Anal. Chem. 2001, 73, 3083-3088
Increasing Bioanalytical Throughput Using pcSFC-MS/MS: 10 Minutes per 96-Well Plate Steven H. Hoke, II,*,† John A. Tomlinson, II,† Renee D. Bolden,‡ Kenneth L. Morand,† J. David Pinkston,‡ and Kenneth R. Wehmeyer†
Health Care Research Center, The Procter & Gamble Company, P.O. Box 8006, Mason, Ohio 45040, and Miami Valley Laboratories, The Procter & Gamble Company, P.O. Box 538707, Cincinnati, Ohio 45253
The utility of packed-column supercritical, subcritical, and enhanced fluidity liquid chromatographies (pcSFC) for high-throughput applications has increased during the past few years. In contrast to traditional reversed-phase liquid chromatography, the addition of a volatile component to the mobile phase, such as CO2, produces a lower mobile-phase viscosity. This allows the use of higher flow rates which can translate into faster analysis times. In addition, the resulting mobile phase is considerably more volatile than the aqueous-based mobile phases that are typically used with LC-MS, allowing the entire effluent to be directed into the MS interface. High-throughput bioanalytical quantitation using pcSFC-MS/MS for pharmacokinetics applications is demonstrated in this report using dextromethorphan as a model compound. Plasma samples were prepared by automated liquid/liquid extraction in the 96-well format prior to pcSFC-MS/MS analysis. Three days of validation data are provided along with study sample data from a patient dosed with commercially available Vicks 44. Using pcSFC and MS/MS, dextromethorphan was quantified in 96-well plates at a rate of ∼10 min/plate with average intraday accuracy of 9% or better. Daily relative standard deviations (RSDs) were less than 10% for the 2.21 and 14.8 ng/mL quality control (QC) samples, while the RSDs were less than 15% at the 0.554 ng/mL QC level. In the health care industry, the measurement of drug levels in biological fluids answers key questions that are asked throughout drug discovery and development. The more rapidly these measurements are obtained, the more quickly drugs progress toward regulatory approval.1-3 To facilitate this process and to effectively utilize capital-intensive tandem mass spectrometry equipment, it is important to minimize analysis times, where possible. With the continual development of techniques and approaches for enhancing LC-MS/MS throughput, bioanalytical analysis times have been decreasing over the past few years. * Corresponding author: (e-mail)
[email protected]. † Health Care Research Center. ‡ Miami Valley Laboratories. (1) Baillie, T. A.; Pearson, P. G. In Mass Spectrometry in Biology and Medicine; Burlingame, A. L., Carr, S. A., Baldwin, M. A., Eds.; Humana Press: Totowa, NJ, 2000. (2) Muck, W. Pharmazie 1999, 54, 639-644. (3) Watt, A. P.; Morrison, D.; Evans, D. C. Drug Discovery Today 2000, 5, 1724. 10.1021/ac0014820 CCC: $20.00 Published on Web 05/15/2001
© 2001 American Chemical Society
Currently, typical LC-MS/MS analysis times for bioanalytical methods are in the range of 1-4 min,4-6 with isocratic analyses being on the shorter end of that range and gradient determinations taking relatively longer. There are also a few reports of very highthroughput LC-MS/MS methods with analysis times of less than 1 min/sample that were achieved using short columns, high flow rates, and/or parallel separations.7-9 Potential limitations for higher throughput, quantitative LCMS/MS analysis using parallel separations and multisprayer interfaces include cross-talk and a loss of data acquisition speed due to the delay that is required between the sampling of each effluent stream. For both high-throughput serial and parallel separations, there are also limitations due to the viscosity and volatility of the LC mobile phases. Typical flow rates of water and methanol or acetonitrile mixtures in the 0.5-1.5 mL/min range produce high back pressure and/or sacrifice MS sensitivity, depending on the diameter of the column, volatility of the mobile phase, viscosity of the mobile phase, and ionization mode. In contrast, the use of very high mobile-phase velocities (>5 mL/ min) is readily possible with the addition of a viscosity lowering agent such as CO2,10-11 making this approach an attractive alternative for increasing the throughput of bioanalyses. The resulting mobile phase also has the advantage of being more volatile than reversed-phase solvent mixtures, and it is therefore more compatible with the mass spectrometer because the interface can effectively dry several milliliters of mobile phase per minute. While CO2 is relatively nonpolar, mixtures of CO2 and polar organic solvents retain the polarity and solvating power of the polar organic until significant levels (40-60%) of CO2 are added.12 The general approach of adding CO2 to achieve a supercritical, (4) Steinborner, S.; Henion, J. Anal. Chem. 1999, 71, 2340-2345. (5) Eichhold, T. H.; Kuhlenbeck, D. L.; Baker, T. R.; Stella, M. E.; Amburgey, J. S.; deLong, M. A.; Hartke, J. R.; Cruze, C. A.; Pierce, S. A.; Wehmeyer, K. R. J. Chromatogr., B: Biomed. Sci. Appl. 2000, 741, 213-220. (6) Romanyshyn, L.; Tiller, P. R.; Hop, C. E. C. A. Rapid Commun. Mass Spectrom. 2000, 14, 1662-1668. (7) Heinig, K.; Henion, J. J. Chromatogr., B: Biomed. Sci. Appl. 1999, 732, 445-458. (8) Hiller, D. L.; Brockman, A. H.; Goulet, L.; Ahmed, S.; Cole, R. O.; Covey, T. Rapid Commun. Mass Spectrom. 2000, 14, 2034-2038. (9) Bayliss, M. K.; Little, D.; Mallett, D. N.; Plumb, R. S. Rapid Commun. Mass Spectrom. 2000, 14, 2039-2045. (10) Anton, K., Berger, C., Eds. Supercritical Fluid Chromatography with Packed Columns, Techniques and Applications; Chromatographic Science Series 75; Marcel Dekker: New York, 1998. (11) Berger, T. Packed Column SFC, RSC Chromatography Monographs; The Royal Society of Chemistry: Cambridge, 1995.
Analytical Chemistry, Vol. 73, No. 13, July 1, 2001 3083
Figure 1. Structures of (A) DEX and (B) d3-DEX.
subcritical, or enhanced fluidity separation requires the same instrumentation and all three will be referred to as “pcSFC” in this report. Although rarely used for bioanalytical analyses, the successful application of pcSFC to increase bioanalytical throughput was recently reported using the chiral determination of ketoprofen as an example.13 In the present paper, the time savings that can be achieved by using a nonchiral, isocratic pcSFC separation and MS/MS detection are investigated by quantifying a basic antitussive, dextromethorphan (DEX, Figure 1A), in human plasma. The reported bioanalytical analysis times for DEX have mirrored that of the overall industry. Early bioanalytical approaches reported for DEX quantitation in biomatrixes include LC-UV,14 LC-flourescence,15 and GC with nitrogen phosphorus detection.16 The typical analysis times of these methods range from 8 to 20 min per sample. However, recently reported methods for DEX quantitation in plasma have largely been LC-MS/MS based and have used liquid/liquid extraction (LLE)17,18 and dilute-and-shoot19 schemes for sample preparation. The LC-MS/MS analysis times for these methods approach 1 min/sample. Presently, pcSFC is utilized for very high throughput quantitation of DEX. By using a 2 × 10 mm cyano (CN) column and a 7.5 mL/min mobile-phase flow rate, the average analysis time is reduced to ∼6 s/sample while providing other desirable bioanalytical attributes such as sensitivity, linearity, accuracy, precision, specificity, and ruggedness. EXPERIMENTAL SECTION Chemicals and Reagents. DEX (Figure 1A) was obtained from the United States Pharmacopeial Convention (Rockville, MD). The stable-isotope-labeled internal standard, [2H3]dextromethorphan (d3-DEX, Figure 1B) was prepared at the Procter & Gamble Health Care Research Center (Mason, OH). Radiolabeled dextromethorphan, [N-methyl-3H]dextromethorphan, (3H-DEX) was obtained from New England Nuclear Life Science Products (Boston, MA). Ethyl ether (reagent grade), methanol (12) Yuan, H.; Olesik, S. V. Anal. Chem. 1998, 70, 1595-1603. (13) Hoke, II, S. H.; Pinkston, J. D.; Bailey, R. E.; Tanguay, S. L.; Eichhold, T. H. Anal. Chem. 2000, 72, 4235-4241. (14) Park, Y. H.; Kullberg, M. P.; Hinsvark, O. N. J. Pharm. Sci. 1984, 73, 2429. (15) East, T.; Dye, D. J. Chromatogr. Biomed. Appl. 1985, 338, 99-112. (16) Pfaff, G.; Briegel, P.; Lamprecht, I. Int. J. Pharm. 1983, 14, 173-189. (17) Eichhold, T. H.; Greenfield, L. J.; Hoke, II, S. H.; Wehmeyer, K. R. J. Mass Spectrom. 1997, 32, 1205-1211. (18) Wehmeyer, K. R.; Hoke, II, S. H.; Eichhold, T. H.; McCauley-Myers, D. L.; Bolden, R. D. J. Chromatogr., B in press. (19) McCauley-Myers, D. L.; Eichhold, T. H.; Bailey, R. E.; Dobrozsi, D. J.; Best, K. J.; Hayes, II, J. W.; Hoke, II, S. H. J. Pharm. Biomed. Anal. 2000, 23, 825-835.
3084 Analytical Chemistry, Vol. 73, No. 13, July 1, 2001
(HPLC grade), and sodium bicarbonate (reagent grade) were purchased from J. T. Baker (Phillipsburg, NJ). Distilled-deionized water was obtained from a Barnstead NanoPure II system (Dubuque, IA). Carbon dioxide for pcSFC was purchased from Air Products and Chemicals, Inc. (Allentown, PA) and blank human plasma was collected from volunteer donors at Procter & Gamble (Mason, OH). Instrumentation for Sample Handling and LLE. A MicroLab AT Plus 2 (Hamilton Co., Reno, NV) was used to perform all liquid transfers for the LLE extraction including manipulation of samples from a single-tube format into the 96-well plate format, preparation of standards, and addition of internal standard to the samples, quality controls (QCs), and standards. Furthermore, the Hamilton Microlab AT Plus 2 was used for the addition of organic solvent and buffer necessary for the extraction procedures, as well as for transferring the ether extraction solution to clean polypropylene tubes. To achieve the best accuracy and precision, the standard and internal standard solutions were dispensed 0.3 mm from the bottom of the polypropylene tubes. The standards, internal standard, and buffer were aspirated using liquid-level sensing with a single prewet mixing step before solution transfer. The transfer of ether was performed using a preset depth without liquid-level sensing. A Tomtec Quadra 96 (Hamden, CT) was used to add solvent for reconstitution of samples after drying. The Tomtec is a 96channel liquid handling system which permits the parallel transfer of 96 samples. Preparation of DEX Plasma Standards. Initial stock solutions of DEX and d3-DEX were prepared in methanol. DEX stock standard solutions and d3-DEX stock internal standard solutions were prepared in 0.1% NaCl/MeOH (50/50; v/v) and stored at -20 °C. Sodium chloride was added to facilitate the use of liquidlevel sensing on the Hamilton MicroLab AT Plus 2. Working plasma standards were prepared daily by first pipetting 20 µL of a 75.8 ng/mL d3-DEX stock solution (1.52 ng) into each of 96 empty 1.2-mL polypropylene tubes (National Scientific Supply Co., Claremont, CA), followed by 20 µL of a 1 M sodium carbonate buffer (pH 10.54). Then, 10 µL of the appropriate DEX standard stock was added to selected tubes to provide DEX standard mass loadings that ranged from 0.0512 to 13.1 ng/tube. Finally, an aliquot (0.2 mL) of blank human plasma was added to each standard tube providing DEX standard concentrations ranging from 0.256 to 65.6 ng/mL. This resulted in a total of nine standards providing a calibration range of 256 times the lowest standard. Preparation of Quality Control Samples. A DEX QC stock solution at 14.8 ng/mL was prepared by adding a small aliquot of the appropriate DEX stock standard solution to a 10-mL volumetric flask and diluting to volume with blank human plasma. Midlevel and low-level QC stock samples were prepared by serial dilution with blank human plasma to yield DEX levels of 2.21 and 0.554 ng/mL. The stock QC plasma samples were stored at -70 °C. On each validation day, working QC samples were prepared in a 96-well plate format using the Hamilton MicroLab AT Plus 2 system. An aliquot (0.2 mL) of a given QC stock solution was pipetted into a corresponding polypropylene tube already containing 20 µL of the 75.8 ng/mL d3-DEX solution and 20 µL of the 1 M sodium carbonate pH 10.54 buffer. On each validation day, replicate (n ) 48) QC samples were prepared at each level.
Figure 2. Schematic diagram of the autosampler configuration used for sequentially diverting flow to each of the eight injector ports.
Preparation of Pharmacokinetic Samples. Plasma samples obtained from a subject dosed with Vicks Formula 44 cough syrup (30 mg of DEX HBr; Procter & Gamble, Cincinnati, OH), were prepared for analysis in a 96-well plate format using the Hamilton MicroLab AT Plus 2 instrument. A 200-µL aliquot of each subject sample was added into separate tubes already containing 20 µL of a 75.8 ng/mL d3-DEX solution and 20 µL of the 1 M sodium carbonate pH 10.54 buffer. Samples were then prepared for analysis by LLE as described below. LLE Extraction. A 600-µL aliquot of ethyl ether was added to plastic tubes containing the standards, QCs, and study samples using the Hamilton Microlab AT Plus 2 liquid-handling system. The tubes contained in the 96-well rack were then covered with a 96-well mat cap (Microliter, Suwanee, GA). An aluminum block was placed on top of the mat cap to prevent cross-contamination, and the entire assembly was placed in a multitube vortex (VWR, South Plainfield, NJ) and clamped into place. DEX was subsequently extracted into the ether layer by mixing with the multitube vortex for 5 min at midlevel power to allow thorough mixing of the samples. Following extraction, the tubes in the 96-well rack were placed in a dry ice/acetone bath to freeze the plasma layer. Freezing the plasma layer simplified the transfer of the ether and greatly reduced the possibility of transferring plasma. Additionally, freezing the plasma layer minimized the potential for crosscontamination between the sample tubes by condensing the ether vapors away from the top of the closely spaced sample tubes. It was important to keep the mat cap cover secured with the aluminum block during freezing in order to prevent the ether pressure from pushing the mat cap out of the tubes. After the plasma layer was frozen, the mat cap was carefully removed and a portion of the ether layer (400 µL) was transferred with the Hamilton Microlab AT Plus 2 to a 96-well rack of clean polypropylene tubes. The samples were evaporated to dryness with nitrogen using a SPE Dry-96 concentrator (Resolution Systems, Inc., Holland, MI). Heat was applied to the samples for 15 min after evaporating to dryness. Finally, the Tomtec Quadra 96 system was used to add 200 µL of methanol to reconstitute the dried extracts. The samples were covered with a mat cap and vortexed at the lowest power setting for 5 min to facilitate
reconstitution while preventing contact of sample solutions with the mat cap. The mat cap was carefully removed and the Tomtec Quadra 96 was used to transfer the samples, in parallel, from the individual tubes into the autosampler vials of a 96-well plate. Following the LLE and reconstitution, the standards, QCs and samples were analyzed by pcSFC-MS/MS. pcSFC-MS/MS Instrumentation. The pcSFC solvent delivery system was composed of a Gilson (Middletown, WI) modular system that included a model 308 control pump, designed to deliver CO2, two model 306 auxiliary pumps for the delivery of conventional organic and aqueous mobile phases, and a model 811C dynamic mixer. The system was configured using a series of two- and three-way valves such that it could be converted from LC to SFC mode and vice versa in 5-10 min. The autosampler used for this work was a Gilson 215 liquidhandling system with a model 889 eight-port injector. Modifications to the autosampler included the replacement of the static splitter/combiner with a dual-headed Valco Cheminert model C5, eight-position valve (Valco Instruments Co. Inc., Houston, TX). The autosampler was used to simultaneously load samples from 8 wells of a 96-well plate into the eight injectors. The full mobilephase flow was then sequentially diverted to each of the eight injectors using one side of the valve. For analysis of a given sample, the flow passed through the appropriate injection loop and was then directed through the other side of the valve which recombined the eight possible flow paths prior to entering the column (Figure 2). The injection volume was 10 µL, and the delay between injections was 5 s. Details of this modified autosampler were recently reported along with its use for the characterization of synthetic organic libraries in the flow injection mode.20 The separation was performed using a Javelin BetaBasic guard column (2 × 10 mm, 5 µm) with a CN stationary phase (Keystone Scientific, Bellefonte, PA). The mobile phase was a mixture of carbon dioxide and methanol (35:65) that was maintained at a flow rate of 7.5 mL/min. The flow and dimensions of the interface transfer line21 were such that the postcolumn pressure was (20) Morand, K. L.; Burt, T. M.; Regg, B. T.; Chester, T. L. Anal. Chem. 2001, 73, 247-252. (21) Baker, T. R.; Pinkston, J. D. J. Am. Soc. Mass Spectrom. 1998, 9, 498-509.
Analytical Chemistry, Vol. 73, No. 13, July 1, 2001
3085
maintained at a level to ensure the CO2/MeOH mixture was one phase throughout the column.22 The mass spectrometer was a PE Sciex API III+ (Thornhill, ON, Canada). For the pcSFC separation, a modified TurboIonSpray source was used, which allows the addition of a makeup flow using a tee junction; however, no postcolumn makeup flow was used for this application, nor was the flow split prior to introduction into the mass spectrometer. The details of this modified source are described elsewhere.21 The TurboProbe temperature and nitrogen gas flow rate were 470 °C and 8 L/min, respectively, and the nebulizer gas pressure was 65 psi (nitrogen). Protonated analyte ions were generated using ESI and orifice potentials of 4000 and 70 V, respectively. Collisional activation was achieved using argon as the collision gas at a thickness of 270 × 1013 molecules/cm2 and a collision energy of 30 eV. The selected reaction monitoring (SRM) transition m/z 272 to 147 was monitored for detection of DEX, while the transition m/z 275 to 150 was monitored for d3-DEX. The dwell time for each transition was relatively short at 75 ms to ensure that enough points were collected across the narrow pcSFC peaks. Quantitation of DEX. Peak areas of the chromatographic peaks were determined using the PE-Sciex software package, MacQuan version 1.5. Calibration curves for DEX were constructed by plotting peak area ratios of DEX/d3-DEX versus DEX concentrations and fitting these data to a 1/x2 linear regression plot. DEX concentrations of samples and QCs were interpolated from this line. Suppression of DEX in Plasma Extracts. The potential suppression of the electrospray signal by components in the extracted plasma sample matrix was determined through comparison of instrumental response to DEX spiked into blank plasma extracts versus DEX spiked into methanol solutions. Blank plasma was extracted using the described LLE procedure. The reconstituted extract (170 µL) was transferred to an autosampler vial already containing 20 µL of d3-DEX internal standard solution producing a concentration of 7.58 ng/mL and 10 µL of a DEX standard producing a 32.8 ng/mL DEX solution. The neat methanol solutions were prepared analogously with the exception that DEX and d3-DEX were spiked into pure methanol rather than a methanol reconstitution of a plasma extract. Samples were prepared in replicate (n ) 4). Suppression caused by the plasma matrix was determined by dividing the peak areas produced for each compound from the analysis of the spiked plasma extracts by the corresponding peak areas obtained from the spiked methanol solutions. Human Pharmacokinetic Study. A subject was orally dosed with 30 mg of DEX HBr from a commercially purchased Vicks Formula 44 cough syrup. Blood samples (10 mL) were obtained at 0, 0.16, 0.33, 0.5, 1, 1.5, 4, 8, 12, 16, and 26 h postdose using collection tubes containing sodium heparin as the anticoagulant. The blood was immediately placed on ice and subsequently processed by centrifugation to yield plasma. The resulting plasma samples were then stored in polypropylene cryovials at -70 °C until analysis. RESULTS AND DISCUSSION pcSFC Chromatographic Profiles of DEX and d3-DEX. Using a 2 × 10 mm CN column, the flow rate and methanol (22) Chester, T. L.; Pinkston, J. D. J. Chromatogr., A 1998, 807, 265-273.
3086 Analytical Chemistry, Vol. 73, No. 13, July 1, 2001
Figure 3. Ninety-six sequential injections achieved by using pcSFC-MS/MS for the repeated analysis of d3-DEX in methanol (40 pg on column).
content of the pcSFC mobile phase were optimized so that the analyte retention time was approximately 2-3 times the void time. This allowed rapid throughput while still taking advantage of the separation portion of the analysis to minimize suppression and add selectivity. Back-pressure limitations and the mobile-phase drying rate prevented using faster flow rates for the analysis of DEX. During normal operation, the back pressure was ∼4000 psi, which is within 25% of the practical system limit of 5000 psi imposed by the Valco dual-headed, eight-port valve. Also, with a 7.5 mL/min flow rate and a 35:65 ratio of CO2 to MeOH, mobilephase drying in the mass spectrometer interface was adequate. However, a further significant increase in the flow rate above 7.5 mL/min rendered the mass spectrometer unable to effectively dry the mobile phase, which resulted in a loss of sensitivity. Figure 3 shows the analysis of an entire 96-well plate. This figure was produced by repeated injection of 40 pg of d3-DEX in MeOH. The analysis time for the entire plate was 10 min and 12 s. On the 10-min time scale, it is difficult to observe the resolution and peak shape; therefore, various expansions of the time scale are also shown. The top trace shows two individual peaks separated by the injection-to-injection time of 5 s. For quantitative method validation and application, files were closed and opened after every eight injections. These conditions allowed the collection of data from an entire column (8 wells) of a 96-well plate every 45 s. Much of the autosampler rinsing and loading of the subsequent column of eight samples occurred during data acquisition so that the time between each set of eight samples was reduced to ∼6 s.
Table 1. Average Recoveries of DEX QC Samples for 3 Days of Validation Obtained Using pcSFC-MS/MS (n ) 48 at Each Level)
Figure 4. Chromatograms produced by the analysis of an eightwell column containing calibration standards prepared in human plasma. The top trace represents DEX ranging from 32.8 ng/mL to 0.256 ng/mL. The bottom trace represents the internal standard at 7.58 ng/mL.
Figure 5. Chromatograms of DEX generated by eight replicate injections of (A) plasma blanks, (B) 0.554 ng/mL QCs, (C) 2.21 ng/ mL QCs, and (D) 14.8 ng/mL QC samples. The chromatograms of the internal standard are not shown.
Figure 4 shows the analysis of an eight-well column containing calibration standards ranging from 32.8 to 0.256 ng/mL. Figure 5 shows resolution, peak shape, and signal-to-noise ratio obtained with eight replicate injections of plasma blanks and QC samples at low, medium, and high levels. Batches as large as 440 injections were analyzed using this method; the total analysis time for that number of samples was less than 48 min. The speed advantage is apparent versus the highest throughput LC-MS/MS method reported for DEX analysis in plasma.19 In this previous report, the analysis time is relatively fast for LC-MS/MS at 1.1 min/sample, but it would still require 8.1 h to complete a batch of 440 injections. The throughput advantage of pcSFC-MS/MS provides the ability to immediately assess batch performance and allows data processing
day
DEX concn (ng/mL)
mean recovery (%)
RSD (%)
1
0.554 2.21 14.8
99 95 93
11.7 9.0 7.5
2
0.554 2.21 14.8
99 98 98
14.4 9.4 5.7
3
0.554 2.21 14.8
92 94 91
14.2 8.0 5.1
and reporting to commence very quickly. No problems with back pressure or loss of sensitivity were noted during analyses. Recovery and Stability. Recovery of DEX from plasma using the liquid/liquid extraction procedure was determined by spiking 3H-DEX at 26 700 dpm into four 200-µL human plasma aliquots. The resulting extracts were counted using standard techniques with a Packard 2550 TR/LL scintillation counter (Meriden, CT). The extraction efficiency was determined to be 68.5% with a 2.0% relative standard deviation (RSD) (n ) 4). Stability of DEX was established in previous studies for up to 3.5 h in whole blood at ambient temperature, 4 h in plasma at ambient temperature, and for three freeze-thaw cycles. In all cases, average DEX recoveries were found to be within 12% of the spiked concentrations, indicating that stability of DEX is not an issue.17 Stability was not further investigated as part of the current study. Specificity and Linearity. Specificity of the method for DEX determination in human plasma was established by showing that six different sources of human plasma did not contain any interference for DEX or d3-DEX. Calibration of DEX was performed using nine standards to cover the concentration range of 0.256-65.6 ng/mL. The lowest calibration standard, 0.256 ng/ mL, was the method lower limit of quantitation (LLOQ). The average recovery of the lowest standard for 3 days of validation was 106% with a 7.1% RSD. Typical recoveries for all calibration standards were within (10% of theoretical, and all correlation coefficients were greater than 0.996. Accuracy and Precision. Accuracy and precision data were obtained from 3 days of spike and recovery experiments in human plasma. Table 1 displays the average daily recovery and the precision (% RSD) at each of the three QC levels. The average recoveries for all QC levels and all 3 days of validation ranged from 91 to 99%. The precision was acceptable at all QC levels and improved with increasing QC concentration. For the low-level QC samples, the RSDs ranged from 11.7 to 14.4%; for the midlevel QCs, the RSDs ranged from 8.0 to 9.4%, and the RSDs for the high-level QCs ranged from 5.1 to 7.5%. As shown in Figures 4 and 5, there was some variability observed in the signal intensity from injection to injection. This was corrected by use of a stablelabeled internal standard so that the accuracy and precision expected of bioanalytical assays were obtained. Matrix Suppression. Suppression was evaluated because of concerns that unretained material might cause signal suppression since DEX and d3-DEX eluted ∼2 s after the void. However, the Analytical Chemistry, Vol. 73, No. 13, July 1, 2001
3087
oral dose of DEX HBr from Vicks 44. This profile shows a Cmax of 8.5 ng/mL at 4 h postdose. The pcSFC-MS/MS methodology was sensitive enough to measure the plasma concentration of the 26-h sample at 1.3 ng/mL. Repeat analysis of the study samples resulted in less than 15% deviation from the original values.
Figure 6. DEX levels in human plasma after dosing with 30 mg of DEX HBr from Vicks 44.
percentage of signal suppression for DEX spiked into plasma extracts at 32.8 ng/mL was only 25.7%. For d3-DEX spiked into plasma extracts at 7.58 ng/mL, the level of suppression was 23.5%. The RSDs of the DEX and d3-DEX peak areas (n ) 4) were e12% for measurements made in both the plasma extracts and the methanol solutions used to calculate suppression. While some suppression was observed, it was insignificant and did not impact bioanalytical quantitation. Human Pharmacokinetic Profile Following Oral Dosing of DEX. The pcSFC-MS/MS method was shown to be useful for analysis of PK study samples. Figure 6 shows a plot of DEX plasma levels versus postdose sampling time following a 30-mg
3088
Analytical Chemistry, Vol. 73, No. 13, July 1, 2001
CONCLUSIONS A pcSFC-MS/MS method was developed and validated for quantitation of DEX in human plasma for concentrations ranging from 0.256 to 65.6 ng/mL. The throughput advantages of using a volatile, low-viscosity mobile phase were demonstrated by combining a multiplexed autosampler, a short 2 × 10 mm column, and a flow rate of 7.5 mL/min to reduce the time for target compound quantitation of DEX in plasma extracts to 10 min and 12 s for a 96-well plate. This represents a significant advance over typical LC-MS/MS methods. Compared with the highest throughput LC-MS/MS method reported for DEX analysis in plasma, a 10-fold increase in throughput is realized as the LC-MS/MS method requires 106 min or 1.8 h for the analysis of a single 96well plate. Increasing the throughput of bioanalyses is critical for the rapid reporting of data that facilitate the drug development process and for the effective utilization of capital-intensive mass spectrometry instrumentation. In addition to increasing throughput, relevant bioanalytical attributes were provided by the pcSFC-MS/MS approach including high part-per-trillion sensitivity as well as good specificity, linearity, accuracy, and precision. Received for review December 15, 2000. Accepted March 19, 2001. AC0014820
