Evaluation of a Four-Channel Multiplexed Electrospray Triple

Anal. Chem. 2001, 73, 1740-1747

Evaluation of a Four-Channel Multiplexed Electrospray Triple Quadrupole Mass Spectrometer for the Simultaneous Validation of LC/MS/MS Methods in Four Different Preclinical Matrixes Liyu Yang,† Thierry D. Mann,‡ David Little,§ Ning Wu,† Robert P. Clement,† and Patrick J. Rudewicz*,†

Schering-Plough Research Institute, 2015 Galloping Hill Road, Kenilworth, New Jersey 07033-1300, Micromass, Inc., 100 Cummings Center, Beverly, Massachusetts 01915-6101, and Micromass, UK Limited, Floats Road, Wythenshawe Manchester, M23 9LZ, UK

A four-channel multiplexed electrospray interface on a triple quadrupole mass spectrometer was evaluated for the simultaneous validation of LC/MS/MS methods for the quantitation of loratadine and its metabolite, descarboethoxyloratadine, in four different biological matrixes. The assays were performed in rat, rabbit, mouse, and dog plasma from 1 to 1000 ng/mL using 96-well solid-phase extraction for sample preparation. The limit of quantitation of 1 ng/mL corresponded to 5.56 pg of each analyte injected on-column. For the drug, quality control samples (n ) 6 at four concentrations) had precision ranging from 0.967 to 16.0% and accuracy ranging from -8.44 to 10.5% across all four species. For the metabolite, the precision ranged from 0.684 to 11.0% and the accuracy was between 6.36 and -9.06%. Intersprayer cross talk for the multiplexed electrospray ion source was evaluated as a function of analyte concentration and was less than 0.08% at concentrations as high as 1000 ng/mL. These results demonstrate the feasibility of using parallel analysis to reduce the time required for method validation and to increase sample throughput in drug development studies. Quantitative bioanalysis plays a major role in the drug development process. Once an appropriate compound is selected from drug discovery, quantitative methods need to be developed and validated to determine the concentration of the drug and, if necessary, metabolites in biological matrixes. These methods are used to support several activities in drug development, including formulations research, GLP toxicology, clinical pharmacology, and clinical research studies. Atmospheric pressure ionization (API) liquid chromatography/tandem mass spectrometry (LC/MS/MS) is the most widely used analytical technique for this purpose. Most assays are done in the multiple reaction monitoring (MRM) mode * Corresponding author: (phone) 908-740-6513; (fax) 908 740-4474; (E-mail) [email protected]. † Schering-Plough Research Institute. ‡ Micromass, Inc. § Micromass, UK Limited.

1740 Analytical Chemistry, Vol. 73, No. 8, April 15, 2001

using either a structural analogue or a stable isotope-labeled analogue as an internal standard. In an effort to reduce the time required to get a drug to market, bioanalytical laboratories in the pharmaceutical industry continually search for new sample preparation and LC/MS/MS techniques to increase sample throughput. One successful approach has been off-line automated sample preparation in the 96-well plate format.1-4 Several methodologies have been used with 96-well plate sample preparation, including protein precipitation,5 liquid-liquid extraction,6 and the use of solid-phase extraction plates.7 Automated 96-well sample preparation in the batch mode has also been combined with fast chromatography. This has been used for the quantitation of benzodiazepines in human urine for the analysis of 1000 samples in 12 h.8 In addition, the quantitative determination of five estrogen receptor modulators in human plasma with a total chromatographic run time of less than 30 s has been described.9 On-line sample processing has also been used for highthroughput LC/MS/MS analysis. This includes traditional columnswitching techniques where one reusable column is used for sample extraction followed by elution onto a second column for separation of the analytes.10-12 Another design is the Prospekt system that utilizes disposable columns for the on-line extraction (1) Henion, J.; Brewer, E.; Rule, G. Anal. Chem. 1998, 70, 650A-656A. (2) Kaye, B.; Herron, W. J.; Macrae, P. V.; Robinson, S.; Stopher, R.; Venn, F.; Wild, W. Anal. Chem. 1996, 68, 1658-1660. (3) Allanson, J. P.; Biddlecombe, R. A.; Jones, A. E.; Pleasance, S. Rapid Commun. Mass Spectrom. 1996, 10, 811-816. (4) Simpson, H.; Berthemy, A.; Buhrman, D.; Burton, R.; Newton, J.; Kealy, M.; Wells, D.; Wu, D. Rapid Commun. Mass Spectrom. 1998, 12, 75-82. (5) Watt, A. P.; Morrison, D.; Locker, K. L.; Evans, D. C. Anal. Chem. 2000, 72, 979-984. (6) Ramos, L.; Bakhtiar, R.; Tse, F. L. S. Rapid Commun. Mass Spectrom. 2000, 14, 740-745. (7) Peng, S. X.; King, S. L.; Bornes, D. M.; Foltz, D. J.; Baker, T. R.; Natchus, M. G. Anal. Chem. 2000, 72, 1913-1917. (8) Zweigenbaum, J.; Henion, K.; Steinborner, S.; Wachs, T.; Henion, J. Anal. Chem. 1999, 71, 2294-2300. (9) Zweigenbaum, J.; Henion, J. Anal. Chem. 2000, 72, 2446-2454. (10) McLoughlin, D. A.; Olah, T. V.; Gilbert, J. D. J. Pharm. Biomed. Anal. 1997, 15, 1893-1901. (11) Needham, S. R.; Cole, M. J.; Fouda, H. G. J. Chromatogr., B 1998, 718, 87-94. 10.1021/ac0012694 CCC: $20.00

© 2001 American Chemical Society Published on Web 03/15/2001

of samples.13 Turbulent flow chromatography combined with column switching has also been used for the analysis of multiple analytes in drug discovery studies.14 In our laboratory, we have explored the use of a four-channel electrospray ion source on a triple quadrupole mass spectrometer for parallel analysis.15 With this ion source, the effluent from four HPLC columns is introduced into the mass spectrometer simultaneously. A four-channel multiplexed electrospray ion source interfaced to a time-of-flight mass spectrometer for the qualitative analysis of a mixture of four test compounds has been reported.16 We have investigated this technology to increase the speed of bioanalytical method validation in support of preclinical drug development studies. In the preclinical area, bioanalytical methods are used to support toxicokinetic studies to evaluate systemic exposure of the drug and metabolites and to help correlate exposure to any target organ toxicity.17 It is required that these toxicology studies are carried out according to the principals of good laboratory practices (GLPs). Hence, the bioanalytical methods used to support such studies need to be fully validated according to the latest U.S. Food and Drug Administration (FDA) and pharmaceutical industry guidelines.18 To provide bioanalytical support for a preclinical toxicology program, we perform a complete LC/MS/MS method validation using plasma from each toxicology species. For example, a validation would be done using rat, mouse, dog, and rabbit plasma. With one electrospray LC/MS system and a traditional single sprayer ion source, each run of a validation is often done on successive days. In this paper, the evaluation of a four-channel electrospray ion source for the simultaneous validation of bioanalytical methods in four preclinical species is described. The advantages as well as the disadvantages of this parallel approach for increasing throughput are also discussed. This evaluation was done using loratadine, a long-acting tricyclic antihistamine with selective peripheral histamine H1receptor antagonistic activity. To support clinical studies, an LC/ MS/MS method with 96-well plate solid-phase extraction was developed and validated for both loratadine and its metabolite, descarboethoxyloratadine.19 This assay had a quantitative range of 0.025-10 ng/mL for both the drug and the metabolite. As part of our efforts to evaluate the multiplexed electrospray (MUX) interface for preclinical support, we changed the concentration range of the loratadine assay to 1-1000 ng/mL, this level being (12) 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. (13) Beaudry, F.; Le Blanc, J. C. Yves; Coutu, M.; Brown, N. K. Rapid Commun. Mass Spectrom. 1998, 12, 1216-1222. (14) Wu, Jing-Tao; Zeng, H.; Qian, Mingxin; Brogdon, B. L.; Unger, S. E. Anal. Chem. 2000, 72, 61-67. (15) Yang, Liyu; Wu, Ning; Clement, R.; Rudewicz, P. Paper presented at the 48th American Society for Mass Spectrometry and Allied Topics, Long Beach, CA, June 11-15, 2000. (16) de Biasi, W.; Haskins, N.; Organ, A.; Bateman, R.; Giles, K.; Jarvis, S. Rapid Commun. Mass Spectrom. 1999, 13, 1165-1168. (17) Cayen, M. N. Toxicol. Pathol. 1995, 23, 148-157. (18) 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. (19) Yang, Liyu; Clement, R.; Beaudry, F.; Grandmaison, C.; Di Donato, L.; Masse, R.; Rudewicz, P. Paper presented at the 48th American Society for Mass Spectrometry and Allied Topics, Long Beach, CA, June 11-15, 2000.

Figure 1. Chemical structures of (A) SCH 29851 (loratadine); (B) SCH 34117 (descarboethoxyloratadine), and (C, D) their respective isotopically labeled internal standards.

more representative of what is often required to support toxicokinetic studies for new chemical entities. EXPERIMENTAL SECTION Materials and Reagents. The analytes, SCH 29851 (loratadine) and SCH 34117 (descarboethoxyloratadine), and their isotopically labeled internal standards, 2H4-SCH 29851 and 2H4SCH 34117 (see Figure 1), were synthesized by Schering-Plough Research Institute (Kenilworth, NJ). Rat, mouse, rabbit, and dog plasma, with EDTA as anticoagulant, were purchased from Bioreclamation Inc. (Hicksville, NY). Optima grade (99.9%) acetonitrile and Optima grade (99.9%) methanol were obtained from Fisher Scientific Co. (Pittsburgh, PA). Formic acid (minimum 95%, ∼5% water, and 0.5% acetic acid) was obtained from Sigma Chemical Co. (St. Louis, MO). ReagentPlus grade (99.99%) ammonium acetate and glacial acetic acid (99.99%) were obtained from Aldrich. Milli-Q water used in this study was purified inhouse with an A10 Millipore water purification system (Millipore Corp., Bedford, MA). Sample Preparation. Analyte stock solution containing SCH 29851 and SCH 34117 at 100 µg/mL each was prepared in methanol. Internal standard (IS) stock solution containing 2H4-SCH 29851 and 2H4-SCH 34117 at 100 µg/mL each was also prepared in methanol. Calibration curve standards (STD) were prepared at 10 concentrations: 1, 2, 5, 10, 20, 50, 100, 200, 500, and 1000 ng/mL. Quality control (QC) samples were prepared at four concentrations: QC limit of quantitation (LOQ) at 1 ng/mL, QC low at 3 ng/mL, QC medium at 400 ng/mL, and QC high at 800 ng/mL. STD 10 at 1000 ng/mL was prepared by diluting 250 µL of the analyte stock solution into 25 mL of blank plasma. Standards 1-9 were prepared by serial dilutions of STD 10 with blank plasma. Similarly, QC high at 800 ng/mL was prepared by diluting 200 µL of the analyte stock solution into 25 mL of blank plasma. QC samples at the LOQ, low, and medium levels were prepared by serial dilutions of QC high with blank plasma. Internal standard working solution (ISWS) at 40 ng/mL in 100 mM ammonium acetate pH 6.0 was prepared by diluting 200 µL of IS stock solution (100 µg/mL) into 500 mL of 100 mM ammonium acetate solution at pH 6.0 in a 500-mL volumetric flask. Analytical Chemistry, Vol. 73, No. 8, April 15, 2001

1741

Figure 2. Schematic diagram of the four-channel multiplexed electrospray ion source showing the direction of the ions coming from sprayer 1 into the sampling rotor, moving into the sampling cone and then into the hexapole ion lense.

Extraction Procedure. Five hundred microliters of each plasma sample was aliquoted into a Coster cluster tube (Fisher Scientific, Fair Lawn, NJ) and arranged into the 96-well format. The extraction procedure was carried out using a TOMTEC Quadra 96 model 320 liquid-handling system (TomTec, Hamden, CT) and 3M Empore C18 SD 96-well extraction disk plate (Varian Associates, Sugar Land, TX). Two hundred and fifty microliters of 2H4-SCH 29851 and 2H4-SCH 34117 ISWS (40 ng/mL) was added to each sample tube, except for the double blanks in which 250 µL of 100 mM ammonium acetate, pH 6.0, was added instead of ISWS. The C18 disk plates were conditioned with 0.8 mL of methanol and 0.8 mL of Milli-Q water. Each sample was loaded onto the plates in two aliquots. Once the sample had completely flowed through the plate, the plate was washed with 0.8 mL of Milli-Q water and 0.8 mL 20:80 solution B (2 mM ammonium acetate, 0.1% acetic acid, 0.1% formic acid in 50:50 (v/v) acetonitrile in methanol): solution A (2 mM ammonium acetate, 0.1% acetic acid, 0.1% formic acid in water). The analytes and their respective IS were then eluted into a 1-mL 96-well collection plate (Fisher Scientific Co., Pittsburgh, PA) using 150 µL of solution B. The C18 disk plates were then washed with 300 µL of solution A. This yielded ∼450 µL of processed sample solution. After mixing, 5 µL was injected directly into the LC/MS/MS system. At the LOQ, this represented 5.56 pg of each analyte injected onto the HPLC column. Liquid Chromatography/Mass Spectrometry. The autosamplere consisted of a Gilson 215/889 multiple injection module (Gilson, Inc., Middleton, Wisconsin). The Gilson 215/889 multiple injection module has four injector needles that can draw four samples simultaneously from a 96-well plate and then inject 1742

Analytical Chemistry, Vol. 73, No. 8, April 15, 2001

the samples onto four individual LC columns simultaneously. An HP 1100 pump (Hewlett-Packard, Palo Alto, CA) was used to deliver a total flow rate of 800 µL/min and split to 4 × 200 µL/min into four LC columns. Postcolumn split introduced a total of 4 × 60 µL/min of the liquid flow into the mass spectrometer. The remaining flow was directed to waste. Four HPLC columns and guard columns (BDS-C8, 100 × 2 mm and 20 × 2 mm, 5 µm, Keystone Scientific Inc., Bellefonte, PA) were used in the parallel LC/MS/MS system. Separation of SCH 29851 and SCH 34117 was achieved under isocratic conditions with 85:15 methanol/25 mM ammonium formate, pH 3.5, as the mobile phase. The total run time (including cycle time) was 3.5 min. The mass spectrometer used in this study was a Micromass Quattro Ultima triple quadrupole equipped with a four-channel multiplexed electrospray ion source (Micromass UK Limited, Manchester, England). Figure 2 is a schematic diagram of the four-channel multiplexed electrospray ion source. The MUX interface consists of four electrospray probes and a sampling rotor positioned coaxially with the sampling cone. The effluents from four HPLC columns were continuously electrosprayed at 60 µL/ min using the four electrospray probes. However, at any one time, the position of the sampling rotor allows only the spray from one probe to be admitted into the sampling cone of the mass spectrometer. A programmable stepper motor, which is controlled by the Micromass Masslynx data system, controls the position of the sampling rotor. This allows the data system to track the data from each of the four sprays separately. The minimum time needed to step from one sprayer to another sprayer (interspray step time) is 50 ms.

The mass spectrometer was operated in positive ion multiple reaction monitoring mode. The following MRM transitions were monitored for the analytes and internal standards: SCH 34117, m/z 311-259; 2H4-SCH 34117, m/z 315-263; SCH 29851, m/z 383-337; 2H4-SCH 29851, m/z 387-341. Dwell time for each transition was 50 ms, with 20 ms interdwell delay and 50 ms interspray step time. Total cycle time was 1.24 s. A positive voltage of 4 kV was applied to each of the four electrospray probes. Source temperature was 150 °C. Desolvation gas was heated to 400 °C. One hundred and twenty volts was applied to the sampling cone. Collision energy was 20 eV for all the analytes and internal standards. Masslynx version 3.4 software was used for data acquisition and processing. RESULTS AND DISCUSSION Precision and Accuracy of the Assay. For the MUX interface evaluation, a one-run method validation over the concentration range of 1-1000 ng/mL was performed simultaneously in dog, rat, mouse, and rabbit plasma. For each species, the run contained duplicate calibration curve standards at 10 concentrations, QC samples at 4 concentrations (n ) 6 at each concentration), and 4 matrix blanks. (The total number of injections per species was 48.) Two 96-well plates were arranged so that the autosampler simultaneously injected four of the same sample type, each in a different species. For example, calibration curve standards at 1 ng/mL in rat, mouse, rabbit, and dog plasma were injected simultaneously. Data for each injection were acquired into a unique and separate data file. For both SCH 29851 and SCH 34117, peak area ratios (analyte/ IS) for calibration curve standards were plotted versus concentration and fit to a quadratic regression with 1/y weighting. All calibration curves for SCH 29851 and SCH 34117 had correlation coefficients greater than 0.99. Quadratic fits for the calibration curves were also obtained using a single inlet electrospray ion source. The deviation from linearity at the higher end of the curves is most probably due to a nonlinear response of the electrospray ionization process at higher analyte concentrations.20 The performance of the assay for SCH 29851 and SCH 34117 was determined by assessing the precision (% CV) and accuracy (percent difference from nominal) for QC samples in replicates of six at four concentration levels (Tables 1 and 2). For SCH 29851, the precision ranged from 0.967 to 16.0% and accuracy ranged from -8.44 to 10.5% of nominal across all four species. For SCH 34117, the precision ranged from 0.684 to 11.0% and the accuracy was between 6.36 and -9.06%. For both the drug (SCH 29851) and the metabolite (SCH 34117), the precision and accuracy were within current FDA-recommended acceptance criteria of (15% at the low, medium, and high QC levels and (20% at the LOQ QC level. The ranges in precision and accuracy in Tables 1 and 2 may be attributed to several sources. Most of the variation in the measurements is derived from sample preparation, extraction, and matrix effects on the ionization process. Furthermore, with the MUX interface, fewer scans are obtained for a given analyte peak, which may lead to more variation in peak area measurements relative to a single sprayer experiment. This may result in a wider range in precision and accuracy with MUX, especially at the LOQ. (20) Kebarle, P.; Tang, L. Anal. Chem. 1993, 65, 972A-986A.

Table 1. Quality Control Sample Concentrations of SCH 29851 (Within-Run Precision and Accuracy) QC LOQ QC low QC medium QC high 1.00a 3.00 400 800 (ng/mL) (ng/mL) (ng/mL) (ng/mL) mean (rabbit plasma) % CV n mean % diff

0.982 6.92 6 -1.83

2.82 3.20 6 -5.94

409 1.15 6 2.20

829 1.75 6 3.60

mean (mouse plasma) % CV n mean % diff

1.06 3.75 6 5.83

3.10 2.84 6 3.33

414 1.11 6 3.43

833 2.65 6 4.18

mean (rat plasma) % CV n mean % diff

1.11 16.0 6 10.5

2.75 2.18 6 -8.44

409 1.03 6 2.25

836 1.21 6 4.44

mean (dog plasma) % CV n mean % diff

0.952 10.6 6 -4.83

2.90 2.24 6 -3.44

417 0.967 6 4.15

823 2.06 6 2.89

a

Nominal value.

Table 2. Quality Control Sample Concentrations of SCH 34117 (Within-Run Precision and Accuracy) QC LOQ QC low QC medium QC high 1.00a 3.00 400 800 (ng/mL) (ng/mL) (ng/mL) (ng/mL) mean (rabbit plasma) % CV n mean % diff

1.05 6.47 6 4.67

2.95 2.12 6 -1.67

422 1.54 6 5.61

851 2.00 6 6.36

mean (mouse plasma) % CV n mean % diff

1.02 8.54 6 2.17

3.08 2.50 6 2.56

407 1.73 6 1.80

816 2.05 6 2.00

mean (rat plasma) % CV n mean % diff

1.05 9.98 6 4.83

2.73 2.55 6 -9.06

401 0.684 6 0.255

810 2.18 6 1.30

mean (dog plasma) % CV n mean % diff

0.973 11.0 6 -2.67

2.90 2.25 6 -3.39

413 0.724 6 3.36

821 1.29 6 2.62

a

Nominal value.

Representative ion chromatograms for LOQ QC samples for SCH 29851 and SCH 34117 extracted from rat plasma are shown in Figure 3. The chromatograms show good signal-to-noise ratios for both analytes at the LOQ. Although the LOQ concentration is 1 ng/mL, only 5 µL out of 450 µL of processed sample was injected, which corresponds to 5.56 pg injected on-column. This small injection volume was used so that we could adequately test the absolute detection limits of a MUX triple quadrupole system. If the entire sample were injected, a lower LOQ could theoretically be achieved with the MUX interface. Chromatograms from extracted blank dog plasma with internal standard for SCH 29851 and SCH 34117 are shown in Figure 4. The signal for both analytes is negligible, demonstrating that there are no endogenous components in the plasma that interfere with Analytical Chemistry, Vol. 73, No. 8, April 15, 2001

1743

Figure 3. Multiple reaction ion chromatograms for a limit of quantitation quality control sample in rat plasma showing from bottom trace to top trace SCH 29851, SCH 34117, 2H4-SCH 29851, and 2H4-SCH 34117.

Figure 4. Multiple reaction ion chromatograms for a blank plus internal standard sample in dog plasma showing from bottom trace to top trace SCH 29851, SCH 34117, 2H4-SCH 29851, and 2H4-SCH 34117.

the analytes and there is very little carryover of analyte from previous injections. However, because the MUX interface cosprays four LC effluents continuously into the mass spectrometer, the potential for intersprayer cross talk needed to be investigated with another experiment as described below. Intersprayer Cross Talk. One general aspect of the MUX interface that would be of particular concern when study samples of unknown concentration are being analyzed is the potential for cross talk between the four sprayers. An experiment was designed to investigate cross talk during the course of analysis at different concentration levels. Standards and QC samples from dog plasma were used in this experiment. For sprayers 2, 3, and 4, a validation run was injected with each sprayer using the 1744 Analytical Chemistry, Vol. 73, No. 8, April 15, 2001

same injection order. For sprayer 1, only solvent blanks were injected. Intersprayer cross talk was found to be relatively low but dependent upon the concentration of the samples cosprayed at sprayers 2, 3, and 4. At concentration levels less than 50 ng/mL, no detectable cross talk was observed. Figure 5 shows the representative chromatograms for different channels at 100 ng/mL and 1000 ng/mL. Cross talk for SCH 29851 and SCH 34117 was approximately 0.01% at 100 ng/mL and 0.08% at 1000 ng/mL. In the four-species validation experimental design, cross talk between the sprayers was eliminated because injection orders across the four channels were identical. However, when unknown samples are being analyzed, cross talk may present limitations in

Figure 5. Multiple reaction ion chromatograms for a 100 ng/mL sample in dog plasma through sprayer 2 (top left panel) and a solvent blank through sprayer 1 (top right panel) a 1000 ng/mL dog plasma sample through sprayer 2 (bottom left) and a solvent blank through sprayer 1 (bottom right). For each panel, bottom trace to top trace: SCH 29851, SCH 34117, 2H4-SCH 29851, and 2H4-SCH 34117.

the dynamic range of the assay. This is due to the current acceptance criterion for blank samples: The peak area of a blank sample should be less than 20% of the peak area of the LOQ calibration curve standard. The parallel LC/MS/MS system could also be used to analyze different compounds on different channels. With this application, cross talk between the channels would not be an issue as long as the compounds have different MRM transitions. In an effort to reduce intersprayer cross talk, further work on the design of the MUX interface is ongoing. Comparison of the MUX Interface with Conventional Single Sprayer Interface. The obvious advantage of the MUX interface is its throughput: it decreases the total analysis time by a factor of 4. In this validation study, there were 48 samples for each validation run and a 3.5-min run time. Using a single sprayer interface to analyze samples from four validation runs, the total analysis time would be 11 h and 12 min. Using the MUX interface, the total analysis time was 2 h and 48 min. The sensitivity of the MUX interface, however, is lower than the conventional single sprayer interface. Flow injection experiments were carried out to evaluate the relative sensitivities of the two interfaces. Five microliters of a neat solution containing both SCH 29851 and SCH 34117 at 1 ng/mL and their internal standards at 100 ng/mL (neat LOQ) were analyzed with the MUX and the single sprayer interface. Except for the different ion sources, all experimental parameters were kept the same. Figure 6 shows the neat LOQ signals with both interfaces. The results

indicated a ∼3-fold decrease in signal using the MUX interface. There could be several factors contributing to the decrease in signal. First, with the MUX interface, the position of the sprayers could not be optimized as with the single sprayer interface. In addition, the electrospray desolvation efficiency in the MUX interface may be lower than the single sprayer interface. With MUX, although each LC flow is split to 60 µL/min postcolumn, a total flow of 240 µL/min is introduced into the ion source region which may cause a decrease in the efficiency of the electrospray desolvation process. Furthermore, with the single sprayer, the desolvation gas blows coaxially to the sprayer whereas, with the MUX interface, the desolvation gas blows countercurrent against the spray thereby reducing the desolvation efficiency. Another concern with the MUX interface with a quadrupole mass analyzer is that the total cycle time is longer than with a single sprayer interface. As an example, in this study, we monitored four transitions. The chromatographic peak widths were on average 15 s. The conventional single sprayer was set up using a dwell time of 200 ms for each transition, with 20-ms interdwell delay. This gave a total cycle time of 0.88 s, and 17 data points could be detected across the peak for each transition. When the MUX interface was used, the dwell time for each transition was reduced to 50 ms in order to maintain an acceptable number of data points per peak. Taking into consideration a 20ms interdwell delay and a 50-ms interspray step time, the total cycle time was 1.24 s. With these settings and MUX data Analytical Chemistry, Vol. 73, No. 8, April 15, 2001

1745

Figure 6. Multiple reaction ion chromatograms for a neat solution containing both SCH 29851 and SCH 34117 at 1 ng/mL and their internal standards at 100 ng/mL (neat LOQ) comparing the MUX (top four chromatograms) with the single sprayer (bottom four). From bottom trace to top trace: SCH 29851, SCH 34117, 2H4-SCH 29851, and 2H4-SCH 34117.

acquisition, 12 data points could be detected across each 15-s peak. Hence, when MUX is used with a quadrupole mass analyzer, it is important to consider the dwell time (minimum of 10 ms) and chromatographic peak width since these parameters will ultimately limit the number of analytes and internal standards that may be monitored in an assay. Originally, the MUX technology was interfaced to a time-offlight (TOF) mass spectrometer.16 With the fast acquisition capacity of the TOF instrument (>10 spectra/s), a MUX interface containing eight channels may be used while still maintaining mass spectral integrity. With the MUX interfaced to a TOF, fast chromatography with narrow chromatographic peak widths may be achieved. The disadvantages of a MUX-TOF for quantitation are reduced sensitivity and limited dynamic range relative to a triple quadrupole in the MRM mode. Advantages and Disadvantages of the Current Parallel LC/ MS/MS Setup. The parallel LC/MS/MS system used in this 1746 Analytical Chemistry, Vol. 73, No. 8, April 15, 2001

investigation consists of one autosampler (Gilson 215/889 multiple injection module) and one HPLC system (HP 1100). Other options are four HPLC systems with one or four autosamplers. Advantages of a single HPLC system setup are simplicity and low cost, and it saves laboratory space. A disadvantage is difficulty in regulating the flow rate across the four channels. Using one HPLC system, the liquid flow is split evenly into four channels using a Valco splitter. From that point, the flow rates are pressure regulated. To maintain the same flow rate across the channels, care must be taken to ensure that the back pressures in the four channels are kept the same. Connection tubing, columns, and guard columns all generate back pressure. Columns and guard columns from the same manufacturer, even from the same batch, most likely generate different back pressures. In our first attempt to do simultaneous validations, variation for retention times across the channels was greater than 20% even with the same length of connection tubing for each channel. We then took the approach

of starting with tubing of approximately the same length, checking the flow rate at the end and adjusting the length according to the difference in flow rate. This allowed for a more even split of liquid flow across the four channels, as is demonstrated by the very close agreement of retention times across the channels. A parallel LC/MS/MS design utilizing four separate HPLC systems has the advantage of better flow control and, therefore, consistency in retention time. Another advantage is the choice of HPLC conditions. In the situation when different compounds are analyzed in different channels, each compound could be optimized with its choice of HPLC column, mobile phases, gradient, etc. Disadvantages of the multisystem approach are that it takes more laboratory space and it also costs more than the single HPLC system design. CONCLUSIONS We have described the use of a four-channel multiplexed electrospray ion source for the simultaneous validation of four

assays in four different matrixes. The validation experiments in each matrix met current FDA acceptance criteria for precision and accuracy, demonstrating that this ion source may be used to support GLP-regulated drug development studies. The MUX interface decreased the total analysis time by a factor of 4. Cross talk between sprayers was evaluated as a function of concentration and shown to be approximately 0.01% at 100 ng/mL and 0.08% at 1000 ng/mL. This may present limitations on the dynamic range of assays. These results demonstrate that 96-well sample processing in conjunction with the MUX interface may decrease the time required for multiple validations and increase sample throughput in a drug development bioanalytical laboratory.

Received for review October 27, 2000. Accepted January 26, 2001. AC0012694

Analytical Chemistry, Vol. 73, No. 8, April 15, 2001

1747