Anal. Chem. 1997, 69, 4519-4523
Articles
Tandem-in-Time Mass Spectrometry as a Quantitative Bioanalytical Tool David T. Rossi,* Keith L. Hoffman, Nancy Janiczek-Dolphin, Howard Bockbrader, and Tresavon D. Parker
Bioanalytical Core Group, Department of Pharmacokinetics and Drug Metabolism, Parke-Davis Pharmaceutical Research, Division of Warner-Lambert Company, 2800 Plymouth Road, Ann Arbor, Michigan 48189
Tandem-in-time mass spectrometry, as implemented on an ion-trap detector (ITD), is the process whereby precursor ions are created, stored in a radio frequency (rf) trapping field, and then sequentially fragmented to form product ions by application of additional rf waveforms. As with any form of tandem mass spectrometry (MS/MS), tandem-in-time MS is highly selective, by virtue of both mass discrimination and specific gas-phase chemistry. Beyond this, however, tandem-in-time MS offers ion throughput efficiency and cost advantages over either quadrupole or sector instruments. This paper will describe the use of capillary gas chromatography combined with tandem-in-time mass spectrometry to quantify a novel therapeutic agent extracted from human plasma. For an example compound, a quantitation limit of 25 pg/mL (S/N ≈ 10, 15 fmol on-column) was attained out of plasma. The interday imprecision was e12.2% over a dynamic range extending to 10 ng/mL. Due to favorable ionization conditions for the test analytes, electron ionization resulted in formation of M+ ions, with very little fragmentation, allowing for maximum assay sensitivity. Although method characterization and validation demonstrated adequate instrumental performance, some lack of ruggedness was encountered during routine application. When coupled to an appropriate chemical separation such as liquid or gas chromatography, tandem mass spectrometry (MS/ MS) demonstrates impressive selectivity and sensitivity and has been used to solve numerous quantitative bioanalytical problems, including trace quantitation of xenobiotics and other clinically significant analytes in biological fluids.1 Although there are appreciable losses in ion transmission relative to single mass spectrometry experiments, MS/MS more than compensates by strongly discriminating against chemical noise, resulting in real gains in selectivity and quantitation limits.2 To date, a majority of bioanalytical applications employ a “tandem-in-space” approach whereby ions are produced, separated, fragmented, and separated again as they traverse sequential compartments of a mass spectrometer.3 These individualized compartments require discrete mass separators, usually quadrupoles, additional ion lenses, and multiple vacuum pumps. (1) Kinter, M. Anal. Chem. 1995, 67, 493R. (2) Burlingame, A. L.; Boyd, R. K.; Gaskell, S. J. Anal. Chem. 1996, 68, 599R. (3) Brittain, R.; Feigel, C. Am. Lab. 1994, Feb, 44. S0003-2700(97)00247-3 CCC: $14.00
© 1997 American Chemical Society
By contrast, the tandem-in-time approach is an MS/MS technique which can be implemented on a quadrupole ion-trap mass spectrometer.4,5 After ion production, separation, and storage, precursor ions contained in a storage field can be fragmented to product ions using energy imparted as a radio frequency (rf) waveform and through incidental collision with neutral gas (He) molecules. This process has been termed collisional activation (CA) or collision-induced dissociation (CID) and is widely used in many types of MS/MS experiments.6 The rf waveform, applied through the endcaps of the ion trap, can be resonant with the oscillating frequency of the target ion (resonant fragmentation) or not (nonresonant fragmentation). Although either type of waveform can lead to useful fragmentation, having both options helps assure that a satisfactory fragmentation can be obtained. Once formed, the characteristic product ions can be separated by the highly efficient ion manipulation capabilities of the ion trap. Tandem-in-time mass spectrometry is a very efficient MS/MS experiment because ions are not transported from one chamber to another as they are produced, stored, fragmented, and separated. Rather, a single chamber, utilizing a quadrupole rf field, is used to perform all functions. A desirable byproduct of this high efficiency is that multiple quadrupole analyzers, numerous ion lenses, and multiple pumping systems are no longer required. Many tandem-in-time MS experiments can be performed with benchtop ion-trap instruments of modest cost.7 Also, if any spectral scanning is required, ion transmission rates will be higher for ion traps than for quadrupoles.4 Higher pressures (10-3 Torr), which are problematic for quadrupole systems, are actually beneficial to ion traps by providing improved mass resolution and sensitivity.8 Although there have been numerous recent bioanalytical applications of tandem-in-space mass spectrometry, tandem-in-time mass spectrometry applications have, thus far, been few and limited. The determination of volatile organics in pharmaceuticals9 and applications involving environ(4) McLuckey, S. A.; Van Berkel, G. J.; Goeringer, D. E.; Glish, G. L. Anal. Chem. 1994, 66, 689A. (5) McLuckey, S. A.; Van Berkel, G. J.; Goeringer, D. E.; Glish, G. L. Anal. Chem. 1994, 66, 737A. (6) Lorquet, J. C. Mass Spectrom. Rev. 1994, 13, 233. (7) Wehmeyer, K. R.; Knight, P. M.; Parry, R. C. J. Chromatogr. 1996, 676, 53. (8) Stafford, G. C.; Kelley, P. E.; Syka, J. E. P.; Reynolds, W. E.; Todd, J. F. J. Int. J. Mass Spectrom. Ion Processes 1984, 60, 85. (9) Schuberth, J. Anal. Chem. 1996, 68, 1317.
Analytical Chemistry, Vol. 69, No. 22, November 15, 1997 4519
Figure 1. Chemical structures for (a) I and (b) II.
mental pollutants10 and food aroma components11 are recent noteworthy examples. This work demonstrated the utility of iontrap mass spectrometry coupled to capillary gas chromatography for the trace quantitation of a novel therapeutic drug in human plasma. A robotically automated version12,13 of a previously reported solid-phase extraction14 was used for sample preparation. Product ions, formed by tandem-in-time mass spectrometry, were monitored and used in quantification of the analytes. Precision and accuracy were defined over the dynamic range of the methodology. The selectivity and quantitation limits of the methodology were defined and compared to previously reported work using gas chromatography with mass spectrometric detection.15 The tandem-in-time methodology was applied to quantification of a novel cholinesterase inhibitor being tested in human subjects for the first time. EXPERIMENTAL SECTION Materials. Compound I, (CI-1002 HCl, >99% purity, molecular mass of free base ) 269.4 Da), 1,3-dichloro-6,7,8,9,10,12-hexahydroazepino[2,1-b]quinazoline monohydrochloride, and compound II (internal standard, PD 146023-2, >99% purity, molecular mass of free base ) 283.2 Da), 1,3-dichloro-6,7,8,9,10,12-hexahydro-2methylazepino[2,1-b]quinazoline monohydrochloride, were prepared by Parke-Davis Pharmaceutical Research (Ann Arbor, MI). The structures for these analytes are given in Figure 1. Ultrapure helium (99.999%) was purchased from AGA (Maumee, OH). Sodium phosphate, acetonitrile, dioxane, methanol, water, and trifluoroacetic acid were either reagent or HPLC grade and were purchased from EM Science (Gibbstown, NJ) or Mallinkrodt (Paris, KY) and used as received. Solid-phase extraction (SPE) columns (Bond Elut C-18 200 mg, 3 mL cartridges) were purchased from Varian Sample Preparation Products (Harbor City, CA). Blank human plasma was obtained from Interstate Blood Bank Inc., (Memphis, TN). Preparation of Standards and Controls. A fresh aqueous stock solution containing 200 µg mL-1 of I (free-base equivalents) was prepared for each batch run. A 2.0 µg mL-1 solution was prepared from this by volumetric dilution with water. This dilute stock solution was further diluted volumetrically with water to (10) PlomLey, J. B.; March, R. E.; Mercer, R. S. Anal. Chem. 1996, 68, 2345. (11) Huston, C. K. In Food Aroma Analysis:Science and Technology; Marsilli, R., Ed.; Marcel Dekker: New York, 1996. (12) Hoffman, K. L.; Andress, L. D.; Parker, T. D.; Guttendorf, R. J.; Rossi, D. T. Lab. Robot. Autom. 1996, 8, 237. (13) Parker, T. D.; Wright, D. S.; Rossi, D. T. Anal. Chem. 1996, 68, 2437. (14) Rossi, D. T.; Overmyer, S. K.; Lewis, R. C.; Narang, P. K. J. Chromatogr. 1991, 566, 257. (15) Hoffman, K. L.; Rossi, D. T.; Wright, D. S. J. Pharm. Biomed. Anal. 1995, 13, 979.
4520
Analytical Chemistry, Vol. 69, No. 22, November 15, 1997
prepare working standards at 0.1, 0.2, 0.4, 1.0, 2.0, 4.0, 20.0, and 40 ng mL-1. A 250 µL aliquot of each working standard was added to 1.0 mL of blank human plasma to prepare standards for calibration over the range from 25 pg mL-1 to 10 ng mL-1. A 200 µg mL-1 (free-base equivalents) stock solution of II was prepared for each batch run. This solution was diluted with water to prepare a stock II solution of 2.0 µg mL-1. This 2.0 µg mL-1 solution was diluted with water to prepare a working internal standard solution of 100 ng mL-1. Addition of 250 µL aliquots of the working internal standard solution to human plasma samples gave a concentration of 25 ng mL-1. Human plasma quality controls containing 0.075, 0.75, and 7.5 ng mL-1 I (free-base equivalents) were prepared by diluting aliquots of stock solutions with human plasma. Greater than 95% of these quality controls were of the biological component. Controls were subdivided into 3.5 mL aliquots and stored frozen in plastic screw-cap test tubes until use. Instrumentation. A Varian Saturn II GC/MS equipped with an 8200 autosampler and a Compaq Prolinea 4/33 data system was utilized. Sample injection was performed with a 1078 universal capillary injector (UCI) and silanized glass on-column inserts. Sample injection volume was 5.0 µL, sandwiched by air gaps of 0.5 µL. After septum penetration, pre- and postinjection delays were 0.03 and 0.00 min, respectively. Sample injection was at 95 °C, at a rate of 5 µL s-1. At 20 s postinjection, the injector manifold was heated to 300 °C at 150 deg min-1 and held for 10.0 min. To eliminate solvent from the injection, an injector purge valve was held open until 0.33 min postinjection. The temperature of the transfer line was held constant (280 °C). Electron ionization (EI) was performed at 70 eV at a manifold temperature of 220 °C. Positive ions were monitored. For competitive performance experiments, a Hewlett-Packard (Wilmington, DE) gas chromatograph-mass selective detector was used. The ion trap dimensions were 1.564 cm (end cap-to-end cap distance), and the ring electrode diameter was 2.0 cm. The electric field amplitude (voltage divided by distance) was evaluated over the range from 35.8 to 44.8 V/cm. A DB-5ms fused silica capillary column (15 m × 0.25 mm i.d., 0.25 µm film thickness; J&W, Folsom, CA) was utilized. The column was initially operated at 130 °C for 2.30 min, after which the temperature was increased to 300 °C at 20 deg min-1 for 8.5 min, with a 4.0 min hold time. The carrier gas was helium with a column inlet pressure of 15 psi. The Saturn automated tuning program was used to tune the mass spectrometer, resulting in typical filament emission currents, multiplier voltages, and automatic gain control (AGC) settings of 20 µA, 1600 V, and 20 000, respectively. The AGC is set prior to each mass spectral scan and is used to correct for space charge effects and other spurrious signal fluctuations by adjusting the number of ions in the trap. This technique improves both linearity and precision of response. Selection of tandem-in-time conditions for I (parent ion mass window of 267 ( 1.5 Da) involved evaluation of nonresonant CID waveform amplitudes from 56 to 70 V, with 20 ms residence times. To further improve signal-to-noise ratio (S/N), the mass range for I was limited from m/z 210 to 270, with a data acquisition rate of 500 ms-1. This limited scan range allowed four individual scans (microscans) to be averaged for each mass scan. II was added to each sample as an internal standard and was monitored in a tandem mode, using the reaction m/z 283 going to m/z 235 with a scan rate of 500 ms-1. Data collection
began 7.0 min after injection, with a first acquisition segment duration of 1.8 min (I) and a second segment duration of 3.2 min (II). The relative numbers of ions for each analyte present at any given moment were determined from mass spectral peak heights at m/z 239 (IS) and 225 (I). Chromatographic retention times for I and II were 8.3 and 9.0 min, respectively. Assay Procedures. To 1.0 mL of human plasma were added 0.250 mL of I working standard or water (for samples, quality controls, or blanks), 0.250 mL of internal standard working solution, and 1.0 mL of sodium phosphate buffer (pH 6.0; 0.05 M). Using a previously described batch-robotic solid-phase extraction system,12-14 solid-phase cartridges (200 mg, C18) were conditioned by rinsing with 3 mL each of methanol, acetonitrile, water, and sodium phosphate (5 in. Hg for ∼5 s). Samples were robotically loaded on cartridges at full vacuum (20 in. Hg) to start all the plasma flowing evenly. Vacuum was then automatically reduced (5 in. Hg), and the remaining sample volumes were drawn through the cartridges. Cartridges were robotically washed with 1 mL of water and 1 mL of acetonitrile (5 in. Hg) and dried at full vacuum for 10 min. Analytes were eluted into 10 mm × 75 mm glass tubes with 1 mL of 2% trifluoroacetic acid in HPLC grade methanol (5 in. Hg). Samples were dried under nitrogen (10 psi, 50 °C), reconstituted with 30 µL of dioxane/acetonitrile (75:25 v/v), and transferred into low-volume conical glass injection vials. Volumes of 5.0 µL of each sample or standard were injected into the GC column. Data Reduction. The assay was validated over a I plasma concentration range of 25-10000 pg mL-1 by assaying eight calibration standards and three quality control samples in triplicate, in three separate batch runs. The best-fit line was determined by least-squares linear regression of peak area ratio (peak area of I at m/z 225/peak area of II at m/z 281) vs concentration from each batch run, using a weighting factor of 1/concentration × 2.17 Concentrations of I in quality controls and samples were calculated using peak-area ratios and regression parameters. RESULTS AND DISCUSSION Large Volume Injection. A programmable split valve, located inside the universal capillary injector, was operated in coordination with injection to allow the introduction of large sample volumes. The large-volume sample injection technique used here was one of several reviewed by Mol et al.16 and has been referred to as “solvent-split injection”. In this approach, a sample is injected with the split exit open at a temperature (95 °C) below the solvent boiling point (101 °C for dioxane). After evaporation of the solvent, the split valve is closed, and the analytes retained on the liner are transferred to the analytical column in splitless mode by rapidly increasing the injector temperature. Glass wool in the liner prevents the sample from being pushed to the bottom of the injector, avoiding a flooding of the column inlet. The system under discussion here was limited to an injection volume of 5 µL, due to capacity of the available syringes. Although a systematic study of the parameters associated with the largevolume injector has not been completed and will be the topic of a future report, the use of large-volume injection had generally (16) Mol, H. G. J.; Janssen, H. G.; Cramers, C. A.; Brinkman, U. A. T. J. High Resolut. Chromatogr. 1995, 18, 19. (17) 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.
Figure 2. Relative ion abundances versus nonresonant CID waveform amplitude for precursor ion (m/z 267) and product ions of I.
positive effects on this analytical method. System performance was generally good, as gauged by detection limit (4 pg on-column), precision of the assay (∼8% RSD from 0.75 to 7.5 ng/mL), and column efficiency (N ) 1.1 × 105 theoretical plates). Mass Spectra. The EI mass spectra for the compound selected to illustrate this approach have been reported.15 Briefly, the spectrum for I includes a predominant ion (m/z 267, 100%) with an associated isotopic cluster (m/z 265 (42%), 269 (70%), and 270) at lower abundances. Very little fragmentation was noted, with only one other ion exceeding 10% relative abundance (m/z 225)and none exceeding 15%. Because of this minimal fragmentation noted with electron ionization (EI), chemical ionization (CI) was not considered. The spectrum obtained for II displayed the most abundant ion (M+) at 281 Da, with an isotopic cluster similar to that for I. A minor (∼10%) fragment was noted at m/z 239 Da. These mass spectra were useful when selecting ion-molecule reaction products to monitor during the tandem-in-time experiments described below. Recovery from Human Plasma. Manual extraction (n ) 3) mean (%RSD) I recoveries were 97.5 (7.0)%, 94.6 (7.1)%, and 99.8 (8.4)% at 1.5, 7.0, and 20 ng mL-1, respectively.15 The pooled mean (%RSD) recovery at all levels was 96.4 (7.5)%. Mean (%RSD) internal standard recovery was 104 (7.6)% at 25 ng mL-1. Recent work using the robotic system described here12,13 has demonstrated that differences in recoveries between automated and manual extractions are minimal. Tandem-in-Time MS Performance. A significant advantage of the benchtop ion trap is that method optimization can involve adjustment of only one parameter, the rf waveform amplitude applied to the end caps. This parameter is increased from low to high during consecutive injections (or consecutive mass spectral scans of a single injection) to evaluate the extent of CID of the analyte(s). For I, a summary representative of this experiment is shown in Figure 2. As the amplitude, applied to the end caps as a nonresonant rf waveform, is increased, the precursor ion (m/z 267) becomes increasingly less durable. This ion decreases precipitously in relative abundance from 60 to 63 V of applied rf amplitude and is virtually undetectable at an amplitude of 66 V. A concurrent increase in product ion (m/z 225) is observed until a plateau is reached at 61 V. At amplitudes higher than about 63 V, secondary product ions (m/z 189 and 155) are created at the expense of the m/z 225 ion, which drops off sharply at amplitudes Analytical Chemistry, Vol. 69, No. 22, November 15, 1997
4521
greater than 69 V. These secondary product ions are the result of CID involving the m/z 225 ion and are equivalent to MS/MS/ MS or (MS3) experiments. For I, the optimal waveform amplitude (65 V) was graphically selected at the plateau region of product ion formation, providing the best sensitivity (Figure 2). A residual trace of precursor ion (m/z 267) was allowed to remain in the chromatogram, for use as a qualitative marker. This process was repeated for II (m/z 283) and daughter ion (m/z 239), with an amplitude of 64 V being optimal. Throughout this process, the reaction time was optimized at 20 ms, and the mass isolation window was set at 3 amu. Product ions were monitored for I, resulting in large selectivity and sensitivity increases relative to the selected ion monitoring of parent ion experiments reported previously. A comparison of chromatograms from plasma extracts collected by selected-ion monitoring of precursor and product ions shown in Figure 3A and B, respectively, illustrates this. The precursor ion chromatogram shows the analyte peak at 500 pg/mL, surrounded by a large number of comparably sized matrix peaks. Due to this chemical noise, the limit of quantitation is only 500 pg/mL, or 33 pg oncolumn. By monitoring a product ion at m/z 225, a significant reduction in chemical noise was obtained, and the quantitation limit decreased to 25 pg/mL or 4 pg on-column, with a S/N of ∼10. Enhancement was also made for II, but because this compound was added to each sample as an internal standard, the enhanced sensitivity was not as useful. Other small changes to the temperature program and carrier gas flow rate caused the observed changes in retention. Some of the improvement in quantitation limit resulted from the larger sample volume injected (2.5-fold increase), while the remainder (8-fold increase) was due to the monitoring of product ions. A representative blank chromatogram for the product ion experiment, Figure 3C, shows the complete absence of interfering peaks at retention times for either I or II. Precision and Accuracy. Within- and between-day accuracy and precision for the quantitation of I in human plasma were evaluated at 75, 750, and 7500 pg/mL, and the results are summarized in Table 1. With one exception, within-day precision was less than or equal to 14.5% RSD. Within-day relative errors ranged from -15.6 to 13.1% but were generally less than 10%. Between-day imprecision ranged from 8 to 12%, with relative errors of 9.5% or less. These results are widely recognized as acceptable for bioanalytical quantitation.17 It is possible that assay precision could be improved by mass spectral peak area rather than by peak height, but this option was not evaluated. Technique Comparison. The quantitation limits (defined here as a chromatographic peak of S/N ) 10 )17 for I were evaluated by three gas chromatography detection techniques: (1) quadrupole mass-selective detection, (2) ion trap utilizing a single MS experiment, and (3) ion trap using tandem-in-time mass spectrometry. The results, summarized in Table 2, indicate that the tandem-in-time MS technique has an 8-fold lower limit of quantitation than the single MS experiment performed on the ion trap, and the single MS ion trap experiment has an approximately 8-fold lower limit of quantitation than an experimental with a quadrupole mass-selective detector with selected ion monitoring. Although it is likely that different results would be obtained for other compounds and instruments, at least two groups have (18) Long, G. L.; Winefordner, J. D. Anal. Chem. 1983, 55, 712A.
4522 Analytical Chemistry, Vol. 69, No. 22, November 15, 1997
A
B
C
Figure 3. (A) Precursor ion chromatogram (m/s 267) for a human plasma extract containing 500 pg mL-1 of I. (B) Product ion chromatogram (m/z 225) for human plasma sample containing 25 pg mL-1 of I. (C) Product ion chromatogram (m/z 225) for human plasma blank sample extract.
reported similar results for ion-trap MS and MS/MS experiments.7,19 Application to Clinical Samples. To evaluate the routine quantitation capabilities of the tandem-in-time mass spectrometry technique, the methodology described and characterized above was applied to heparinized plasma samples collected during a pharmacokinetic study in which human subjects received escalating oral doses of drug (I). From these data, plasma concentration versus time profiles were constructed and used to guide dose escalation. Quantitation with the tandem-in-time approach demonstrated excellent sensitivity and selectivity and was capable of quantifying I over the range from 25 to 10 000 pg/mL of human (19) Consroe, P.; Kennedy, K.; Schram, K. Pharmacol. Biochem. Behav. 1991, 40, 517.
Table 1. Within- and Between-Day Accuracy and Precision for I at Three Levels level (pg/mL)
day 1 day 2 day 3 interday
%RSD %RE %RSD %RE %RSD %RE %RSD %RE
75
750
7500
6.3 0.0 8.6 -15.6 16.0 -12.7 12.2 -9.5
7.2 -2.3 a 3.2 10.5 -8.1 8.3 -3.2
14.5 7.5 4.8 13.1 2.3 6.9 8.1 9.2
a n ) 2 at this level. One replicate was excluded due to poor chromatography. %RSD are percent relative standard deviations. %RE are percent relative error.
Table 2. Comparative Limits of Quantitation (Defined as S/N ) 10) for I instrument type
amount on column
quadrupole MSD ion trap - single MS ion trap - MS/MS
270 pg, 1000 fmol 33 pg, 125 fmol 4 pg, 15 fmol
plasma. Although a formal methods comparison study was not conducted, the plasma levels obtained from the tandem-in-time method agreed with those obtained from a reference method (HPLC-UV) from 6 to 32% for a limited number (n ) 4) of samples. Further, the tandem-in-time mass spectrometry method extended the quantitation limit by approximately 1 order of
magnitude, relative to the reference method, and was useful for determining plasma levels of drug at doses as low as 1 mg. One disadvantage noted for the GC/MS technique, which was thought to be a result of inadequate/incompatible sample cleanup, was the occasional splitting of chromatographic peaks, which increased the overall variability of the method and decreased ruggedness. An increase in chromatographic peak splitting was also attributed to a deteriorated filament, used as an electron gun for the analyte ionization process. Replacement of this component significantly decreased the frequency of split chromatographic peaks. CONCLUSIONS The high degree of selectivity, exceptionally high sensitivity and efficiency, ease of collecting qualitative information, and modest cost are factors which should expand the use of tandemin-time mass spectrometry in the future. Although method characterization and validation demonstrated adequate instrumental performance, some lack of ruggedness was encountered during routine application and was thought to have arisen from inadequate sample cleanup, which was incompatible with capillary gas chromatography using large injection volumes. We expect a greater utilization of tandem-in-time mass spectrometry as advances in instrumentation are made. Received for review March 4, 1997. Accepted September 5, 1997. AC970247N X
Abstract published in Advance ACS Abstracts, October 15, 1997.
Analytical Chemistry, Vol. 69, No. 22, November 15, 1997
4523
