Anal. Chem. 2000, 72, 61-67

Direct Plasma Sample Injection in Multiple-Component LC-MS-MS Assays for High-Throughput Pharmacokinetic Screening Jing-Tao Wu,* Hang Zeng, Mingxin Qian, Bernice L. Brogdon, and Steve E. Unger

Stine Haskell Research Center, DuPont Pharmaceuticals Company, P.O. Box 30, Newark, Delaware 19714

The simultaneous dosing of numerous compounds followed by multiple-component analysis using LC-MS-MS (the N-in-1 approach) has significantly improved the throughput of the drug-screening process. However, plasma samples still need to be extracted before LC-MSMS analysis, which frequently limits the throughput of the assay. In this work, a high-throughput on-line extraction technique has been developed for multiple-component LC-MS-MS assays using a high-flow column-switching technique. In N-in-1 LC-MS-MS assays, high sensitivity is required since the dose level is generally reduced to minimize drug-drug interactions. In addition, good chromatographic separation is essential to minimize interference and suppression effects. The direct plasma sample injection method developed in this work has successfully met the two requirements for multiple-component LCMS-MS assays in high-throughput pharmacokinetic screening. Plasma samples containing a large number of potential drug candidates were directly injected onto an extraction column operated under a flow rate sufficiently high to exhibit a turbulent-flow profile. The extracted analytes were then eluted onto an analytical column via column switching for LC-MS-MS analysis. The use of turbulent flow resulted in a faster and more rugged extraction with reduced carryover compared with results obtained under laminar-flow conditions. Meanwhile, the use of a column-switching method maintained the chromatographic resolving power and high sensitivity of the LC-MS-MS assay. Separation efficiency, dynamic range, accuracy, and precision comparable with those of solidphase extraction have been achieved with the turbulentflow column-switching technique. As a result, this technique has been successfully and routinely used for highthroughput pharmacokinetic screening. Pharmacokinetic (PK) screening plays an important role in modern drug discovery. In an accelerated discovery process, PK screening is usually performed at a very early stage in the screening strategy. Therefore, limitations in PK screening can largely affect the overall throughput of the drug discovery process. With the development of combinatorial chemistry, the number of compounds that can be synthesized within a short period of time has increased dramatically. To increase the throughput of 10.1021/ac990769y CCC: $19.00 Published on Web 12/04/1999

© 2000 American Chemical Society

PK screening, simultaneous dosing of numerous compounds followed by multiple-component analysis using LC-MS-MS (the N-in-1 approach) has been developed. As a parallel PK screening process to resource parallel synthesis, the N-in-1 approach has proved to be an effective way to improve the throughput of PK screening.1-5 Samples from PK studies are generally in biological fluids such as plasma, serum, and urine. These sample matrixes are not directly compatible with LC-MS analysis since they block LC columns and contaminate the ion source. Therefore, sample pretreatment is required before LC-MS analysis.6 This has been done using protein precipitation, liquid-liquid, or solid-phase extraction. This process is labor intensive and time consuming and can become a bottleneck in the high-throughput PK screening. The automation of extraction procedures helps and has been executed in both serial7,8 and parallel9,10 formats. An alternative approach to automated sample extraction is to use column switching.11-15 In this technique, extraction is performed on either a reusable extraction column or a disposable cartridge. The extracted analytes are subsequently eluted onto a second column for separation. As an on-line extraction technique, it dramatically simplifies or eliminates most of the tedious steps in sample extraction. Although column switching with a reusable (1) Berman, J.; Halm, K.; Adkison, K.; Shaffer, J. J. Med. Chem. 1997, 40, 827829. (2) Olah, T.; McLoughlin, D.; Gilbert, J. Rapid Commun. Mass Spectrom. 1997, 11, 17-23. (3) Bryant, M.; Korfmacher, W.; Wang, S.; Nardo, S.; Nomeir, A.; Lin, C. J. Chromatogr. 1997, 777, 61-66. (4) Hop, C.; Wang, Z,; Chen, Q.; Kwei, G. J. Pharm. Sci. 1998, 87, 901-903. (5) Allen, M. C.; Shah, T. S.; Day, W. W. Pharm. Res. 1998, 15, 93-97. (6) Henion, J.; Brewer, E.; Rule, G. Anal. Chem. 1998, 70, 650A-656A. (7) Simpson, H.; Berthemy, A.; Buhrman, D.; Burton, R.; Newton, J.; Kealy, M.; Wells, D.; Wu, D. Rapid Commun. Mass Spectrom. 1998, 12, 75-82. (8) Janiszewski, J.; Schneider, R. P.; Hoffmaster, K.; Swyden, M.; Wells, D.; Fouda, H. Rapid Commun. Mass Spectrom. 1997, 11, 1033-1037. (9) Allanson, J. P.; Biddlecombe, R. A.; Jones, A. E.; Pleasance, S. Rapid Commun. Mass Spectrom. 1996, 10, 811-816. (10) Cai, J.; Henion, J. D. J. Chromatogr., B. 1997, 691, 357-370. (11) van der Hoeven, R. A. M.; Hofte, A. J. P.; Frenay, M.; Irth, H.; Tjaden, U. R.; van der Greef, J.; Rudolphi, A.; Boos, K.-S.; Varga, G. M.; Edholm, L. E. J. Chromatogr. A 1997, 762, 193-200. (12) Beaudry, F.; LeBlanc. J.; Coutu, M.; Brown, N. Rapid Commun. Mass Spectrom. 1998, 1216-1222. (13) McLoughlin, D. A.; Olah, T. V.; Gilbert, J. D. J. Pharm. Biomed. Anal. 1997, 15, 1893-1901. (14) Needham, S. R.; Cole, M. J.; Fouda, H. G. J. Chromatogr., B 1998, 718, 87-94. (15) Pretorius, V.; Smuts, T. W. Anal. Chem. 1966, 38, 274-281.

Analytical Chemistry, Vol. 72, No. 1, January 1, 2000 61

extraction column is perhaps the most straightforward and economical way to perform on-line solid-phase extraction, it has not been widely used owing to lengthy method development times, pressure buildup, carryover, and its relatively slow speed. Another technique that can be used for direct plasma sample injection is turbulent flow chromatography. Most chromatography systems are operated under laminar-flow conditions. However, if the flow rate is increased to a very high level (normally with a linear flow rate > 5 cm/s), the flow will start to exhibit some turbulent-flow properties. The higher the flow rate, the more turbulent the flow will become. Turbulent-flow chromatography refers to separation systems operated under a turbulent-flow profile. It has been demonstrated that improved separation efficiency has been achieved in an open tubular column operated under turbulent-flow conditions.16 Recently, the use of turbulentflow conditions in a packed column has generated greater interest, particularly for the direct injection of plasma samples.17,18 A high flow rate with only a moderate back-pressure could be achieved in these packed columns due to the use of large packing materials (30-60 µm in diameter). The use of large packing materials also allows the use of large end-column frits (10-40 µm). Consequently, large protein molecules in the plasma sample can easily pass through these columns as waste, without precipitating. This makes turbulent-flow chromatography a fast and rugged extraction technique. However, using large packing materials considerably decreases the separation efficiency. Also, to take the advantage of the reduced band broadening under a turbulent-flow profile, the analytes can only be slightly retained on the stationary phase.18 It is thus difficult to perform a sample-stacking and -focusing step for enhanced sensitivity. In multiple-component LC-MS-MS assays, it is important to maintain sufficient resolving power to avoid potential interference among the compounds and signal suppression in MS detection. The dose level in an N-in-1 study is usually reduced to minimize potential drug-drug interactions. Also, the detection duty cycle is reduced to monitor multiple analytes. The reduced dose level and detection duty cycle demand high sensitivity in LC-MSMS detection. Therefore, a rugged, fast, and automated extraction method that could maintain the high separation efficiency and high sensitivity of the LC-MS-MS assay is of great value for high-throughput N-in-1 PK screening. Although neither columnswitching nor turbulent-flow chromatography alone completely fulfills these requirements, a combination of these two techniques can be used to meet these goals. The use of turbulent-flow conditions in the extraction column solves some of the common problems with conventional column-switching methods, such as speed, pressure buildup, and carryover methods. Meanwhile, the addition of column switching adds high separation efficiency and sensitivity that is usually not sufficient with turbulent-flow chromatography alone. Because of the extremely fast flow rate used for the extraction column, the entire extraction process, which includes loading, washing, flushing, and reequilibrating, can almost always be performed in a time frame shorter than that of the analytical run. Therefore, performing the extraction and (16) Jemal, M.; Qing, Y.; Whigan, D. B. Rapid Commun. Mass Spectrom. 1998, 12, 1389-1399. (17) Ayrton, J.; Dear, G. J.; Leavens, W. J.; Mallett, D. N.; Plumb, R. S. Rapid Commun. Mass Spectrom. 1997, 11, 1953-1958. (18) Gulochon, G.; Martin, M. Anal. Chem. 1982, 54, 1533-1540.

62 Analytical Chemistry, Vol. 72, No. 1, January 1, 2000

Figure 1. Schematic diagram of the turbulent-flow column-switching LC-MS-MS apparatus.

separation in parallel can completely eliminate both the labor and time involved in sample preparation. In this work, a turbulent-flow column-switching system has been set up and on-line sample preparation methods have been developed and optimized to directly inject plasma samples for multiple-component LC-MS-MS analysis. Using a 10-in-1 LCMS-MS assay, the effect of using turbulent flow versus laminar flow on pressure buildup and carryover has been studied. The overall performance of this technique has been compared with that of solid-phase extraction. The effectiveness of this technique was demonstrated in a 14-in-1 PK study in the dog. EXPERIMENTAL SECTION Apparatus. Simplicity is an additional advantage of the turbulent-flow column switching method since it only involves the use of conventional HPLC equipment. A schematic diagram of the instrument setup is shown in Figure 1. Two binary highpressure mixing HPLC pumps were used for this system. An HP1100 pump (Hewlett-Packard, Waldbronn, Germany) was used to deliver a high flow through a reversed-phase extraction column (OASIS extraction column, 1 × 50 mm, 30 µm particle hydrophilic-lipophilic balanced copolymer; Waters Corp., Milford, MA), to load and wash the sample, and subsequently to flush and equilibrate the extraction column. Water (solvent A) and 0.1% methylamine (hydrochloride) and 0.1% ammonium acetate in methanol (solvent B) were used as the solvents for this pump. A Shimadzu 10A pump (Shimadzu, Tokyo) was used to deliver a gradient flow to elute the analytes from the extraction column and to perform the separation on the analytical column. A flow rate of 0.4 mL/min was used for the analytical column with dimensions of 2 × 150 mm. Two types of C18 columns (Symmetry from Waters Corp. and Develosil-MG from Phenomenex, Inc., Torrance, CA) were used as the analytical columns for the 10in-1 and the 14-in-1 assays reported in this work, respectively. Water and acetonitrile both with 0.1% formic acid were used as solvents for this pump (solvents C and D, respectively). A softwarecontrolled electrically actuated high-pressure six-port switching valve was preinstalled in the Shimadzu column oven. A CTC HTS PAL autosampler (LEAP Technologies, Carrboro, NC) was used to inject plasma samples. This autosampler allowed the use of two separate wash solvents. Water with 0.1% formic acid (solvent E) and methanol with 0.1% methylamine and 0.1% ammonium acetate (solvent F) were used as wash solvents for the autosampler, respectively. To avoid column blockage resulting from protein

precipitation by the organic mobile phase, after each injection the aqueous solvent (solvent E) was first used to wash away the plasma residue before washing with an organic solvent (solvent F). The additives in solvents E and F were used to reduce carryover. After the autosampler injected a plasma sample, the HP1100 pump started to deliver solvent A at 4 mL/min to load the sample onto the extraction column and subsequently to clean the sample. The extraction column was directed to waste during this time. The proteins and other matrix components in the plasma sample were removed while the analytes were retained on the extraction column. This loading and washing step was completed after 1 min when the switching valve was switched to the other position. The extraction column was then in the flow path of the Shimadzu pump. The Shimadzu pump immediately started a gradient using solvents C and D to elute the analytes from the extraction column to the analytical column using a flow rate of 0.4 mL/min. This elution step was completed in 2 min when the switching valve was switched back to the original position. The HP1100 pump then started to deliver solvent B at 4 mL/min for 2 min to flush the column. Meanwhile, the analytes were being separated on the analytical column. Finally 1 minute of solvent A delivered by the HP1100 pump at 4 mL/min was used to reequilibrate the extraction column for the next sample. Mass Spectrometer. A Micromass (Manchester, U.K.) Quattro LC triple-quadrupole mass spectrometer with a Z-spray ionization source was used. The positive-ion electrospray ionization mode was used for all studies. The mass spectrometer was operated in the multiple-reaction-monitoring mode with a slightly opened resolution on the first mass analyzer and unit mass resolution on the second analyzer. A dwell time of 100 ms per ion pair was used. The flow from the analytical column at 0.4 mL/ min was directed to the electrospray source without splitting. System Control and Synchronization. A Pentium 200 computer with Masslynx 3.2 (Micromass) software was used to control the mass spectrometer, the HP1100 pump, and the CTC autosampler via a TDAT board, an HPIB board, and RS232 communication, respectively. The Shimadzu pump software controlled the Shimadzu pump and the preinstalled switching valve. The Quattro LC, HP1100, Shimadzu 10A, and the CTC autosampler were also connected by contact closures to synchronize time events. Sample extraction and separation were performed in parallel. After a start signal was sent from Masslynx, the CTC autosampler injected the sample and sent out a contact closure signal to the HP1100 to start its time program. According to the preset values in the time program, the HP 1100 then sent out two contact closure signals to the mass spectrometer and the Shimadzu pump to start the data acquisition and the time program on the Shimadzu pump, respectively. The switching valve information was preset in the Shimadzu pump time program. At a specified time in the HP1100 program, the pump sent out another contact closure signal to the CTC autosampler to inject the next sample. The extraction of the next sample then started while the analytical column was still analyzing the previous sample. Sample Preparation. For direct injection, 200 µL of the plasma sample was added to a 12 mm autosampler vial, followed by 20 mL of internal standard (2 µg/mL in water) and 20 µL of water. The samples were ready for injection after vortexing. For standard and quality control samples, 20 µL of standard spiking

Figure 2. Chemical structures of the 10 compounds and the internal standard used for the evaluation of the turbulent-flow columnswitching system.

solution and 20 µL of internal standard solution were added to a 12 mm autosampler vial containing 200 µL of blank plasma. An injection volume of 100 or 120 µL was used in this work. For solidphase extraction used as a comparison in this work, a 96-well solidphase extraction plate with the same sorbent chemistry as the extraction column was used (OASIS, Waters Corp.). It was preconditioned with 1 mL of methanol followed by 1 mL of water before applying the plasma samples described above. The plate was washed with 1 mL of water and eluted with 1 mL of methanol. The eluent was then evaporated to dryness and reconstituted in 100 µL of 0.1% aqueous formic acid/acetonitrile (95/5). An injection volume of 50 µL was used which gave a sample amount equivalent to that with 120 µL of plasma directly injected. Materials. All commercially available compounds and mobilephase additives were purchased from Sigma (St. Louis, MO). The structures of the commercially available compounds used in this work are listed in Figure 2. Compounds in the 14-in-1 PK study were synthesized at the DuPont Pharmaceuticals Co. (Wilmington, DE). RESULTS AND DISCUSSION Ideally, a good automation tool for high-throughput PK screening should provide fast, rugged, and simple sample preparation without sacrificing the sensitivity, specificity, and accuracy Analytical Chemistry, Vol. 72, No. 1, January 1, 2000

63

Figure 3. Total ion chromatogram and MRM chromatograms of the LC-MS-MS assay for the 10 compounds with 100 µL direct injection of a 250 ng/mL chimpanzee plasma standard. A 2.0 × 150 mm C18 column (Symmetry, Waters Corp.) was used as the analytical column, which was operated in the gradient mode with a flow rate of 0.4 mL/min.

of the LC-MS-MS assay. The turbulent-flow column-switching technique combines the speed and ruggedness of turbulent-flow chromatography with the focusing and resolving power of column switching, which leads to uncompromising sensitivity, specificity, and accuracy of the assay results. In addition, this technique is also of great simplicity and is as reliable as conventional HPLC. To evaluate the performance and value of the turbulent flow column switching technique for multiple-component LC-MSMS analysis, 10 commercially available drug compounds were chosen. The structures of these compounds are shown in Figure 2. Four of these compounds (alprazolam, oxazepam, temazepam, and estazolam) were close structural analogues while the other six compounds had some diversity in structure. The purpose of this pool selection was to mimic the common situations in most N-in-1 PK screenings. Compounds synthesized by the same laboratory usually have close structural similarities while compounds across different laboratories exhibit more diversity. In Figure 3 are shown the total ion chromatogram (TIC) and multiple-reaction-monitoring (MRM) chromatograms of a direct injection of 100 µL of a 250 ng/mL chimp plasma standard of a mixture of the above 10 compounds. A high aqueous back-stacking step was used when the analytes were back-flushed onto the analytical column. This stacking step and the subsequent gradient elution focus the analyte and reduce band broadening resulting from the relatively large injection volume in the column-switching technique. As a result, good separation quality and peak shapes were achieved for the 10 compounds with a total run time of 10 min, including the extraction time. While the column-switching technique provided the separation efficiency and sensitivity required in multiple-component LCMS-MS assays, the use of turbulent-flow on-line extraction greatly improved ruggedness. The extraction column contained 30 µm 64 Analytical Chemistry, Vol. 72, No. 1, January 1, 2000

Figure 4. Comparison of back-pressure buildup as a function of the number of plasma samples injected using turbulent-flow (4 mL/ min) and laminar-flow (0.5 mL/min) column-switching methods. Each injection was 100 µL of blank dog plasma.

i.d. packing material and 10 µm end-column frits, which allowed large protein molecules to easily pass through the column without plugging. The reduced accumulation of protein improved analyte recovery and reduced pressure buildup. In Figure 4, the column back-pressure is shown as a function of the number of plasma samples injected for the same extraction column operated under high-flow conditions (4 mL/min) and under conventional laminarflow conditions (0.5 mL/min). Blank dog plasma with an injection volume of 100 µL was used in this study. Each data point was an average of three separate tests. When the column was operated under laminar-flow conditions, the column back-pressure increased

Table 1. Comparison between Turbulent-Flow Column-Switching and Manual Solid-Phase Extractionsa turbulent flow column switching compound

dynamic range (ng/mL)

% CV

alprazolam carbamazepine estazolam fenfluramine haloperidol oxazepam phentolamine puromycin temazepam triprolidine

1-1000 1-1000 1-2500 1-1000 1-2500 1-2500 1-2500 1-2500 1-2500 1-2500

average a

manual solid-phase extraction mean %

dynamic range (ng/mL)

% CV

mean %

8.9 3.1 12.4 3.0 4.4 14.1 5.6 3.4 9.5 12.6

-8.2 -3.1 6.8 4.1 12.9 3.2 -4.9 4.2 6.5 -12.8

1-1000 1-1000 1-2500 1-2500 1-2500 1-1000 1-2500 2.5-2500 1-2500 1-2500

6.2 13.7 3.2 8.4 7.0 11.2 12.9 4.7 10.1 9.2

4.2 -5.8 5.0 12.8 4.8 -11.6 1.4 4.7 -10.8 6.3

7.7

6.7

8.7

6.7

% CV and mean % are calculated at 50 ng/mL with n ) 3.

quickly with the number of plasma samples injected. After injection of 80 plasma samples, the back-pressure increased about 7-fold. This dramatic pressure increase made the laminar-flow technique unsuitable for batches of >80 samples. In two of the three tests, the column pressure exceeded the preset limit (400 bar) before 120 plasma sample injections were completed. When the column was operated under turbulent-flow conditions, however, there was little change in back-pressure after injecting 120 plasma samples. This effect is most probably due to a unique mechanism associated with a turbulent-flow profile. When a particle enters a packed column, there are numerous pathways that the particle can take. Some of these pathways contain flow channels with smaller diameters while others may be much larger. Under laminar-flow conditions, a particle could enter and plug a channel if the channel has a size smaller than or similar to that of the particle itself. When more and more channels become plugged, the column backpressure continues to increase until the column is fully plugged. Under turbulent flow, a particle that flows into a smaller channel could be back-flushed out by the eddies generated from the turbulent flow. The particle could then take a different pathway and enter a different channel. The process is repeated until the particle finds a larger channel and passes through without causing any pressure buildup. In our laboratory, one extraction column can be routinely used for 200-300 plasma sample injections without causing significant back-pressure increase. An additional advantage of using turbulent-flow extraction is reduced carryover. Carryovers for the above 10 compounds using turbulent-flow extraction, laminar-flow extraction, and solid-phase extraction were measured by injecting blank chimp plasma (or blank extracts for solid-phase extraction) following a 2500 ng/ mL standard. During the extraction column wash step, a 2 min wash at 4 mL/min and a 4 min wash at 0.5 mL/min with the flushing solvent were used in turbulent-flow and laminar-flow extractions, respectively. For all of the 10 compounds, the turbulent-flow extraction showed a reduced carryover compared with that of laminar-flow extraction. This is probably due to the large volume of wash solvent that could be applied within a very short period of time in the turbulent-flow extraction. When the carryovers of turbulent-flow extraction and solid-phase extraction were compared, turbulent-flow extraction still showed a higher carryover for many of the compounds. The average carryover for

laminar-flow, turbulent-flow, and solid-phase extractions were 0.25 ( 0.15%, 0.14 ( 0.07%, and 0.07 ( 0.04%, respectively. The added carryover for either column-switching method may result from the extraction column and the fact that plasma samples were more viscous and more difficult to wash away in the autosampler and transfer tubing. To eliminate a possible bias resulting from elevated carryover for highly absorptive compounds, a reverse sampling sequence was used. In this reverse sampling sequence, predose and late time point samples were assayed first, while samples from early time points or those expected at maximum concentrations were assayed last. Several blank plasma samples were injected between samples from different animals. With this protocol, possible bias from carryover was kept to minimum. While acceptable for unblinded discovery applications, this approach is unacceptable for wide dynamic range GLP studies. The overall performance of the turbulent-flow column-switching technique was compared with that of off-line manual solid-phase extraction as follows. A 400 µL aliquot of the 10-in-1 plasma standard curve and three sets of plasma quality control samples were prepared, and each sample was then split into two equal aliquots. One aliquot was prepared using solid-phase extraction before being analyzed by LC-MS-MS. The other aliquot was directly injected onto the turbulent-flow column-switching system. Injection volumes were adjusted so that equivalent amounts of the analyte were injected. A comparison of the exact recoveries was not possible since the peak shapes and patterns were somewhat different for the two methods. The dynamic ranges, coefficients of variance, and mean differences for the quality control samples are summarized in Table 1, which clearly indicates that the two methods were fully comparable. As demonstrated by the above 10-in-1 assay, the turbulentflow column-switching technique can provide fast and rugged online sample extraction without sacrificing separation efficiency and sensitivity. Consequently, it can be used as a high-throughput tool for multiple-component LC-MS-MS assays in N-in-1 PK screening. Figure 5 illustrates the MRM chromatograms of 14 compounds in a plasma sample collected 1 h after simultaneously infusing 14 compounds intravenously into a dog. The purpose of this 14-in-1 study was to screen and identify the structural analogues with better PK profiles. Therefore, most of the compounds in this pool were close structural analogues, and 8 of Analytical Chemistry, Vol. 72, No. 1, January 1, 2000

65

Figure 5. MRM chromatograms of a plasma sample collected 1 h after intravenous dosing of 14 drug compounds into a dog. The injection volume was 100 µL. A 2 × 150 mm Develosil-MG C18 column (Phenomenex, Torrance, CA) was used as the analytical column.

the 14 compounds showed a major product ion peak at m/z 157. Although there were other product ion peaks that may have been unique for each of these 8 compounds, the intensities of the other ion peaks were lower by at least 1 order of magnitude. Therefore, the base peak at m/z 157 was selected to achieve the sensitivity needed using a reduced dose (0.5 (mg/kg)/compound) in the N-in-1 study. With 8 compounds sharing the same product ion, it was important that the assay have sufficient chromatographic resolving power to avoid possible interference in detection, as was shown in the MRM chromatograms of compounds C3 and C4. Compounds C3 and C4 are close structural analogues that share the same product ion while their parent ions differ by only 2 Da. Because of the isotopic interference and the limited mass resolution on a quadrupole instrument, a false peak was formed in the MRM chromatogram for C3, as indicated by an arrow in the Figure 5. However, the separation efficiency provided by column switching was sufficient to resolve this false peak from the analyte peak of C3 and thus eliminated the interference. The importance of using chromatographic separation to improve assay specificity was also demonstrated in the analysis of C1, where two interfering peaks (marked by asterisks) resulting from endogenous components in the animal were resolved from the analyte peak. Due to its extreme polarity, the peak for C12 showed considerable tailing. It could not be effectively stacked using as weak a mobile phase as 97% aqueous solvent when eluted from the extraction to the analytical column. However, even with such a polar compound, the turbulent-flow conditions could still effectively extract it from the plasma and produce satisfactory results. Two internal standards were selected and used for this assay. The early eluted internal standard (IS 1) was mostly used for 66

Analytical Chemistry, Vol. 72, No. 1, January 1, 2000

compounds that eluted before 7 min, and the late eluted internal standard (IS 2) was mostly used for compounds that eluted after 7 min. The use of two internal standards with different chromatographic retentions better compensated for variations in sample extraction in large-number multiple-component assays. As a result, a dynamic range from 2.5 nM (≈1 ng/mL) to 2500 nM was achieved for most of the compounds in this pool. In Figure 6 are shown the plasma concentration-time profiles of the 14 compounds following an intravenous infusion into a male beagle dog at a dose of 0.5 (mg/kg)/compound . Taking advantage of the high sensitivity provided by the turbulent-flow column-switching LC-MS-MS assay, we measured plasma concentrations up to 24 h. Most of the compounds in this pool had very short half-lives with concentrations decreasing quickly to very low levels. As indicated by the concentration profiles, the screening could effectively differentiate compounds with different PK profiles. Compound 7 was identified as a lead due to its low systematic clearance. The PK parameters generated from the turbulent-flow column-switching multiple-component LC-MSMS assay are summarized in Table 2. This turbulent-flow column-switching method has been successfully and routinely used to perform on-line sample preparation for 100% of the samples in our laboratory. This includes samples in a variety of biological matrixes, such as plasma, serum, urine, bile, synovial fluid, and even whole blood. Although this method is most suitable for N-in-1 multiple-component assays, discrete studies can also take advantage of this technique to achieve rugged and sensitive assays. On the basis of the assay results from a large number of compounds, this turbulent-flow column-switching method is also

Figure 6. Plasma concentration profile for the 14 compounds in the 14-in-1 PK study after intravenous dosing at 0.5 mg/kg into dog. Table 2. Summary of Pharmacokinetic Parameters of the 14 Compounds after Intravenous Infusion into Dogs at 0.5 (mg/kg)/Compounda (Data listed is an average from two animals) compd

t1/2 (h)

Cl ((L/h)/kg)

Vss (L/kg)

% Fe

1 2 3 4 5 6 7 8 9 10 11 12 13 14

1.9 2.4 2.3 3.0 4.4 4.8 3.8 1.3 0.9 5.0 1.0 1.7 1.7 1.0

0.6 0.8 0.5 0.7 0.6 0.6 0.1 0.9 1.2 0.6 0.6 0.3 0.6 0.9

0.5 0.5 0.6 0.6 0.5 0.7 0.2 0.8 0.7 0.5 0.4 0.2 0.4 0.5

2.4 4.2 15.2 2.1 4.0 0.0 95.4 0.6 0.1 0.1 9.5 2.2 0.1 0.4

a Data listed are averages from two dogs. Symbols: t 1/2, half-life; Cl, systematic clearance; Vss, volume of distribution; % Fe, fraction of the parent compound excreted through urine.

applicable to poorly water-soluble and highly protein-bound compounds. For poorly water-soluble compounds, the standards can be prepared in organic solvents such as methanol or a mixture of methanol and water. However, to avoid protein precipitation in the sample, pure acetonitrile should not be used as the solvent for standards and the spiking volume of the standards should be limited to no more than 20% of the plasma volume. This turbulentflow column-switching method works well for most of the highly protein-bound compounds without any modification. For a very

few compounds that show extremely strong protein binding, this method is also applicable after some small modifications during loading or sample preparation. A 0.5 min low-flow (0.5 mL/min) loading step can be used before the high-flow washing step. This will allow more contact time between the analytes and the extraction sorbent. Acidifying the plasma samples in 0.5% formic acid is another effective way to help reduce protein binding. CONCLUSIONS As an on-line sample preparation technique that eliminates labor and time involved in sample extraction, the turbulent-flow column-switching technique successfully combines the speed and ruggedness of turbulent-flow extraction and the sensitivity and separation efficiency of column-switching methods. Its performance is comparable to that of solid-phase extraction in terms of dynamic range, lower limit of quantitation, assay accuracy, and precision. It is a valuable tool for eliminating the sample extraction bottleneck in high-throughput PK screening. As demonstrated in a 14-in-1 PK study, this technique was fully capable of meeting the requirements for multiple-component LC-MS-MS assays and has been successfully applied to large N-in-1 PK studies in our laboratory. ACKNOWLEDGMENT We thank the Medicinal Chemistry group for synthesizing some of the compounds and the Animal Resources group for performing the animal studies reported in this work. Received for review July 13, 1999. Accepted October 22, 1999. AC990769Y

Analytical Chemistry, Vol. 72, No. 1, January 1, 2000

67