents 50 µg/mL VPA in plasma. Table 1. Analytical Figures of Merit precision [VPA] (µg/mL)

mean peak ht (cm)

std dev

175 125 50

3.70 2.50 1.00

0.200 0.167 0.063

accuracy

% RSD

calcd [VPA] (µg/mL)

% DFS

5.405 6.693 6.324

168.232 116.206 51.173

3.867 7.035 2.345

are within 15%. The DR and TR are 2000 and 9.65 × 10-4, respectively. The methods used to test Cm , TR, DR, and the LOD (S/N ) 3) are accurate within experimental error, based on close agreement of the values between Clim and Cdet.9 In the present system, 1036 valproic acid molecules are required to transfer one rhodamine 800 molecule. This is similar to that obtained elsewhere.36 The DR obtained in the present study can be improved by a factor of 10 with the use of a microbore column.20 A problem that was observed with the present system was the drift in the baseline, which may partly be due to the inadequate temperature control of the chromatographic instrumentation. Validation of Method for Valproic Acid in Plasma. The calibration curve constructed for valproic acid extracted from plasma exhibited a correlation coefficient of 0.9932. The extraction procedure demonstrated recoveries of 73 and 72%, respectively, for 50 and 100 µg/mL concentrations of valproic acid added to blank plasma. The selectivity of the method was validated by observation of no interfering peaks in extracts of blank plasma at the elution window for valproic acid (Figure 4). The precision and accuracy of the method based on peak heights was within 7.0% RSD and 8.0% DFS, respectively (Table 1). The LOD (S/N ) 3) for extracted valproic acid was determined to be 11.54 µg/ (36) Ishii, D.; Takeuchi, T. J. Liq. Chromatogr. 1988, 11, 1865-1874.

Analytical Chemistry, Vol. 68, No. 21, November 1, 1996

3767

mL. The indirect method may therefore be advantageous for the routine analysis of valproic acid in clinical samples . CONCLUSIONS The synthesis of a reagent from rhodamine 800 as a precolumn reagent for amine containing analytes was unsuccessful due to the loss of fluorescence upon derivatization. Rhodamine 800 has been found to be suitable as a mobile phase additive for RP-HPLCindirect VDLIF detection of valproic acid in plasma. The laboratory-constructed VDLIF detector was useful for quantitation of valproic acid in the indirect mode. The sensitivitity of the constructed detection system can be improved by incorporating a differential measurement approach.20 This technique involves splitting the diode laser beam into two and focusing both the beams at different positions on the flow cell. The precision and

3768

Analytical Chemistry, Vol. 68, No. 21, November 1, 1996

sensitivity of the analytical measurement is enhanced because the effects of the background signal can be cancelled, which results in the magnification of the analyte response thereby improving DR. The differential measurement approach is performed in combination with high-frequency modulation and lock-in detection. The use of a microbore column would also enhance analyte detectability. The drift in the baseline of the chromatographic system can be reduced substantially by improved temperature control of the chromatographic system. Received for review June 4, 1996. Accepted August 5, 1996.X AC960548H X

Abstract published in Advance ACS Abstracts, September 15, 1996.