Strategy of Accelerated Method Development for ... - ACS Publications

Oct 26, 2009 - The recovery information of the analyte(s) of interest .... effects in the positive ion electrospray mode.7,9r11,13 A data set ... wate...
1 downloads 0 Views 1MB Size
Download PDF
Anal. Chem. 2009, 81, 9225–9232

Strategy of Accelerated Method Development for High-Throughput Bioanalytical Assays Using Ultra High-Performance Liquid Chromatography Coupled with Mass Spectrometry Guowen Liu, Heidi M. Snapp, Qin C. Ji,* and Mark E. Arnold Bioanalytical Sciences, Bristol-Myers Squibb Co., Route 206 and Province Line Road, Princeton, New Jersey 08543 Here we report a strategy for rapid method development of high-throughput bioanalytical assays using ultra highperformance liquid chromatography coupled with tandem mass spectrometry (uHPLC-MS/MS). First, a data set was established for the removal of representative phospholipids under different sample treatments to guide subsequent method development for various compounds. The recovery information of the analyte(s) of interest under different extraction conditions was then obtained during method development. With the recovery profiles and the pre-established knowledge of phospholipids removal in place, an optimal extraction condition was identified to give not only a satisfactory recovery but also a good cleanup of the sample. A rapid LC or uHPLC method was developed without the need of extensive column wash after the elution of the analyte. This strategy was demonstrated through the method development of a uHPLC-MS/MS bioanalytical assay for the quantitation of ketoconazole in human plasma with liquid-liquid extraction using a hexane and ethyl acetate solvent system. The retention time for ketoconazole through an isocratic elution was 18 s. Good accuracy and precision were obtained. Assay ruggedness was demonstrated by consistent internal standard responses and retention time for 500 sequential injections. In addition, consistent results were obtained for incurred sample reanalysis. In the bioanalytical area, rapid method development and highthroughput bioanalytical assays are critical for not only reducing cost but also for shortening the drug development time cycle. It can help the pharmaceutical industry accelerate regulatory filings at different drug development stages and bring the medicine to patients earlier. Although the feasibility of high-throughput bioanalysis has been demonstrated about 10 years ago with the liquid chromatography coupled with tandem mass spectrometry (LC-MS/MS) analysis of 2000 samples per 24 h,1,2 it has only become more practical with recent advances in LC, such as sub-2 * To whom correspondence should be addressed. Qin C. Ji, Ph.D, e-mail: [email protected]. (1) Zweigenbaum, J.; Heinig, K.; Steinborner, S.; Wachs, T.; Henion, J. Anal. Chem. 1999, 71, 2294–2300. (2) Zweigenbaum, J.; Henion, J. Anal. Chem. 2000, 72, 2446–2454. 10.1021/ac901316w CCC: $40.75  2009 American Chemical Society Published on Web 10/26/2009

µm particle columns and pumps that can operate at high back pressures (up to ∼15 000 psi). As the dominant technology in bioanalysis of small molecule drug candidates for many years, liquid chromatography coupled with mass spectrometry (LC-MS) has become a well-established technology with regard to assay development. One area that is well understood is the effect of the biological matrix background on assay performance.3,4 The choices for reducing matrix effects are often either to develop a good chromatographic method to separate the analyte from interference matrix background and clean up the column after each injection (i.e., less downstream matrix effects) or to have a good sample cleanup to remove major interference matrix background prior to LC-MS analysis. Plasma is the most common matrix for biological samples, and the abundant phospholipids present in plasma are one of the biggest concerns for assay performance.3,5 When a significant amount of phospholipids are present in the final samples ready for injection, it will be very important to clean these highly retentive phospholipids off the LC column so that there will be no downstream ion suppression during subsequent LC-MS analysis. One approach to reduce downstream matrix effects is to use a precolumn backflush between sample injections.4,6 However, it could be difficult to implement this with a short run time (e.g., less than 30 s) and high column back pressure. A more frequently used approach is to have a column wash stage and recondition the column after the analyte has eluted. However, this approach adds a significant amount of time to the LC run, which greatly affects the throughput. With the potential risk from phospholipids, examination of phospholipids in the final samples during method development is currently a common practice5,7-11 across the pharmaceutical (3) Jemal, M.; Xia, Y. Curr. Drug Metab. 2006, 7, 491–502. (4) Chang, M.; Ji, Q.; Zhang, J.; El-Shourbagy, T. Drug Dev. Res. 2007, 68, 107–133. (5) Wu, S.; Schoener, D.; Jemal, M. Rapid Commun. Mass Spectrom. 2008, 22, 2873–2881. (6) Ji, Q.; Rodila, R.; Gage, E.; El-Shourbagy, T. Anal. Chem. 2003, 75, 7008– 7014. (7) Du, L.; White, R. Rapid Commun. Mass Spectrom. 2008, 22, 3362–3370. (8) Chambers, E.; Wagrowski-Diehl, D.; Lu, Z.; Mazzeo, J.; Chambers, E. J. Chromatogr., B: Anal. Technol. Biomed. Life Sci. 2007, 852, 22–34. (9) Little, J.; Wempe, M.; Buchanan, C. J. Chromatogr., B: Anal. Technol. Biomed. Life Sci. 2006, 833, 219–230. (10) Ismaiel, O.; Halquist, M.; Elmamly, M.; Shalaby, A.; Karnes, H. J. Chromatogr., B: Anal. Technol. Biomed. Life Sci. 2007, 859, 84–93. (11) Ismaiel, O.; Halquist, M.; Elmamly, M.; Shalaby, A.; Thomas Karnes, H. J. Chromatogr., B: Anal. Technol. Biomed. Life Sci. 2008, 875, 333–343.

Analytical Chemistry, Vol. 81, No. 22, November 15, 2009

9225

Figure 1. Flowchart for the accelerated method development strategy.

industry. This is often done by monitoring the elution profile of the analyte and the phospholipids or through monitoring the LC-MS chromatographic ion suppression/enhancement of the analyte. However, no matter what compound is tested, phospholipids removal behavior from plasma under specific extraction conditions will be the same. Therefore, this information could be used repeatedly for different compounds and at different laboratories. Unfortunately, this information is not currently generated systematically and often is not well organized, retained, or utilized again for future method development. In our opinion, the evaluation of matrix background (specifically phospholipids) during method development for every new compound/each new method is a repetitive effort, unnecessarily generating information that already exists, and as a result slows down the method development process. Establishing a database for such information covering most commonly used sample cleanup techniques/ conditions will be very helpful, especially for people doing method development routinely. As shown in Figure 1, here we propose a strategy for fast method development of rapid uHPLC-MS bioanalytical assays. First, a database is established for sample cleanup under different sample preparation conditions. Second, recovery information of the analyte(s) of interest is profiled under different extraction conditions. If a good recovery of the analyte can be achieved where the sample cleanup is satisfactory, these extraction conditions can be used for sample preparation. If, however, satisfactory recovery of the analyte can only be achieved with poor sample cleanup, then a different extraction procedure will need to be explored or another approach to clean up the samples will need to be applied, either prior to or during LC-MS/MS analysis time. In real situations, scientists can usually predict a good recovery condition of a specific compound based on their experience and the physicochemical properties of the compound. Once an optimal condition is identified, a rapid LC or uHPLC method can be readily developed without extensive column wash after the elution of the analyte(s) at the end of the each run. In addition, the evaluation of recovery can be done simultaneously for multiple compounds. This could allow a simultaneous development of extraction 9226

Analytical Chemistry, Vol. 81, No. 22, November 15, 2009

conditions for multiple assays, as is frequently needed in the discovery setting. This strategy was demonstrated through method development of a uHPLC-MS/MS bioanalytical assay for the quantitation of ketoconazole in human plasma samples. A data set was created for sample cleanup after liquid-liquid extraction (LLE), represented by the removal of phosphotidylcholine (PC), lysophosphatidylcholine (lysoPC), and sphingomyelins (SM) in the final extracted samples. It is known that PC, lysoPC, and SM account for more than 85% of the phospholipids found in human plasma,12 which can be detected using a precursor ion scan of m/z 184 in positive ion electrospray.13 It is also known that these lipids are the major cause for significant LC-MS/MS matrix ionization effects in the positive ion electrospray mode.7,9-11,13 A data set for the removal of these phospholipids by monitoring precursor ions of m/z 184 in positive ion electrospray should give a good indication of sample cleanup. Throughout this article, PC was used to represent PC, lysoPC, and SM. The extraction solvents used for LLE were ethyl acetate and hexane. Hexane generally provides better extraction efficiency for nonpolar compounds, and ethyl acetate usually results in good extraction efficiency for polar compounds. Therefore, a mixture of ethyl acetate and hexane provides an extraction solvent system that covers a wide range of compounds.4,14 EXPERIMENTAL SECTION Chemical, Reagents, Materials, and Apparatus. Chemicals and Reagents. HPLC grade acetonitrile, methanol, isopropanol, ethyl acetate, hexane, and ACS grade ammonium acetate were purchased from J.T. Baker (Phillipsburg, NJ). Formic acid (SupraPur grade), acetic acid (ACS grade), and ammonium hydroxide (ACS grade) were purchased from EMD Chemicals (Gibbstown, NJ). Human K2EDTA plasma was obtained from Bioreclamation (Hicksville, NY). Ketoconazole was purchased (12) Phillips, G.; Dodge, J. J. Lipid Res. 1967, 8, 676–681. (13) Xia, Y.; Jemal, M. Rapid Commun. Mass Spectrom. 2009, 23, 2125–2138. (14) Rodila, R.; Kim, J.; Ji, Q.; El-Shourbagy, T. Rapid Commun. Mass Spectrom. 2006, 20, 3067–3075.

from Sigma-Aldrich (St. Louis, MO) and ketoconazole-d8 (internal standard) was purchased from United States Biological (Swampscott, MA). Other compounds were drug candidates in development at Bristol-Myers Squibb (Princeton, NJ). Materials. Vials used for standard and quality control samples were 2.0 mL polypropylene vials from VWR International (Bridgewater, NJ). Tubes used for liquid-liquid extraction were polypropylene MicroTubes from National Scientific (Claremont, CA). The caps used to seal the liquid-liquid extraction plate are Piercable TPE capband-8 cluster from Micronic (AR Lelystad, The Netherlands). Plates used for sample collection and injection were Costar Brand 96-well assay blocks from VWR International (Bridgewater, NJ). Apparatus. A Tomtec Quadra 96 model 320 robotic liquid handler, obtained from Tomtec (Hamden, CT), was used for adding organic solvent and transferring samples during sample preparation. An SPE Dry 96 sample concentrator, obtained from Jones Chromatography (Lakewood, CO), was used for solvent evaporation of liquid-liquid extracts. uHPLC-MS/MS Equipment. All sample analyses were performed on a Sciex API 4000 triple quadrupole mass spectrometer (Applied Biosystems/MDS SCIEX, Concord, Ontario, Canada), which was controlled by Analyst 1.4.2 software. This software was also used for data acquisition and processing. The mass spectrometer was directly coupled with a uHPLC system from Leap Technology (Carrboro, NC), which consisted of a Leap HTC-PAL autosampler, a Flux 4x Ultra mobile phase delivery pump, and a HotDog 5090 column heater from Prolab (Reinach, Switzerland). The separation was achieved on an Acquity UPLC BEH Shield RP18 column (1.7 µm, 2.1 mm × 50 mm) from Waters (Milford, MA). uHPLC-MS/MS Methods. The mobile phases consisted of 0.1% formic acid in water (A) and 0.1% formic acid in acetonitrile (B) for PC monitoring and recovery screening. The gradient for PC monitoring started with 40% B and stayed at 40% B for 0.1 min, then changed to 95% B in 0.05 min and stayed at 95% B for 4.3 min, then switched back to 40% B in 0.05 min and stayed at 40% B until the next injection. The gradient elution for analytes recovery screening started with 5% B and stayed at 5% B for 0.1 min, then increased B to 95% in 1.4 min and stayed at 95% B for 0.45 min, then switched back to 5% B in 0.05 min and stayed at 5% B until the next injection. The injection volume was 5 µL, and the flow rate was 0.8 mL/min for both PC monitoring and recovery screening. The detection was in positive ion electrospray mode by precursor ion scan (precursors of m/z 184 with a mass range from m/z 400 to m/z 1000) for PC monitoring and by selected reaction monitoring (SRM) for analyte recovery screenings. For PC monitoring, the collision induced dissociation (CID) energy was 35 eV, the electrospray voltage was 4500 V, and the probe temperature was maintained at 650 °C. The total run time was 5 min. For analytes recovery screening, the electrospray voltage was set at 1500 V and the probe temperature was 550 °C. Five compounds were monitored using this method, with a stable isotope labeled internal standard (IS) for each compound. There were 10 SRM transitions in the MS method and the dwell time was set as 5 ms for each transition. The total run time was 2.5 min. For the quantitative analysis of ketoconazole, the mobile phase consisted of water/acetonitrile (44/56, v/v) with 0.1%

ammonium hydroxide and 10 mM ammonium bicarbonate. Separation was achieved by isocratic elution. The injection volume was 5 µL, and the flow rate was 1.0 mL/min. The detection was by positive ion electrospray SRM (m/z 531 > 489 for ketoconazole and m/z 539 > 497 for ketoconazole-d8). The electrospray voltage was set at 4500 V, and the probe temperature was 550 °C. The CID energy was 43 eV, and the declustering potential (DP) was 120 V. The dwell time was set as 20 ms for each transition. The column temperature was maintained at 40 °C in all cases. Sample Preparation. PC Monitoring and Data Set Establishment. To prepare samples for the method development of PC monitoring, protein precipitated (PPT) plasma samples were first prepared. Specifically, 3 mL of human plasma were mixed with 9 mL of acetonitrile; then, the sample was centrifuged for 5 min at 3200 rpm. The supernatant from PPT was used to prepare serially diluted standards to establish standard curves using a solvent mixture of acetonitrile/water (20/80, v/v). Specifically, the relative PC content was assigned as 100% for undiluted supernatant from the PPT and 0.5% for a 200 times diluted PPT sample. To establish a data set for PC removal under different conditions, 3 extraction buffers (basic, 50 mM ammonium acetate in water with 1% ammonium hydroxide, approximately pH 10; neutral, 50 mM ammonium acetate in water, approximately pH 7; and acidic, 50 mM ammonium acetate in water with 1% formic acid, approximately pH 3) and 11 different extraction solvent combinations of hexane and ethyl acetate (100/0, 90/10, 80/ 20 . . . 20/80, 10/90, 0/100, v/v) were screened. A volume of 50 µL of human plasma were mixed with 50 µL of water, 100 µL of extraction buffer, and 600 µL of extraction solvent. The samples were shaken for 20 min and centrifuged at 3200 rpm for 5 min; then, 400 µL of supernatant was transferred into a collection plate, dried down under nitrogen flow, and reconstituted in 200 µL of acetonitrile/water (20/80, v/v) with 0.1% formic acid. Neat solution blanks were used to evaluate the background level for this method and served as 100% PC removal, which were used for background subtraction. Supernatant from protein precipitation samples (with final plasma amounts equivalent to the liquid-liquid extraction samples) served as 0% PC removal, which were used to normalize the final data. Analyte Recovery Screening. To demonstrate the feasibility that the recovery of multiple compounds can be done together, four randomly selected Bristol-Myers Squibb proprietary compounds were tested together with ketoconazole. Recovery samples were prepared by spiking five compounds into human plasma. These samples were screened for their recovery of all five compounds under the same conditions used for PC removal data set establishment (see PC Monitoring and Data Set Establishment). Specifically, 50 µL of the spiked human plasma (containing 500 ng/mL of ketoconazole) were put through the same extraction procedures as that used for creating the PC removal data set. After the samples were dried down, they were reconstituted using 200 µL of 0.1% formic acid in acetonitrile/water (20/80, v/v), which contained the stable labeled IS (50 ng/mL for ketoconazole-d8) for each compound. Assay Performance Evaluation for the Quantitation of Ketoconazole in Human Plasma. Calibration Standards (STD) and Quality Control (QC) Samples Preparation. Ketoconazole and ketoconazole-d8 stock solutions (1.0 mg/mL) were prepared in a mixture Analytical Chemistry, Vol. 81, No. 22, November 15, 2009

9227

of acetonitrile and water (50/50, v/v). A volume of 50 µL of the ketoconazole stock solution were added to 4.950 mL of control human plasma to make a plasma working solution. Eight standard levels, 1.00, 2.50, 10.0, 50.0, 100, 400, 750, and 1000 ng/mL, were prepared by appropriately diluting the working solution into the control human plasma. Six QC levels, 1.00, 3.00, 40.0, 500, 800, and 10 000 ng/mL (dilution QC), were prepared by adding the appropriate volume of working solution into control human plasma. Liquid-Liquid Extraction. A volume of 50 µL of STDs, QCs, testing samples, and blanks were mixed with 50 µL of internal standard working solution (100 ng/mL of ketoconazole-d8 in acetonitrile/water, 20/80, v/v), 100 µL of 50 mM ammonium acetate buffer with 1% ammonium hydroxide, and 600 µL of mixture solvent of hexane/ethyl acetate (50/50, v/v). The samples were shaken for 20 min and centrifuged at 3200 rpm for 5 min; then, 400 µL of supernatant was transferred into the collection plate, dried down under nitrogen flow, and reconstituted in 200 µL of acetonitrile/water (20/80, v/v) with 10 mM ammonium bicarbonate and 0.1% ammonium hydroxide. Incurred Samples Reanalysis and Assay Ruggedness Test. A total of 24 incurred samples (real samples from the in vivo study) were pooled and used to evaluate the reproducibility of this method. After the incurred sample reanalysis test was finished, the extracted samples were then used for the assay ruggedness test by sequentially injecting 500 injections into the uHPLC-MS/MS system. Data Processing. For PC monitoring, the entire trace of elution profiles from the LC-MS precursor ions scan of m/z 184 were manually integrated using Sciex Analyst software version 1.4.2 and the resulting area was used for quantitation. For analyte recovery screening and ketoconazole quantitation, the peak areas of all analytes and their IS were automatically determined using Sciex Analyst software version 1.4.2. For ketoconazole quantitation, a calibration curve was derived from the peak area ratios (ketonconazole/ketoconazole-d8) using weighted linear-squares regression of the area ratio versus the concentration of the standards. A weighting of 1/X2 (where X is the concentration of a given standard) was used for curve fitting. The regression equation for the calibration curve was used to back-calculate the measured concentration for each standard and QC. The results were compared to the theoretical concentration to obtain the accuracy, expressed as a percentage of the theoretical value, for each standard and QC. RESULTS AND DISCUSSIONS uHPLC-MS/MS Method for Monitoring PC. In order to establish a data set of PC removal under various sample cleanup conditions, a reliable assay which can assess the amount of PC is required. Reproducibility, lack of carryover, and throughput are the main criteria for assay performance. In most of the literature we found,5,10,11 the monitoring of the PC is performed using the same assay as for the analyte(s) of interest. Here we have developed a specific method to monitor PC to establish the data set. As described in uHPLC-MS/MS Methods, a uHPLC-MS/ MS method was developed to monitor PC. Total ion chromatograms for all precursor ions of m/z 184 over the range of m/z 400-1000 are shown in Figure 2. The majority of PC is eluted out within 3 min under the conditions used. The total elution time 9228

Analytical Chemistry, Vol. 81, No. 22, November 15, 2009

Figure 2. Total ion chromatograms for all precursor ions of m/z 184 from protein precipitated (PPT) human plasma samples. Precursor ions of m/z 184 were monitored over the range of m/z 400-m/z 1000. The entire trace was manually integrated for quantitation. (A) 100% supernatant from a PPT sample, (B) first blank sample (reconstitution solution only) right after the 100% PPT sample, (C) second blank sample after the 100% PPT sample (The peak at the end of traces in panels B and C is spiking noise by a sudden change of mobile phase.), and (D) 200 times diluted PPT sample.

Figure 3. Calibration curve for monitoring PC using precursor ion scan. The curve range is from 0.5% to 100% PPT plasma sample.

was 5 min. The sensitivity, carryover, and linearity of the assay used for monitoring PC were evaluated using the supernatant of plasma protein precipitation (PPT) as the reference standard solution. The total ion chromatogram for an undiluted PPT sample is shown in Figure 2A. The first blank (neat solution) injection after the PPT sample is shown in Figure 2B. The second blank injection after the injection of this sample is shown in Figure 2C. The total ion chromatogram for a sample with 200 times dilution of the PPT supernatant is shown in Figure 2D. The signal intensity (carryover from 2A) in Figure 2B is less than that in Figure 2D, which indicates that the overall carryover is less than 0.5%. Comparing the signal intensity in parts D and C of Figure 2, it is demonstrated that this assay has a good detection of PC even if there is only 0.5% of PC from the PPT supernatant sample remaining. Also, as shown in Figure 3, a good calibration curve over the range of 0.5% to 100% of PPT supernatant can be obtained by using serial dilution samples from the PPT sample with water/ acetonitrile (80/20, v/v). Although absolute quantitation is not necessary, a good calibration curve demonstrates that this assay should give a good evaluation of the PC removal. This method

Figure 4. PC removal profiles for two human plasma lots under different LLE conditions: (A) normal human plasma and (B) human plasma with high fat content. Numbers under the shadow area are multiplied by a factor of 10 for relative phospholipids response. For solvent combination, 100H0E represents hexane/ethyl acetate, 100/0, v/v; PPT represents supernatant of protein precipitation plasma sample. Please note that PPT sample was prepared without pH adjusted.

can be used to generate a data set of PC removal for various sample preparation conditions. PC Removal Data Set for LLE Using the Hexane/Ethyl Acetate Extraction Solvent System. We established a data set for PC removal using different ratios of the hexane and ethyl acetate at various pH conditions. The data set contains results for human plasma under 33 different LLE conditions. Two human plasma lots, one normal and one with high fat content (determined by visual observation), have been tested. The PC removal profiles are shown in Figure 4. The average peak area of three neat solution blank samples within a run served as the background (0%) for that run. All data points in that run have been subjected to background subtraction. The average peak area from three measurements of each lot of PPT plasma sample was used as a reference (100%) to normalize each LLE sample within each run. Each reported number is an average of three measurements (generated from three individual runs). The removal of PC mainly depends on the different combination of hexane and ethyl acetate for both lots of plasma. Generally speaking, the higher the percentage of hexane was, the more PC was removed from the final extracts. Notably, pH also plays an important role in removing PC. The higher the pH was, the more PC was removed from the final extracts. The largest pH effect was observed when 100% ethyl acetate was used. As expected, we did see slight differences in PC removal profiles for these two distinct lots of human plasma. However, the overall trend for both lots of plasma was the same. Under some conditions, there was very little PC remaining (e.g., less than 0.1% at pH 10 with 50/50 hexane/ethyl acetate combination, compared with the 100% PPT supernatant sample). It was noticed that the PC remaining in the high-fat plasma after LLE with 100% ethyl acetate at pH 3 is even higher than that in the PPT reference, which is unexpected. An experiment was carried out to examine the possible reasons behind this (please see details in the Supporting Information, section 1). It was concluded that the way of the referenced PPT sample was prepared accounted for this observation. The reference PPT sample used in this data set was prepared by commonly used PPT

procedures, mixing three volume of acetonitrile with one volume of plasma (organic to aqueous ratio is 3 to 1). Another PPT sample was prepared by mimicking the LLE procedures tested in this paper. A total of 50 µL of human plasma was mixed with 50 µL of water and 100 µL of ammonium acetate buffer (50 mM, pH ∼ 7) and then processed by PPT with 600 µL of acetonitrile (the organic to aqueous ratio was kept as 3 to 1, but the organic part to plasma ratio is now 12 to 1). A PPT sample generated by mimicking the LLE procedures gave much higher PC response than a regular PPT sample did. This is probably due to a higher organic to plasma ratio in the PPT sample mimicking the LLE process. Here we still use PPT generated from a commonly used procedure as a reference to have direct comparison between LLE and PPT. Also, this observation suggests that LLE is not always better than PPT regarding the removal of phospholipids. It is also worthwhile to mention that, ideally, a complete profile of the removal of all types of plasma phospholipids should be established to give a true representation of the sample cleanup. As described in a recent publication,13 it is possible to monitor all types of plasma phospholipids using LC-MS/MS. A precursor ion scan of m/z 184 in the positive ion mode can detect all of the PC, lysoPC, and SM phospholipids; a precursor ion scan of m/z 153 can detect all of the phosphatidic acid (PA), phosphatidylinositol (PI), phosphatidylserine (PS), phosphatidylglycerol (PG), PC, and PE phospholipids; and neutral loss of 141 Da in positive ion mode can detect all of the PE and lyso PE phospholipids. By combination of these three MS detection methods together, all phospholipids can be detected. As shown in the Supporting Information, section 2, experiments were carried out by monitoring a neutral loss of 141 Da in the positive ion mode. Similar results were obtained for PPT sample in our experiment to the results shown in ref 13. A close look at the responses for PPT samples from both ref 13 and our experiment reveals that the absolute response for the neutral loss scan of 141 Da is more than 100 times lower than that for the precursor ion scan of m/z 184. This is either because the amount of PE and lysoPE (detected by neutral loss scan of 141 Da) is much less than that of PC, lysoPC, and SM (detected by the precursor ion scan of m/z 184), or the Analytical Chemistry, Vol. 81, No. 22, November 15, 2009

9229

Figure 5. Recovery profiles for ketoconazole under different pH and extraction solvent combinations. The relative responses are normalized to IS response (example for solvent combination, 100H0E: hexane/ethyl acetate, 100/0, v/v).

sensitivity of the neutral loss scan of 141 Da is much lower than that for a precursor ion scan of m/z 184. Nonetheless, when applied to LLE extracted samples, the neutral loss scan of 141 Da did not generate meaningful information. For the negative ion scan of m/z 153, the experiment was not successful under our conditions. A close look at the data from ref 13 (Figure 5 in the reference), the phospholipids detected in PPT samples by using both negative precursor ion scan of m/z 153 and neutral loss scan of 141 are early eluting phospholipids and have the same elution time as some of the phospholipids detected in the precursor ion scan of m/z 184. Therefore, similar extraction behaviors are expected for those phospholipids detected by precursor ion m/z 153 and neutral loss scan of 141 Da to the phospholipids detected by the precursor ion scan of m/z 184. Importantly, since PC, lysoPC, and SM (detected by precursor ion scan of m/z 184) account for more than 85% of the phospholipids found in human plasma12 and they are the major cause for significant LC-MS/ MS matrix ionization effects in the positive ion electrospray mode,7,9-11,13 the data set for the removal of PC, lysoPC, and SM should be a fair representation of sample cleanup. It is known that glycerophospholipids will undergo basecatalyzed hydrolysis of the acyl chains.15 The hydrolysis process usually happened under harsh basic environment, for example, with sodium hydroxide present. In our experiment, ammonium hydroxide buffer with a pH around 10 was used for sample extraction. Although the risk of hydrolysis of the acyl chains on glycerophospholipids is expected to be low, an experiment was done (see details in the Supporting Information, section 3) to evaluate the pH effects on the profile of phospholipids monitored by our methods. No significant difference was observed for samples treated under different pH. Extraction Recovery of Ketoconazole and Selection of the Extraction Conditions. The extraction recoveries of ketoconazole along with four proprietary Bristol-Myers Squibb compounds were evaluated simultaneously. A uHPLC-MS/MS method was used for screening these compounds. A representative uHPLC-MS/ MS chromatogram is shown in the Supporting Information, section 4. All five compounds were well separated with a run time of 2.5 min. The recovery profile of ketoconazole at various LLE conditions is shown in Figure 5. Generally, the recovery of the ketoconazole was higher at higher pH for the same ratio of hexane and ethyl acetate. By comparison of the PC removal profiles in (15) Hubscher, G.; Hawthorne, J.; Kemp, P. J. Lipid Res. 1960, 1, 433–438.

9230

Analytical Chemistry, Vol. 81, No. 22, November 15, 2009

Figure 6. Representative uHPLC-MS/MS chromatograms of ketoconazole and its IS (ketoconazole-d8): (A) SRM chromatogram for ketoconazole in control plasma blank containing IS only, (B) SRM chromatogram for ketoconazole with the lower limit of quantitation at 1.0 ng/mL, and (C) SRM chromatogram for ketoconazole-d8 at 100 ng/mL.

Figure 4, a good condition (good recovery with minimum PC) at pH 10 with a hexane/ethyl acetate ratio of 50/50 can be easily identified. Less than 0.1% of the PC remained after extraction under these extraction conditions. Therefore, this extraction condition was chosen for further assay validation and evaluation. Simultaneously, recovery profiles for four proprietary BristolMyers Squibb compounds were obtained (data not shown). A good condition was also readily identified for each of these four compounds and successfully applied to method validation and sample analysis. uHPLC-MS/MS Method for the Quantitation of Ketoconazole in Human Plasma. With the knowledge that minimum PC remained (