A Four-Column Parallel Chromatography System for Isocratic or

Jan 3, 2001 - A novel approach to parallel liquid chromatography/tandem mass spectrometry (LC/MS/MS) analyses for pharmacokinetic assays and for simil...
3 downloads 7 Views 83KB Size
Download PDF
Anal. Chem. 2001, 73, 582-588

A Four-Column Parallel Chromatography System for Isocratic or Gradient LC/MS Analyses Colleen K. Van Pelt, Thomas N. Corso,* Gary A. Schultz, Stephen Lowes, and Jack Henion

Advanced BioAnalytical Services, Inc., 15 Catherwood Road, Ithaca, New York 14850

A novel approach to parallel liquid chromatography/ tandem mass spectrometry (LC/MS/MS) analyses for pharmacokinetic assays and for similar quantitative applications is presented. Modest modifications render a conventional LC/MS system capable of analyzing samples in parallel. These modifications involve the simple incorporation of three valves and four LC columns into a conventional system composed of one binary LC pumping system, one autosampler, and one mass spectrometer. An increase in sample throughput is achieved by staggering injections onto the four columns, allowing the mass spectrometer to continuously analyze the chromatographic window of interest. Using this approach, the optimized run time is slightly greater than the sum of the widths of the desired peaks. This parallel chromatography unit can operate under both gradient and isocratic LC conditions. To demonstrate the utility of the system, atorvastatin, five of its metabolites, and their deuterated internal standards (IS) were analyzed using gradient elution chromatography conditions. The results from a prestudy assay evaluation (PSAE) tray of standards and quality control (QC) samples from extracted spiked human plasma are presented. The relative standard deviation and the accuracy of the QC samples did not exceed 8.1% and 9.6%, respectively, which is well within the acceptance criteria of the pharmaceutical industry. For this particular analysis, the parallel chromatography system decreased the overall run time from 4.5 to 1.65 min and, therefore, increased the overall throughput by a factor of 2.7 in comparison to a conventional LC/MS/MS analytical method. With the advent of combinatorial chemistry, pharmaceutical companies are producing drug candidates at increasing rates.1-3 Although these parallel synthetic practices are prevalent in the industry, the liquid chromatography/tandem mass spectrometry (LC/MS/MS) method of analysis has remained a serial technique. Consequently, there is a need to eliminate this bottleneck by generating a means to accelerate the analysis rate of pharmacokinetic assays. Often the solution to this problem is to reduce the chromatographic separation times by hastening the chromatog* Correspondencing author. Phone: (607) 257-0183, x17. Fax: (607) 257-0359. E-mail: [email protected]. (1) An, H. Y.; Cummins, L. L.; Griffey, R. H.; Bharadwai, R.; Haly, B. D.; Fraser, A. S.; Wilsonlingardo, L.; Risen, L. M.; Wyatt, J. R.; Cook, P. D. J. Am. Chem. Soc. 1997, 119, 3696-3708. (2) Cargill, J. F.; Lebl, M. Curr. Opin. Chem. Biol. 1997, 1, 67-71. (3) Czarnik, A. W. Curr. Opin. Chem. Biol. 1997, 1, 60-66.

582 Analytical Chemistry, Vol. 73, No. 3, February 1, 2001

raphy. If the chromatography is accelerated to the point where the peaks of interest are no longer resolved from the “suppression front,” which often consists of polar compounds that are not of interest, the precision of the analysis is sacrificed.4,5 Parallel chromatography offers a solution to achieving the needed higher throughput without having to compromise chromatography and, thus, precision. Several different approaches have been taken to achieve parallel chromatography and, thereby, increase throughput. de Biasi et al. proposed a four-channel multiplexed electrospray interface used in combination with four LC columns and a multiple probe injector. This interface rapidly switches between multiple liquid streams, which unfortunately leads to short dwell times and reduced sensitivity.6 This approach may be useful for unknowns for which sensitivity is not an issue, but not for pharmacokinetic studies in which sensitivity is of great importance. Additionally, because this approach requires modifications to the MS ion source, the user may be limited to a specific vendor. Another approach was taken by Korfmacher et al. who used two separate LC pumping systems, autosamplers, and columns connected via a divert valve to a single mass spectrometer. A major disadvantage of this system is that it requires a significant investment in capital equipment, because it needs the HPLC hardware of two conventional systems.7 Zeng et al. have reported the parallel LC/analyst/ prep LCMS. This system operates two analytical or preparative columns in parallel through a valving system and a dual electrospray ionization interface. The design of this dual interface permits the ionization of the analytes in one line to impact the ionization of the analytes in the other. An additional disadvantage of this system is that two isobaric analytes cannot be quantified simultaneously.8 Here we report a novel approach to parallel chromatography which will be beneficial for pharmacokinetic analysis; for assays involving long run times, such as those encountered with chiral separations; and for all other similar quantitative methods. In most conventional chromatograms, the peak elution time is only a small percentage of the total run time, because there is a significant amount of “idle” time before the compound of interest elutes, as (4) Matuszewski, B. K.; Constanzer, M. L.; Chavez-Eng, C. M. Anal. Chem. 1998, 70, 882-889. (5) Fu, I.; Woolf, E. J.; Matuszewski, B. K. J. Pharm. Biomed. Anal. 1998, 18, 347-357. (6) de Biasi, V.; Haskins, N.; Organ, A.; Bateman, R.; Giles, K.; Jarvis, S. Rapid Commun. Mass Spectrom. 1999, 13, 1165-1168. (7) Korfmacher, W. A.; Veals, J.; Dunn-Meynell, K.; Zhang, X.; Tucker, G.; Cox, K. A.; Lin, C. Rapid Commun. Mass Spectrom. 1999, 13, 1991-1998. (8) Zeng, L.; Kassel, D. B. Anal. Chem. 1998, 70, 4380-4388. 10.1021/ac0006876 CCC: $20.00

© 2001 American Chemical Society Published on Web 01/03/2001

Figure 1. Schematic representation of the parallel chromatography system. The system is composed of a single LC pumping system, an autosampler, and a mass spectrometer that are used in combination with three valves and four LC columns.

well as during the re-equilibration period of the column. The general idea behind our approach to parallel chromatography is to use this idle time by staggering injections onto separate columns such that the chromatographic window of interest is always eluting from one of the columns. The effluent from the particular column having the analytes of interest eluting is diverted to the mass spectrometer. The parallel chromatography system reported here is composed of a single LC pumping system, autosampler, and mass spectrometer used in combination with four columns in parallel configuration, and a valving system. The valving system is composed of three identical valves. Two of the valves work in unison to select which one of the four columns receives the injection while simultaneously providing mobile phase to all four of the columns. The third valve simply selects which column is to be in-line with the mass spectrometer, while the effluents from the other three are directed to a waste container. This parallel chromatography system can easily be modified to perform either isocratic or gradient LC methods. In the assay presented, the throughput was increased by a factor of 2.7, as compared to the conventional method, for the analysis of six compounds and their internal standards. EXPERIMENTAL SECTION Reagents. All analytes, atorvastin (ator), p-atorvastin (p-ator), o-atorvastin (o-ator), atorvastin-lactone (ator-lac), p-atorvastinlactone (p-ator-lac), and o-atorvastin-lactone (o-ator-lac), as well as their deuterated internal standards, atorvastin-d5 (ator-d5), p-atorvastin-d5 (p-ator-d5), o-atorvastin-d5 (o-ator-d5), atorvastinlactone-d5 (ator-lac-d5), p-atorvastin-lactone-d5 (p-ator-lac-d5), and o-atorvastin-lactone-d5 (o-ator-lac-d5), were provided by ParkeDavis (Ann Arbor, MI). Acetic acid and acetonitrile were obtained from J. T. Baker (Phillipsburg, NJ) and Burdick & Jackson (Muskegon, MI), respectively. Deionized, 18 MΩ water was produced with a Milli-Q system (Millipore, Bedford, MA). Human control plasma was obtained from Biological Speciality Corporation (Colmar, PA). Safety Considerations. Human plasma poses a safety risk from potential pathogenic contamination. Every batch of plasma purchased from Biological Speciality Corporation tested negative for hepatitis B surface antigen, anti-HIV-1/HIV-2 antibodies, antihepatitis C antibodies, HIV-1 antigen(s), and in a serological test for syphilis. Regardless, human plasma samples were handled as if they were contaminated with pathogens by wearing appropriate

protective clothing, restriction of work area access, and disinfection of exposed surfaces. HPLC Conditions. Figure 1 is a schematic representation of the instrumentation used in this work. The HPLC system that was used included a binary Shimadzu LC-10AD pump system and a Shimadzu SCL-10A pump controller (Shimadzu, Inc., Columbia, MD). Solvent A was 0.1% acetic acid in water, and solvent B was acetonitrile. The LC gradient began at 20% solvent B, and was then ramped over a 0.1 min interval to reach 70% at 0.5 min. Solvent B was held at 70% until 2.5 min, when it was decreased back to the original 20%. The gradient program ended at 6.57 min. The total flow rate from the HPLC pumping system was 748 µL/ min, which was split equally four ways via a five-port manifold (Valco Instrument Co. Inc., Houston, TX), shown in Figure 1. The mobile phase then entered various lengths of 0.020-in. i.d. polyether ether ketone (PEEK) tubing. The length of this mobile phase delay tubing was dependent on the amount of time the gradient needed to be delayed in each particular line before reaching the head of the column. In this study, line 1 had a delay tubing length of 12 in., line 2 had 5 ft, line 3 had 10 ft, and line 4 had 15 ft, which corresponded to delay times of 0, 1.65, 3.30, and 4.95 min, respectively. Valves 1 and 2 were designed in-house and constructed by Valco Instrument Co. Inc. (Houston, TX). The valves were rotated concurrently and were always in the same relative position in order to select which of the four lines, and therefore which column, would be in-line with the autosampler while continuously providing mobile phase to all four lines. This valve design allows for a given line to receive an injection of sample while maintaining the flow and, thus, allows for separations to occur in the other unselected lines which bypass the injector. The autosampler was a Perkin-Elmer Series 200 autosampler (PerkinElmer Corp., Norwalk, CT), and the injection volume was 20 µL. The autosampler injected every 1.65 min, which corresponds to the sum of the chromatographic window of interest and the time for sufficient baseline to be defined to ensure satisfactory integration of analyte peaks. Four identical Genesis, C-18, 4-µm, 2.1 × 50 mm (Jones Chromatography U.S.A., Inc., Lakewood, CO) LC columns manufactured from the same lot of HPLC packing were positioned following valve 2 in the flow path shown in Figure 1. PEEK tees (Upchurch Scientific, Oak Harbor, WA) with dead volumes of less than 4 µL were located after each column and were used to balance line pressures. Two ends of each tee were connected to Analytical Chemistry, Vol. 73, No. 3, February 1, 2001

583

the inlet and the outlet of the flow, and the third opening of the tee was occupied by a PEEK screw. This screw could be adjusted to equilibrate the back pressures of each of the four lines. Typical back pressure for each of the lines was 445 psi. It should be noted that high-pressure tees or other pressure-regulating devices could be placed prior to the columns to decrease the total system dead volume in future parallel chromatography systems. Following the back pressure adjustment tees in Figure 1 is valve 3, also designed in-house and constructed by Valco Instrument Co. Inc. (Houston, TX). This third valve acted independently of valves 1 and 2. Valve 3 diverted the effluent from the line having the chromatographic window of interest eluting to the MS while directing the effluent from the other three lines to waste. Valve 3 could also have a common waste output rather than independent ones for each line. Balancing Back Pressures of the Four Lines. The back pressures of all four lines needed to be adjusted to the same value. This was performed by changing the pump flow rate from 748 to 187 µL/min, which was the flow rate in a single line. Then three of the outlets on the five-port manifold were blocked with PEEK screws, leaving open only the inlet from the pump and the outlet to the single line whose back pressure was to be measured. Valves 1 and 2 were positioned such that the line whose back pressure was to be measured was in-line with the autosampler. Using the pressure reading from the LC pump, the PEEK screw in the back pressure regulator tee of the particular line was adjusted until the appropriate back pressure was reached. This procedure was then repeated for the other three lines. Once all four lines were adjusted to the same value, the entire process was repeated one final time to ensure that the back pressures of the four lines were matched. MS Conditions. The MS instrument used for this work was a Micromass Quattro II (Cheshire, U.K.) equipped with a Z-spray source and operated in the positive-ion electrospray ionization mode. The Z-spray desolvation temperature and capillary voltage were 390 °C and 3500 V, respectively. The collision energy used was 15 V, and the dwell time for each transition was 50 ms. The transitions monitored for the target analytes and the corresponding internal standards were as follows: ator, m/z 559.4 f m/z 440.3; ator-d5, m/z 564.4 f m/z 445.3; p-ator and o-ator, m/z 575.4 f m/z 440.3; p-ator-d5 and o-ator-d5, m/z 580.4 f m/z 445.3; atorlac, m/z 541.4 f m/z 448.3; ator-lac-d5, m/z 546.4 f m/z 453.3; p-ator-lac and o-ator-lac, m/z 557.4 f m/z 448.3, p-ator-lac-d5 and o-ator-lac-d5, m/z 562.5 f m/z 453.3. Signaled Events in a Parallel Chromatography Cycle. The various components of the parallel chromatography system needed to be configured to act as one integrated system. This was accomplished through contact closures. Figure 2 shows the triggered events during a single parallel chromatography cycle. To initiate the first cycle in a run containing many cycles, the autosampler was manually started. Then, as shown in Figure 2, at 0.01 min, the autosampler triggered the pumps to begin their 6.57-min gradient program. At 0.20 min, the autosampler simultaneously turned valve 3 to position 4, and started a 1.53-min MS run. Then, prior to the autosampler making its second injection, valves 1 and 2 were signaled to turn to position 2 by a contact closure provided by the pumps. The autosampler injected every 1.65 min for a total of four times in a single chromatography cycle. As many of these cycles as necessary were repeated, depending on the number of samples. The above explains how the signaling 584 Analytical Chemistry, Vol. 73, No. 3, February 1, 2001

Figure 2. Schematic representation of the events that occur in a single parallel chromatography cycle. The duration of each cycle is 6.6 min, and during this cycle the autosampler injects every 1.65 min, for a total of four times. In the first cycle of a run containing many cycles, the autosampler is manually started. The autosampler initiates the pump run at 0.01 min in the cycle. Then at 0.2 min in the autosampler run, it signals valve 3 to turn to position 4 and the MS to begin its 1.53 min run. Prior to the autosampler making its second injection, the binary pump is responsible for turning valves 1 and 2 to position 2. At this point, the autosampler injects a second time and the same sequence of events occurs again, for a total of four times in the chromatography cycle. The single 6.6-min cycle shown in the figure can be repeated as many times as necessary.

mechanism in this particular analysis was ultimately controlled by the autosampler; depending on the system, the signal could also be directed from the pumps, mass spectrometer, or custom software. Sample Preparation. The samples were prepared by an automated solid-phase extraction (SPE) using a Quadra 96 (Tomtec, Hamden, CT). An Isolute C-18, 100-mg (Jones Chromatography, Lakewood, CO) SPE block was conditioned first with methanol and then with 100 mM ammonium acetate, pH 4.6. Each 100-µL sample of human plasma was mixed with 0.5 mL of 100 mM ammonium acetate, pH 4.6, before being loaded onto the SPE block. The block was washed with 1 mL of 100 mM ammonium acetate, pH 4.6, followed by 20% methanol in water. Air was then aspirated through the SPE block for 2 min to remove residual water. Once a 96-well polypropylene collection block was placed in the vacuum manifold, the samples were eluted with 95% methanol in water. The elution solvent was evaporated and finally reconstituted in 50 µL of 30% acetonitrile with 0.1% acetic acid in water. RESULTS AND DISCUSSION Figure 3A shows the total selected ion chromatogram (TSIC) of the conventional analysis of atorvastatin and its metabolites. The total time between injections was 4.5 min, although all of the analytes eluted in a 1-min interval. Therefore, the majority of the run time in this analysis was spent either waiting for the compounds to elute or waiting for the column to reequilibrate to the initial conditions. This scenario is common for pharmacokinetic LC/MS analyses.9,10 Consequently, the ability to utilize the (9) Cai, J.; Henion, J. Anal. Chem. 1996, 68, 72-78. (10) Mulvana, D.; Jemal, M.; Pulver, S. C. J. Pharm. Biomed. Anal. 2000, 23, 851-866.

Figure 4. The TSIC profiles for 25 parallel chromatography cycles, or 100 injections, of analytical standard in solvent. Several different concentrations of analyte and IS were injected over the course of the 25 cycles, as noted in the figure. Data for all 100 injections were collected in a single MS file.

Figure 3. A total selected ion current (TSIC) chromatogram of the conventional atorvastatin analysis is shown in A. The peaks of interest elute in a 1-min window, but the total time between injections is 4.5 min. In B, the real time is depicted for the four-column analysis cycle described. A pictorial definition of staggered parallel chromatography is shown in C. The mass spectrometer continuously analyzes the chromatographic window of interest as the successive analyte windows elute from each column in progression. Individual TSIC MS files are shown for each of the four columns relative to their elution times. The injected samples were composed of 50 ng/mL of all six analytes and their internal standards dissolved in mobile phase. Despite resolution differences between the columns, the area ratios of the compounds to their IS remains constant.

intervals in a chromatogram before and after the compounds of interest elute would significantly enhance sample throughput. The objective of the parallel chromatography system presented here is to make use of this idle analysis time by staggering injections onto different columns at timed intervals, allowing the mass spectrometer to continuously analyze the chromatographic window of interest. This is pictorially explained in Figure 3C, which shows the elution of the chromatographic window for each of the four columns in a single chromatography cycle. The LC gradient that was used in the conventional analysis was modified slightly to optimize the parallel chromatography conditions. Once the timing for all of the contact closures was in place, the back pressures of all four lines were matched. The utility of the system was first investigated with analytical standards that were not in a plasma matrix, but were rather just dissolved in mobile-phase solvent. Figure 4 is a chromatogram of a single MS run spanning 25 parallel chromatography cycles. The first six cycles, or 24 injections, are of an analytical standard containing 50 ng/mL of all 6 compounds and their corresponding internal standards. After approximately 43 min in the chromatogram shown in Figure 4, two cycles of 100 ng/mL were injected, each followed by a cycle of a blank sample containing no analyte or internal standard. Following the second cycle of blanks at about 77 min in the chromatogram, four injections of 1 ng/mL of all compounds and their internal standards were made. An enlargement of a portion of the chromatogram containing the blank sample following four injections of 100 ng/mL, as compared to the signal from a 1 ng/mL sample, is shown in Figure 5. From Figure 5, it is

Figure 5. Expansion of a portion of the ion current profile shown in Figure 4, indicating the effective carryover of the parallel chromatography system. Four injections of a sample containing neither analyte nor IS were made following four injections of 100 ng/mL. After the cycle of blanks, four injections of 1 ng/mL were made. The chromatogram indicates that there is some carryover associated with the parallel chromatography system which is comparable to the conventional analysis.

apparent that there is some carryover associated with this parallel chromatography system. However, in the conventional analysis, there are also carryover problems, and blanks are run following each injection of 100 ng/mL in order to minimize these effects. Therefore, although carryover exists with this analysis in the parallel chromatography system, it is not any worse than the conventional method of analysis. Figure 6 shows an enlargement of a later portion of the chromatogram in Figure 4, beginning at about 100 min. In Figure 6, it is apparent that each column gives individually reproducible profiles; however, these profiles vary somewhat between the columns. This column-to-column variability may be due to a variety of reasons, including slightly different back pressures on each column; the assorted lengths of delay tubing between the lines, causing varying degrees of gradient mixing so that slightly different gradients enter each column; and finally, there may be some inherent resolving power variability between the columns. Because the four columns were not acting identically, the area ratio of analyte to internal standard was calculated for the 13 cycles, or 52 injections, shown in the second half of the chromatogram in Figure 4 to determine if the area ratio remained constant between the four columns. All injections were for analytical standard samples of 50 ng/mL of the six analytes and internal standards. The results, shown in Table 1, reveal that the relative standard deviation of the area ratio of the six compounds did not exceed 5.2%. Therefore, although the four columns were Analytical Chemistry, Vol. 73, No. 3, February 1, 2001

585

Table 1. Response from Repeated Injection of 50 ng/mL of Analytical Standard area ratio

mean; n ) 52 SD % RSD

ator/ ator-d5

p-ator/ p-ator-d5

o-ator/ o-ator-d5

ator-lac/ ator-lac-d5

p-ator-lac/ p-ator-lac-d5

o-ator-lac/ o-ator-lac-d5

0.973 0.0163 1.7

0.974 0.0249 2.6

0.967 0.0355 3.7

1.02 0.0524 5.1

1.09 0.0222 2.0

1.17 0.0171 1.5

Figure 6. Expansion of a portion of the ion current profile shown in Figure 4. Each group of peaks represents an injection of 50 ng/mL of analytical standard. When comparing groups of peaks adjacent to one another, variability in the shape is observed. However, when comparing every four groupings of peaks, the shapes are quite similar. Therefore, the performance of each column was reproducible. However, there was variability in peak shape and resolution between columns which did not appear to adversely affect the results.

acting slightly differently, the area ratio remained constant among them, demonstrating that the system can be used for quantitative analysis. The parallel chromatography system was then used to analyze a prestudy assay evaluation (PSAE) tray of samples that contained standards and quality control (QC) samples. All of these samples were from extracted human plasma. Here, the MS run duration was set at 1.53 min, which allowed each injection to have an individual MS file, as depicted in Figure 2. The study was designed such that the atorvastatin acids and the atorvastatin lactones were analyzed independently, which resulted in one set of QCs for the acids and one for the lactones. A representative chromatogram of the 8.0 ng/mL QC for the acid family of compounds is shown in Figure 7. The calibration curves of all six compounds had correlation coefficients greater than 0.99. Tables 2 and 3 provide the results from the two standard curves analyzed in the PSAE tray. The percent deviation from the nominal concentrations of the five standards did not exceed 15%. Table 4 is a summary of the QC data from the PSAE tray. The relative standard deviation (RSD) of the two concentrations of QCs analyzed did not exceed 8.1%, and the accuracy did not exceed 9.6%. The pharmaceutical community has dictated that the RSD and accuracy of standard and QC samples cannot exceed 15%.11 Therefore, the described parallel chromatography system, operated under gradient LC conditions, proves to be an acceptable analysis technique for all six analytes. This parallel chromatography system would be particularly useful for analyses requiring long run times, such as in multianalyte determinations as shown here or in those involved with chiral (11) Shah, V. P.; Midha, K. K.; Dighe, S.; McGilveray, I. J.; Skelly, J. P.; Yacobi, A.; Layloff, T.; Viswanathan, C. T.; Cook, C. E.; McDowall, R. D.; Pittman, K. A.; Spector, S. Pharm. Res. 1992, 9, 588-592.

586 Analytical Chemistry, Vol. 73, No. 3, February 1, 2001

Figure 7. The chromatograms of the selected ion transitions in the 8.0 ng/mL QC extracted from human plasma for the acid family of compounds. Internal standards for all six analytes were present in the sample. This MS file contains the data from only a single injection.

separations. Often, biologically active molecules contain chiral centers in which only one of the stereoisomers has activity. This necessitates the chromatographic resolution of the enantiomers. These chromatographic runs can have durations on the order of 30-50 min for such compounds as 2-(S)-[3,5-bis(trifluoromethyl) benzyloxy]-4-[(3-(5-oxo-1H,4H-1,2,4-triazolo)methyl]-3-(S)-phenylmorpholine,12 and trimipramine.13 With other enantiomeric drugs, such as warfarin, shorter analysis times of 10-20 min are achieved,14 but this is still a tremendously long run time in comparison to high throughput methods. Consequently, long chromatographic runs such as those resolving enantiomers would benefit from parallel chromatography. In addition, this parallel chromatography system is not limited to four columns, because valves to accommodate a larger number of columns could be purchased, which would further enhance the throughput of very long chromatographic runs. Of course, the power of the method would be compromised if the chromatographic window of interest were distributed over a large portion of the run. A standard, single-column LC/MS assay could quickly be adapted to this parallel chromatography system by considering the total run time of the single-column assay, the duration of the chromatographic window of interest, and the position of this window within the total run time. Once these three items are considered, then the timing of the gradient reaching the various columns and the turning of the valves can be calculated. The chromatography may be varied slightly in order to optimize the (12) Zagrobelny, J.; Matuszewski, B. K. Enantiomer 1997, 2, 37-43. (13) Liu, J.; Stewart, J. T. J. Chromatogr. B 1997, 700, 175-182. (14) Takahashi, H.; Kashima, T.; Kimura, S.; Muramoto, N.; Nakahata, H.; Kubo, S.; Shimoyama, Y.; Kajiwara, M.; Echizen, H. J. Chromatogr. B 1997, 701, 71-80.

Table 2. Results from PSAE Tray of Extracted, Spiked Human Plasma Standards ator

p-ator

o-ator

std no.

nominal concn (ng/mL)

measured concn (ng/mL)

% dev from nominal concn

measured concn (ng/mL)

% dev from nominal concn

measured concn (ng/mL)

% dev from nominal concn

1 1 2 2 3 3 4 4 5 5

5 5 10 10 25 25 50 50 100 100

5.16 5.22 10.3 9.31 26.8 24.5 47.4 49.5 104 97.3

3.2 4.3 3.1 -6.9 7.3 -1.9 -5.2 -1 4.0 -2.8

5.22 4.96 10.2 9.3 24.3 25.8 47.1 53.7 100 99.4

4.4 -0.8 1.8 -7.0 -3.0 3.2 -5.7 7.4 0.2 -0.6

4.54 4.61 11.5 9.23 26.5 25.4 43.6 55.3 92.3 108

-9.2 -7.7 15 -7.7 6.1 1.4 -12.8 11 -7.7 7.8

Table 3. Results from PSAE Tray of Extracted, Spiked Human Plasma Standards ato-lac

p-ator-lac

o-ato-lac

std no.

nominal concn (ng/mL)

measured concn (ng/mL)

% dev from nominal concn

measured concn (ng/mL)

% dev from nominal concn

measured concn (ng/mL)

% dev from nominal concn

1 1 2 2 3 3 4 4 5 5

5 5 10 10 25 25 50 50 100 100

4.59 4.95 9.71 10.3 25.3 26.9 48.6 50.1 93.9 106

-8.1 -1.1 -2.9 2.9 1.1 7.6 -2.8 0.2 -6.1 6.3

5.11 5.17 9.52 9.98 25.2 25.2 48.5 48.9 99.6 102

2.1 3.3 -4.8 -0.2 0.6 0.6 -2.9 -2.3 -0.4 1.7

5.18 4.96 9.87 9.97 26.4 24.7 49.6 48.4 99.4 101

3.5 -0.7 -1.3 -0.3 5.7 -1.4 -0.9 -3.2 -0.6 1.4

Table 4. Results from PSAE Tray of Extracted, Spiked Human Plasma QC Samples QC 1 8.0 ng/mL

QC 2 80 ng/mL

QC 3 8.0 ng/mL

QC 4 80 ng/mL

ator

mean SD % RSD % accuracy

7.75 0.141 1.8 -3.1

73.4 1.76 2.4 -8.3

ator-lac

mean SD % RSD % accuracy

7.79 0.627 8.1 -2.7

78.5 3.74 4.8 -1.9

p-ator

mean SD % RSD % accuracy

7.50 0.348 4.6 -6.3

74.1 1.39 1.9 -7.3

p-ator-lac

mean SD % RSD % accuracy

7.83 0.265 3.4 -2.1

77.9 4.78 6.1 -2.6

o-ator

mean SD % RSD % accuracy

7.23 0.222 3.1 -9.6

77.3 5.35 6.9 -3.4

o-ator-lac

mean SD % RSD % accuracy

7.53 0.423 5.6 -5.9

77.1 4.64 6.0 -3.7

number of elution windows that can fit into a chromatography cycle. In addition, the integration of the binary pump, autosampler, valves, and MS needs to be performed in order for the unit to work in a unified manner. The entire development of a parallel chromatography method from a single-column method should take an experienced user no more than a day. CONCLUSIONS Parallel chromatography offers an aid to the LC/MS analysis bottleneck brought about by the nascent combinatorial synthetic techniques. To keep pace with the parallel synthesis of drug candidates, techniques employing parallel analysis need to come to the forefront. In this report, we present a means of performing parallel LC/MS analysis with the addition of only three valves and three columns to a conventional LC/MS system. This system

offers considerable versatility, because valves that can accommodate various numbers of columns can be purchased, and the system is not limited to MS detection. This is a reasonably modest and cost-effective manipulation that can significantly enhance throughput. In addition, this system does not confine the user to a particular vendor’s autosampler, pumps, MS, or electrospray ionization interface. Furthermore, chromatographic separation does not need to be sacrificed in order to achieve higher throughput using this system, and the valving system can be operated under either gradient or isocratic LC conditions. For isocratic conditions, only one LC pump is needed, and no delay volume is required between the pump and the multiple columns. Although the long-term performance of the system has not been tested, its robustness should be comparable to conventional systems, as it is constructed entirely of standard liquid chromaAnalytical Chemistry, Vol. 73, No. 3, February 1, 2001

587

tography components. In the relatively difficult assay presented here, in which 12 compounds were analyzed with a gradient LC method, the throughput was increased by a factor of 2.7, as compared to the conventional analysis. Even higher sample throughput could be achieved, depending on the analysis, with a theoretical run time that is the sum of the widths of the peaks of interest. In practice, however, the run time is slightly longer than this to accommodate valve switching and recording of sufficient baseline to accurately integrate peaks. The system proves to be flexible and easy to operate, because a conventional one-column LC method can quickly be converted to the multiple-column parallel system with minor alterations. Our approach to parallel chromatography has great potential in the LC/MS community,

588

Analytical Chemistry, Vol. 73, No. 3, February 1, 2001

especially for assays with long analysis times, such as those required for chiral separations. ACKNOWLEDGMENT The authors thank Parke-Davis for providing the compounds studied. We would also like to acknowledge Ted Green for providing the conventional LC data of atorvastatin and its metabolites, Kim Wheeler for aiding in the preparation of the PSAE tray, and Dan Mulvana for helpful discussions. Received for review June 8, 2000. Accepted November 16, 2000. AC0006876