Evaluation of a High-Throughput Online Solid Phase Extraction

Sep 21, 2011 - High throughput-solid phase extraction tandem mass spectrometry .... to facilitate very fast online SPE and an API4000 for mass analysi...
0 downloads 0 Views 1010KB Size
ARTICLE pubs.acs.org/ac

Evaluation of a High-Throughput Online Solid Phase Extraction Tandem Mass Spectrometry System for In Vivo Bioanalytical Studies Wenying Jian,*,† Michelle V. Romm,‡ Richard W. Edom,† Vaughn P. Miller,‡ William A. LaMarr,‡ and Naidong Weng† † ‡

Johnson & Johnson Pharmaceutical Research & Development, Raritan, New Jersey 08869, United States Agilent Technologies, Wakefield, Massachusetts 01880, United States

bS Supporting Information ABSTRACT: High throughput-solid phase extraction tandem mass spectrometry (HT-SPE/MS) is a fully automated system that integrates sample preparation using ultrafast online solid phase extraction (SPE) with mass spectrometry detection. HT-SPE/MS is capable of conducting analysis at a speed of 5 10 s per sample, which is several fold faster than chromatographically based liquid chromatography mass spectrometry (LC MS). Its existing applications mostly involve in vitro studies such as high-throughput therapeutic target screening, CYP450 inhibition, and transporter evaluations. In the current work, the feasibility of utilizing HT-SPE/MS for analysis of in vivo preclinical and clinical samples was evaluated for the first time. Critical bioanalytical parameters, such as ionization suppression and carry-over, were systematically investigated for structurally diverse compounds using generic SPE operating conditions. Quantitation data obtained from HT-SPE/MS was compared with those from LC MS analysis to evaluate its performance. Ionization suppression was prevalent for the test compounds, but it could be effectively managed by using a stable isotope labeled internal standard (IS). A structural analogue IS also generated data comparable to the LC MS system for a test compound, indicating matrix effects were also compensated for to some extent. Carry-over was found to be minimal for some compounds and variable for others and could generally be overcome by inserting matrix blanks without sacrificing assay efficiency due to the ultrafast analysis speed. Quantitation data for test compounds obtained from HT-SPE/MS were found to correlate well with those from conventional LC MS. Comparable accuracy, precision, linearity, and sensitivity were achieved with analysis speeds 20 30-fold higher. The presence of a stable metabolite in the samples showed no impact on parent quantitation for a test compound. In comparison, labile metabolites could potentially cause overestimation of the parent concentration if the ion source conditions are not optimized to minimize in-source breakdown. However, with the use of conditions that minimized in-source conversion, accurate measurement of the parent was achieved. Overall, HT-SPE/MS exhibited significant potential for highthroughput in vivo bioanalysis.

B

ioanalytical support is essential for in vivo preclinical and clinical studies throughout the entire drug discovery and development process. Concentration data for drugs and their metabolites in biological matrixes such as plasma, blood, and urine plays an important role in evaluation of pharmacokinetic parameters, safety assessment, and interpretation of efficacy and toxicological observations. The ever-increasing demand for bioanalysis has created opportunities for development of highthroughput bioanalytical approaches. Current approaches for improving method throughput include ultrahigh pressure liquid chromatography (UPLC),1 fused-core particle columns,2 highspeed monolithic chromatography,3 high-speed hydrophilic chromatography (HILIC),4,5 and multiplex chromatography.6 9 However, these approaches may suffer from one or more limitations such as sophisticated instrumentation, high solvent consumption, higher maintenance, or increased cost. Even with fast chromatography, analysis time still exceeds 1 min per sample in most cases. Recently, direct analysis in real time (DART) r 2011 American Chemical Society

ionization mass spectrometry was evaluated for quantitation of drugs in biological matrixes, and it demonstrated potential for high-throughput bioanalysis.10,11 However, this technique is still in its preliminary stage, and the instrumentation has not been adapted for routine bioanalytical support.11 Considerable interest still exists to develop new high-throughput technologies that can provide higher analysis speed. In many laboratories, especially start-ups in developing countries, the cost of purchasing mass spectrometers, the associated maintenance, and the depreciation are still the most expensive investments. A system suitable for rapid analysis of various analytes on a single mass spectrometer would require the least capital investment. HT-SPE/MS is a fully automated system developed in recent years to address the demand for high-speed therapeutic target Received: August 6, 2011 Accepted: September 21, 2011 Published: September 21, 2011 8259

dx.doi.org/10.1021/ac202017c | Anal. Chem. 2011, 83, 8259–8266

Analytical Chemistry screening in early drug discovery. It uses a microscale, reusable SPE system to perform sample extraction prior to analysis by atmospheric pressure ionization-tandem mass spectrometry (APIMS/MS).12 The HT-SPE/MS system was built by coupling specially designed equipment including an autosampler, LC pumps, SPE cartridge, and switching valves. In a typical analysis cycle, the samples are injected by a fast moving sample aspirator and loaded onto the SPE cartridge using an aqueous mobile phase, which washes away the nonretaining components such as salts, buffer, and proteins. The cartridge is then back-flushed using a relatively high organic content mobile phase to directly deliver the sample to the mass spectrometer for analysis. The entire cycle consisting of aspiration, load/wash, elution, and re-equilibrium typically takes 5 10 s, which offers several fold higher speed than all the aforementioned MS-based high-throughput analytical approaches. SPE cartridges packed with different materials including C18, C8, C4, phenyl, cyano, and HILIC, are commercially available to provide extractability for analytes of a broad array of chemical properties. The online SPE system can automatically switch between six cartridges without interrupting an injection sequence, which significantly improves the efficiency of method development and sample analysis. The HT-SPE/MS system is compatible with many types of in vitro samples, such as cell culture extracts and microsomal preparations. A generic protein precipitation is usually employed as a sample pretreatment. Currently, HT-SPE/MS has been successfully applied in various in vitro assays to support high-throughput screening12 18 and absorption, distribution, metabolism, and excretion (ADME) studies including cytochrome P450 (CYP) inhibition, 19,20 metabolic stability, 21 and transporter substrate or inhibitor screening. 22,23 In the current work, the feasibility of utilizing HT-SPE/MS for ultrafast analysis of in vivo samples was evaluated for the first time. The HT-SPE/MS system does not employ chromatographic separation of the eluent from online SPE, and a generic and nonselective protein precipitation method was used as sample pretreatment. Therefore, potential ion suppression caused by matrix components and dosing vehicles could impact the performance of in vivo sample analysis.24 In addition, carry-over from the online extraction cartridges, which are reused for multiple samples, may be exacerbated for in vivo samples, for which the concentrations may be present with a very large dynamic range. These key potential analytical liabilities were systematically investigated for structurally diverse compounds using generic conditions. More importantly, quantitation data for selected analytes were obtained from the HT-SPE/MS system and compared with those from conventional LC MS analysis to evaluate the analytical assay performance, including accuracy, precision, linearity, sensitivity, and potential interference from metabolites.

’ EXPERIMENTAL SECTION Materials. Control (blank) plasma was purchased from Bioreclamation (Hicksville, NJ). Test compounds were obtained from Sigma Chemicals (St. Louis, MO) except for 4-hydroxypropranolol, which was from CDN Isotopes Inc. (Pointe-Claire, Quebec, Canada). Stable isotope labeled test compounds were also from CDN Isotopes Inc. Proprietary compounds and their stable isotope labeled IS were synthesized inhouse at Johnson & Johnson Pharmaceutical Research & Development.

ARTICLE

Sample Preparation. The matrix effects of plasma was evaluated by comparing the analyte peak areas in samples postspiked with analyte to those in the neat solvent. The matrix effect of the dosing vehicle was evaluated by comparing the analyte peak areas in samples added with polyethylene glycol 400 (PEG400) (1%) to those without PEG400. The experiment details can be found in the Supporting Information. Carry-over was evaluated by comparing the analyte peak areas in blanks injected following high concentration samples with those in low concentration samples. The experiment details can be found in the Supporting Information. In the sample preparation for the diclofenac or fluoxetine assays, the calibration standards and quality control (QC) samples were prepared by serial dilution of the stock solution in rat plasma. The samples were prepared by a protein precipitation procedure. Detailed information can be found in the Supporting Information. The JNJ compound 1 study samples in dog plasma and the JNJ compound 2 study samples in human plasma were prepared by a protein precipitation procedure. Instrumentation. The instrumentation for conventional LC MS analysis can be found in the Supporting Information. The Agilent HT-SPE/MS system consisted of a RapidFire 300, which uses high-speed robotics and fast-switching valves to facilitate very fast online SPE and an API4000 for mass analysis. For the matrix effect evaluation and carry-over experiments as well as the fluoxetine assay, a C4 cartridge (Agilent Technologies, Inc., Santa Clara, CA) was used. The samples were loaded on to the SPE cartridge using a combination of 0.09% formic acid and 0.01% triflouroacetic acid in water (v/v/v) (mobile phase A, flow rate 1.5 mL/min) and eluted using a combination of 0.09% formic acid and 0.01% triflouroacetic acid in acetonitrile (v/v/v) (mobile phase B, flow rate 1.25 mL/min). The diclofenac assay and JNJ compound 1 used a C4 cartridge and JNJ compound 2 used a C18 cartridge (Agilent Technologies, Inc., Santa Clara, CA). The samples were loaded onto the cartridge using 0.1% formic acid in water (v/v) (mobile phase A, flow rate 1.5 mL/min) and eluted using 0.1% formic acid in acetonitrile (v/v) (mobile phase B, flow rate 1.25 mL/min). The sample injection needle was washed with water followed by acetonitrile (flow rate 1.25 mL/min). Aspiration time, wash time, elution time, and re-equilibration time were approximately 200, 3000, 3000, and 500 ms, respectively, for a total cycle time of around 8.5 s (JNJ compound 1 had a wash time of 4000 ms for a cycle time of around 9.5 s). All mass spectrometric parameters used were the same as the LC MS method (Supporting Information). The mass spectrometer was controlled by Analyst software (version 1.4.2, AB Sciex, Foster City, CA), and the RapidFire 300 system was controlled by RapidFire-MS Chromatography System software (version 2.0.20, Agilent Technologies Inc., Santa Clara, CA).

’ RESULTS AND DISCUSSION Even though the HT-SPE/MS system has been routinely used in high-throughput screening and in vitro ADME studies, its utilization for in vivo bioanalysis could be challenging due to the complicated biological matrixes and sample types. Biological matrixes such as plasma, urine, or whole blood contain a large amount of interfering components including salts, phospholipids, proteins, and possibly dosing vehicles. If not removed during sample extraction or not separated with chromatography, they may cause matrix effects in mass spectrometry.24 The samples for a HT-SPE/MS assay are generally prepared by a simple, generic 8260

dx.doi.org/10.1021/ac202017c |Anal. Chem. 2011, 83, 8259–8266

Analytical Chemistry

ARTICLE

protein precipitation procedure, extracted by the ultrafast online SPE, and back-flushed directly to the MS system for detection without chromatography. Potentially, a large amount of matrix interference could be retained on the SPE cartridge and then present in the ion source simultaneously with the analyte of interest, causing ion suppression. Another potential issue for the HT-SPE/MS system is carry-over, which has often been shown to be associated with online extraction systems with reusable cartridges or columns. In addition, metabolites may be present in the in vivo samples and cause ion suppression or in-source conversion to the parent drug. Without chromatographic separation, coeluting metabolites may compete for ionization with the parent drug and therefore cause ion suppression.25 More seriously, labile metabolites such as acyl-glucuronides or N-oxides can readily undergo in-source conversion to the parent drug leading to overestimation of parent concentrations.24 These are all the important bioanalytical aspects we evaluated for application of the HT-SPE/MS system for quantitative analysis of in vivo biological samples. Matrix Effects. To evaluate matrix effects on the HT-SPE/MS system, several test compounds covering a wide range of polarity and pKa were selected (Table 1). Rat plasma was processed by a simple protein precipitation procedure, postspiked with the test compound, and injected to the HT-SPE/MS system using a generic SPE operating condition. Correspondingly, neat solutions containing the compound at the same concentration were also

injected. For each type of sample, multiple injections (96 each) were made for detection of any signal drift caused by matrix interference. Figure 1 demonstrates as an example the peak area plot of propranolol and its stable isotope labeled analogue, D7-propranolol, in extracted rat plasma and neat solutions. First of all, it was found that signal intensity was highly consistent across the entire run for both the neat and plasma samples (% coefficient of variation (CV) < 2.5%). This indicated that injection volume and instrument response were very stable on HT-SPE/MS.

Figure 1. The peak area plot of propranolol and D7-propranolol in the matrix effect evaluation experiment.

Table 1. Matrix Effect Evaluation (a) Matrix Factors of Test Compounds in Rat Plasma (n = 96) analyte

stable isotope labeled analyte

propranolol (log P 3.35, pKa 9.5)

0.54

D7-propranolol

0.51

4-hydroxypropranolol (log P 2.26)a fluoxetine (log P 4.27, pKa 10.0)

0.41 0.40

D7-4-hydroxypropranolol D5-fluoxetine

0.41 0.39

norfluoxetine (log P 3.32)a

0.40

D5-norfluoxetine

0.41

diclofenac (log P 4.12, pKa 4.0)

0.27

D4-diclofenac

0.26

nifedipine (log P 2.91, pKa < 1.0)

0.57

NA

NA

nicotine (log P 0.93, pKa 8.5)

0.31

NA

NA

tolbutamide (log P 2.13, pKa 8.7)

0.41

NA

NA

alprenolol (log P 2.75, pKa 9.5)

0.68

NA

NA

(b) Matrix Factors of Test Compounds in Different Lots of Rat Plasma (n = 96) matrix factor

ratio analyte/internal standard

propranolol

D7-propranolol

tolbutamide

propranolol/D7-propranolol

propranolol/tolbutamide

lot 1

0.77

0.70

0.69

1.10

1.12

lot 2

0.72

0.65

0.73

1.10

0.99

lot 3

0.73

0.68

0.73

1.08

1.01

lot 4

0.75

0.67

0.59

1.12

1.28

lot 5 lot 6

0.72 0.71

0.65 0.65

0.61 0.77

1.10 1.09

1.17 0.92

% CV

3.07

3.10

10.49

1.21

12.28

(c) Matrix Factors of Test Compounds in Dosing Vehicle PEG400 (n = 96) PEG400 rat plasma

a

total

diclofenac

0.31

0.27

0.084

D4-diclofenac tolbutamide

0.31 0.20

0.26 0.41

0.081 0.082

pKa values not available. 8261

dx.doi.org/10.1021/ac202017c |Anal. Chem. 2011, 83, 8259–8266

Analytical Chemistry Also, there was no cumulative effect of matrix suppression on the online SPE system under the current operating conditions. Second, the signal intensities in the plasma samples were about half of those in the neat solutions, demonstrating significant ion suppression. More importantly, the response of the stable isotope labeled compounds tracked the nonlabeled compounds very well, implying that ion suppression could be compensated for by using a stable isotope labeled IS. The matrix factor calculated by comparing the average peak area of the 96 injections of postspiked samples to that of the neat samples was 0.54 for propranolol and 0.51 for D7-propranolol, sufficiently close to demonstrate comparable matrix effects (Table 1a). Matrix factors determined for the other compounds and their stable isotope labeled analogues ranged from 0.26 to 0.68 (Table 1a). The % CV of the peak area for the 96 injections of each type of samples was within 8.2% for all the test compounds, which showed that the experiment was highly reproducible. The stable isotope labeled compounds exhibited an almost identical matrix factor to their nonlabeled analogues. As all the above experiments were conducted using a single lot of rat plasma under identical instrument operating conditions, the difference in the matrix factor was solely due to the chemical nature of the compounds. Matrix factor in different lots of matrix was also evaluated for selected compounds, including propranolol, D7-propranolol, and tolbutamide. As demonstrated in Table 1b, the matrix factor in six different lots of rat plasma was determined to be highly consistent for propranolol and D7-propranolol, showing a % CV of 3.07% and 3.10%, respectively. In comparison, tolbutamide experienced higher variability, showing a % CV of 10.49%. For a bioanalytical assay, the effectiveness of the IS to manage variability in different matrix lots can be reflected by the % CV of the ratio of analyte/IS matrix factor. It was found that very little variability (1.21%) was detected when D7-propranolol was designated as IS for propranolol, while tolbutamide as an IS gave much higher variability (12.28%). This experiment demonstrated that on the HT-SPE/MS system, the variability of ion suppression caused by different matrix lots is compound-dependent, and a stable isotope labeled IS can track analyte response much better than a structural analogue. Variability in response between different lots can cause significant errors in quantitation. Also evident in the data from parts a and b of Table 1, the matrix factor can change from experiment to experiment. For example in Table 1a, propranolol, D7-propranolol, and tolbutamide all produced significantly different matrix factors than in Table 1b. Although consistent within the day or the experiment, the matrix factor was found to change when a new SPE cartridge was installed. Therefore, having a reliable IS to correct for matrix effects seems to be critical to the success of HT-SPE/MS. Dosing vehicles such as PEG400, Tween 80, propylene glycol, and hydroxypropyl-β-cyclodextrin, are often used in preclinical formulations and may cause severe ion suppression in MS-based assays. PEG400 often exhibits the most significant suppression for intravenous (IV) dosed samples from rodents.26 28 In the current study, PEG400 was chosen as a representative reagent to elucidate the potential impact of dosing vehicles on HT-SPE/MS assays. The selected test compound was spiked into rat plasma in the presence or absence of PEG400 (1%), subjected to a protein precipitation procedure, and injected to the HT-SPE/MS system. The matrix factor caused by PEG400 was calculated by comparing the analyte peak area in the presence versus that in the absence of PEG400. As shown in Table 1c, the matrix factor from PEG400 was determined to be 0.31 for both diclofenac and

ARTICLE

D4-diclofenac and 0.20 for tolbutamide. The low values suggest that PEG400 was not effectively removed by the online SPE, resulting in significant suppression to the test compounds. The total matrix factor caused by PEG400 in plasma can be estimated by multiplying the matrix factor of plasma by that of PEG400. It was found that the matrix factors were about 0.08 for all three test compounds, meaning that over 90% of the original signal could be lost due to suppression. Overall, the results of the matrix effects evaluation indicated insufficient removal of the matrix interference by the online extraction system on HT-SPE/MS and significant ion suppression for most of the compounds when using generic sample preparation and online SPE conditions. The impact of strong matrix effects on assay sensitivity and quantitation limits will need to be taken into consideration when developing a HT-SPE/MS assay. It also needs to be emphasized that generic SPE operating conditions were used in the current evaluation to minimize method development. For analysis of a specific analyte, several variables such as loading conditions, eluting conditions, and cartridge types could be explored in order to achieve more effective removal of the interfering components while retaining the analyte of interest. In addition, more elaborate off-line sample preparation such as liquid liquid extraction procedures for targeted removal of phospholipids could provide better sample extraction. In the current evaluation, we presented the worst case scenario that requires the least amount of method development effort. Very importantly, stable isotope labeled compounds exhibited an almost identical matrix factor as their nonlabeled analogue in all of the above experiments, which means they can effectively compensate for ion suppression when used as the IS, leading to potentially acceptable assay performance even in the presence of strong matrix effects. Carry-Over. Carry-over could be very significant in an online extraction system due to reuse of the SPE cartridge and the high aqueous content of the loading mobile phase. To evaluate carryover on the HT-SPE/MS system, plasma samples of the test compounds at a relatively high concentration (5000 ng/mL) were injected in 10 replicates, followed by 10 injections of blank samples, using a generic operating condition. The analyte peak area observed in each of the blank samples was compared to the mean peak area of 10 injections of low-concentration plasma samples (10 ng/mL) that were injected before the high-concentration samples. Any value higher than 20% would be considered significant carry-over. Among the nine test compounds, five of them demonstrated no carry-over at all. The other four compounds showed carry-over of different extents, as listed in Table 2. 4-Hydroxypropranolol showed a minor peak (16.5% of the mean of the low concentration samples) in the first blank, and the peak soon decreased and disappeared in the subsequent injections. The other three compounds diclofenac, nicotine, and tolbutamide demonstrated a somewhat decreasing trend as more blanks were injected. However, by the tenth blank, they still showed peaks higher than 20% of the low-concentration samples. This evaluation result demonstrated that carry-over on HTSPE/MS systems could be minimal for some compounds even when a generic SPE operating condition is used. By using generic conditions, we presented the worst case scenario where carry-over was manageable for most of the test compounds, even without extensive method optimization. Because of the fast injection cycle, extra blanks can be inserted after the high concentration samples to manage the carry-over without adding much extra analysis time. Also, for study sample analysis, careful monitoring and a strategic order of injection will be needed, similar to conventional LC MS 8262

dx.doi.org/10.1021/ac202017c |Anal. Chem. 2011, 83, 8259–8266

Analytical Chemistry

ARTICLE

Table 2. Carry-Over of Test Compounds in Blanks Following Injection of High Concentration Samples (% of the Mean Peak Area of 10 Injections of 10 ng/mL Samples) blank no.

4-hydroxypropranolol

diclofenac

nicotine

tolbutamide

1

16.5

206.1

561.2

59.5

2

3.9

176.4

434.3

31.3

3

2.4

125.9

306.2

19.5

4

0.1

82.0

215.6

36.7

5 6

0.0 0.0

63.5 115.6

172.0 132.6

37.7 29.8

7

0.0

136.2

104.6

36.4

8

0.0

33.9

80.4

29.9

9

0.0

27.4

74.4

31.4

10

0.0

22.0

54.7

33.9

assays.29 For certain compounds that tend to show high carryover, the loading conditions, eluting conditions, and SPE cartridge chemistry could be optimized to potentially reduce the retention of residual analytes on the cartridge and divert valves. For compounds that show persistent carry-over, it would be more appropriate to analyze them using regular LC MS or a separate HT-SPE/MS system with more sophisticated and compoundspecific washing. Bioanalytical Assay Performance. To evaluate the feasibility of HT-SPE/MS for in vivo bioanalysis in drug discovery and development, the system was used to analyze both spiked samples and study samples, and the results were compared to those obtained by LC MS methods. Important aspects of bioanalytical assay performance such as accuracy, precision, linearity, sensitivity, and potential interference from metabolites were evaluated. Diclofenac Assay. The main purpose of the current experiment was to evaluate the effectiveness of using a stable isotope labeled IS to manage the impact of matrix effects on a HT-SPE/ MS assay. Calibration standard and QC samples were prepared by spiking diclofenac into rat plasma. The samples were extracted using a protein precipitation procedure, followed by analysis using LC MS and HT-SPE/MS, respectively. The stable isotope labeled analogue, D4-diclofenac, was used as the IS. Each chromatographic peak was 1 2 s wide, and the cycle time was within 9 s. In comparison, it took about 3.5 min (3 min analysis time, plus 0.5 min injection time) to complete each injection on a conventional LC MS system. Overall, the analysis speed was about 20 times faster on the HT-SPE/MS. A lower limit of quantitation (LLOQ) of 10 ng/mL was obtained on both systems. The calibration curve constructed from HTSPE/MS data demonstrated excellent linearity (r2 = 0.9963), comparable to that from LC MS analysis (r2 = 0.9986). The accuracy and precision data are presented in Table 3 (upper). The % bias obtained from the HT-SPE/MS system for the three levels of QC was within (4.9%, and the % CV was within 14.25%, largely comparable to that obtained using LC MS. The low QC demonstrated higher variability on HT-SPE/MS than the conventional LC MS system, indicating that the coeluting matrix may have more impact on the ruggedness of the assay when the analyte concentration is lower. The results suggested that satisfying the bioanalytical performance could be obtained on a HT-SPE/MS system even in the presence of matrix suppression when a stable isotope labeled IS was used. Fluoxetine Assay. Coeluting metabolites may compete with the parent drug for ionization and potentially cause interference.

Table 3. Comparison of Accuracy and Precision of QC Samples of Test Compounds in Rat Plasma mean accuracy (%)

% CV

% bias

Diclofenac QC (n = 6) low QC

LC MS

104.7

3.18

4.7

mid QC

HT-SPE/MS LC MS

100.4 95.1

14.25 2.57

0.4 4.9

HT-SPE/MS

103.0

7.98

3.0

high QC

LC MS

95.2

4.29

4.8

101.6

4.47

1.6

HT-SPE/MS

Fluoxetine QC (Containing Norfluoxetine at 5000 ng/mL) (n = 6) low QC

LC MS

104.4

2.21

4.4

HT-SPE/MS

102.9

10.51

2.9

99.1

1.71

0.9

mid QC

LC MS

high QC

HT-SPE/MS LC MS

102.1 100.5

2.31 2.12

2.1 0.5

HT-SPE/MS

102.9

3.16

2.9

To evaluate the impact of a metabolite on parent drug quantitation on the HT-SPE/MS system, fluoxetine was chosen. Calibration standard and QC samples were prepared by spiking fluoxetine into rat plasma. The des-methyl metabolite of fluoxetine, norfluoxetine, was spiked into all levels of the QC samples at a relatively high concentration (5000 ng/mL). On the conventional LC MS system, there was baseline chromatographic separation of fluoxetine and norfluoxetine, and excellent bioanalytical assay performance for quantitation of fluoxetine was obtained (Table 3, lower). On the HT-SPE/MS system, coelution of the metabolite with the parent drug had minimal impact on the assay, and similar performance was achieved except for a little higher variability with the low QC level (Table 3). Again, similar sensitivity and linearity and much higher speed were achieved. Part of the reason for the absence of interference could be attributed to use of stable isotope labeled fluoxetine as the IS, which could compensate for ion suppression caused by the coeluting metabolite, if there was any. JNJ-Compound 1. For early stage preclinical studies, stable isotope labeled ISs are usually not available because they are expensive and time-consuming to synthesize. Instead, a structural analogue or a default compound is often used as IS. In the current evaluation, study samples from a dog toxicological study were assayed using a structural analogue IS. The concentrations measured on the HT-SPE/MS system were compared to those from a conventional LC-MS system. Figure 2 demonstrates the concentration time profile obtained from both systems for two dogs given low and high doses. There was an excellent match between the profiles obtained from the two systems, and the highest difference between HT-SPE/MS data and LC-MS data was within (15%. It was speculated that even though the matrix factor of the structural analogue IS may most likely be different from that of the analyte, satisfying quantitation results could still be achieved as long as the matrix factor ratio of the two remains relatively constant. In addition, complete coelution of the IS with the analyte on a HT-SPE/MS system subjected them to an identical environment of matrix interference and may have contributed to the effectiveness of the structural analogue IS to compensate for suppression. Overall, even though structural analogue compounds demonstrated larger variability for compensation of matrix effects in different lots 8263

dx.doi.org/10.1021/ac202017c |Anal. Chem. 2011, 83, 8259–8266

Analytical Chemistry

ARTICLE

Figure 2. Plasma concentration time profiles of JNJ-compound 1 in a dog study obtained using an LC MS system vs an HT-SPE/MS system: (A) dog 1 at low dose, (B) dog 1 at high dose, (C) dog 2 at low dose, and (D) dog 2 at high dose.

of matrix as demonstrated in the previous experiment, they could still generate HT-SPE/MS data comparable to those from a LCMS system. JNJ-Compound 2. Co-eluting labile metabolites, such as acylglucuronides and N-oxides, could potentially undergo in-source conversion to the parent drug and cause overestimation of the parent concentration. To evaluate the impact of a labile metabolite on HT-SPE/MS assays, JNJ compound 2 was selected for testing because it was known to have significant plasma levels of an acyl-glucuronide. Samples from a clinical study were assayed using an LC-MS system on which there was complete chromatographic separation of the acyl-glucuronide from the parent drug. The same set of samples was then analyzed on an HT-SPE/MS system where they coeluted. Previous studies have shown that the most critical factor for determining the extent of in-source conversion is DP.30 It was found that by decreasing DP from 70 to 30 V, the conversion of the acyl-glucuronide metabolite back to JNJ compound 2 was reduced from 12% to 1.6%. As demonstrated in Figure 3, the concentrations of JNJ compound 2 determined on an HT-SPE/MS system under DP of 30 coincided reasonably well (within (20%) with those from the LC-MS assay, while those determined under DP of 70 were significantly higher (>20%). The results of this study indicated that potential overestimation of parent drug concentrations caused by coeluting labile metabolites on HT-SPE/MS can be significant, but can be managed by optimizing the ion source conditions, especially DP. However, we need to be mindful when labile metabolites are present in much higher concentrations than the parent drug, or have a much higher extent of in-source conversion, even minor breakdown could cause a significant impact on the parent drug profile. Under these circumstances, HT-SPE/MS may not be a good choice.

’ CURRENT LIMITATIONS AND POTENTIAL APPLICATIONS It is important to be aware of the limitations of the current HT-SPE/MS platform and also to recognize the potential

opportunities provided by the ever-evolving analytical technologies to overcome these limitations. First of all, because of the lack of chromatographic separation, isobaric compounds cannot be differentiated on a HT-SPE/MS system that is coupled with a regular mass spectrometer. However, utilization of a high-resolution mass spectrometer such as time-of-flight (TOF) and Orbitrap MS provides much better capability to resolve isobaric compounds consisting of different elemental compositions. For molecules which have identical molecular formulas, such as chiral compounds or metabolites which are positional isomers (e.g., hydroxy), mass spectrometers equipped with an ion mobility analyzer could potentially separate the gas phase ions based on their different configuration.31,32 Another potential significant issue rising from the lack of separation is in-source breakdown of labile metabolites such as acyl-glucuronides, N-oxides, and sulfate conjugates, which may lead to overestimation of the parent drug concentration. As demonstrated in the previous example, adjustment of ion source parameters, especially the DP, could minimize the conversion. Because of inadequate removal of interferences during the SPE process, and lack of separation from the analyte, ionization suppression caused by biological matrixes or dosing vehicles (especially for rodent studies) could be significant on an HT-SPE/MS system using a nonoptimized, generic method intended for rapid analysis of most analytes. Nevertheless, even with decreased sensitivity, the accuracy and precision of the assay can usually be maintained by using a stable isotope labeled analogue as the IS, which demonstrated adequate compensation for the ion suppression. The approaches to overcome ion suppression and to improve the sensitivity include optimization of the SPE procedure for more complete removal of the matrix components and better selectivity for the analyte, and the use of atmospheric pressure chemical ionization (APCI) mode of ionization which has been demonstrated to be less subject to ion suppression.33,34 Alternatively, the HT-SPE/MS system could potentially be coupled with nanospray mass spectrometry for reduced ion suppression. It has been demonstrated that ion suppression in mass spectrometry was practically absent at low nanoliter/minute LC flow rates.35 If the design of 8264

dx.doi.org/10.1021/ac202017c |Anal. Chem. 2011, 83, 8259–8266

Analytical Chemistry

ARTICLE

readily available. Therefore, the high throughput provided by HT-SPE/MS can be fully realized. Finally, in the postmarketing stage, therapeutic drug monitoring could potentially be conducted by HT-SPE/MS based on a thorough understanding and investigation of the impact of comedications on the HT-SPE/MS assay for the analyte of interest.

Figure 3. Plasma concentration time profile of JNJ-compound 2 in a clinical study obtained using an LC MS system and an HT-SPE/MS system at a DP of 30 and 70 V, respectively.

the HT-SPE/MS system could be modified to accommodate nano flows, the issue of ion suppression could be minimized. Overall, to be certain that HT-SPE/MS assays for in vivo bioanalytial analysis will provide comparable data to LC MS assays, it will involve more up-front method development than that is needed for a conventional LC MS assay, which includes thorough investigation of matrix suppression, metabolite profiles, potential isobaric interferences, and carry-over. Undoubtedly, the full advantage of the high speed of HT-SPE/MS will be for potential analytes with the following characteristics: (1) no isobaric metabolites (if quantification is needed); (2) no labile metabolites that could break down in the ion source significantly; (3) no significant and persistent carry-over; and (4) manageable sensitivity loss due to ion suppression. Under these circumstances, HT-SPE/MS would allow minimal efforts in method development and changeover, and rapid analysis of a large numbers of compounds without delay. For compounds that do not meet the above criteria, LC MS methods will be needed. On the other hand, when thorough HT-SPE/MS method development is required, an advantage is the ultrafast analysis speed, which will give the results rapidly and thus improve the efficiency of method development significantly. Experiments can be conducted and interpreted very quickly. Given the unique advantages and some limitations, we think the HT-SPE/MS system could potentially be a useful tool for high-throughput MS quantitation of in vivo samples. In the early stages of drug discovery, a large number of candidates need to be dosed to animals and analyzed for their pharmacokinetic properties. The ultimate goal is to rapidly screen for “drug-like” characteristics in compounds, rank-order them, and quickly eliminate the poor candidates. A HT-SPE/MS system operated under generic conditions could provide the required throughput with tolerable risk of having issues such as interference or low sensitivity. In contrast, the HT-SPE/MS system may not be suitable for early drug development studies such as good laboratory practice (GLP) toxicology studies because the priority of these studies is not high throughput, and knowledge about the compound, especially the metabolite profile, may not be sufficient to judge if a compound is suitable for HT-SPE/MS analysis. In late-phase clinical studies, and clinical diagnostic studies, large sets of samples and the requirement for efficient analysis at lower cost justifies the up-front burden of method development and investigation on HTSPE/MS. At this stage, the analyte and the metabolite profile are usually well characterized, and the stable isotope labeled IS is

’ CONCLUSIONS In the current study, the HT-SPE/MS system was carefully evaluated for application with in vivo bioanalysis for the first time. Potential analytical liabilities, such as ion suppression and carryover, were systematically investigated using structurally diverse compounds. Important aspects of bioanalytical assay performance including accuracy, precision, linearity, sensitivity, and potential interference from metabolites were evaluated for selected compounds and compared to results from a conventional LC MS system. Generic SPE conditions were employed in the current evaluation to demonstrate the feasibility of using HT-SPE/MS for bioanalysis without extensive method development, which is the normal situation in high-throughput bioanalytical laboratories. The assay results from HT-SPE/MS were found to correlate well with those from a conventional LC MS system, and comparable accuracy, precision, linearity, and sensitivity were achieved. The analysis speed was 20 30 times higher on the HT-SPE/MS than LC MS. Overall, the results demonstrated HT-SPE/MS is potentially a valuable tool for in vivo highthroughput bioanalysis in different stages of the drug discovery and development process. ’ ASSOCIATED CONTENT

bS

Supporting Information. Additional information as noted in text. This material is available free of charge via the Internet at http://pubs.acs.org.

’ AUTHOR INFORMATION Corresponding Author

*Address: 1000 Route 202 South, Raritan, NJ 08869. Phone: 908-927-6584. Fax: 908-541-0422. E-mail: [email protected]. Notes

Authors Wenying Jian, Richard W. Edom, and Naidong Weng have no financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. Authors Michelle V. Romm, Vaughn P. Miller, and William A. LaMarr are all employees of Agilent Technologies, the manufacturer of the Agilent HT-SPE/MS system.

’ ACKNOWLEDGMENT No funding was provided by either Johnson & Johnson Pharmaceutical Research & Development or Agilent Technologies. ’ REFERENCES (1) Mazzeo, J. R.; Neue, U. D.; Kele, M.; Plumb, R. S. Anal. Chem. 2005, 77, 460 A–467 A. (2) Cunliffe, J. M.; Maloney, T. D. J. Sep. Sci. 2007, 30, 3104–3109. (3) Wu, R.; Hu, L.; Wang, F.; Ye, M.; Zou, H. J. Chromatogr., A 2008, 1184, 369–392. (4) Shou, W. Z.; Chen, Y. L.; Eerkes, A.; Tang, Y. Q.; Magis, L.; Jiang, X.; Naidong, W. Rapid Commun. Mass Spectrom. 2002, 16, 1613–1621. 8265

dx.doi.org/10.1021/ac202017c |Anal. Chem. 2011, 83, 8259–8266

Analytical Chemistry (5) Hsieh, Y.; Galviz, G.; Zhou, Q.; Duncan, C. Rapid Commun. Mass Spectrom. 2009, 23, 1364–1370. (6) Korfmacher, W. A.; Veals, J.; Dunn-Meynell, K.; Zhang, X.; Tucker, G.; Cox, K. A.; Lin, C. C. Rapid Commun. Mass Spectrom. 1999, 13, 1991–1998. (7) Hsieh, Y.; Korfmacher, W. A. Curr. Drug Metab. 2006, 7, 479–489. (8) Xu, R. N.; Fan, L.; Rieser, M. J.; El-Shourbagy, T. A. J. Pharm. Biomed. Anal. 2007, 44, 342–355. (9) Jian, W.; Edom, R. W.; Xu, Y.; Weng, N. J. Sep. Sci. 2010, 33, 681–697. (10) Zhao, Y.; Lam, M.; Wu, D.; Mak, R. Rapid Commun. Mass Spectrom. 2008, 22, 3217–3224. (11) Yu, S.; Crawford, E.; Tice, J.; Musselman, B.; Wu, J. T. Anal. Chem. 2009, 81, 193–202. € (12) Ozbal, C. C.; LaMarr, W. A.; Linton, J. R.; Green, D. F.; Katz, A.; Morrison, T. B.; Brenan, C. J. H. Assay Drug Dev. Technol. 2004, 2, 373–382. € (13) Forbes, C. D.; Toth, J. G.; Ozbal, C. C.; Lamarr, W. A.; Pendleton, J. A.; Rocks, S.; Gedrich, R. W.; Osterman, D. G.; Landro, J. A.; Lumb, K. J. J. Biomol. Screening 2007, 12, 628–634. € (14) Quercia, A. K.; Lamarr, W. A.; Myung, J.; Ozbal, C. C.; Landro, J. A.; Lumb, K. J. J. Biomol. Screening 2007, 12, 473–480. (15) Shiau, A. K.; Massari, M. E.; Ozbal, C. C. Comb. Chem. High Throughput Screeening 2008, 11, 231–237. (16) Soulard, P.; McLaughlin, M.; Stevens, J.; Connolly, B.; Coli, R.; Wang, L.; Moore, J.; Kuo, M. S.; LaMarr, W. A.; Ozbal, C. C.; Bhat, B. G. Anal. Chim. Acta 2008, 627, 105–111. (17) Holt, T. G.; Choi, B. K.; Geoghagen, N. S.; Jensen, K. K.; Luo, Q.; LaMarr, W. A.; Makara, G. M.; Malkowitz, L.; Ozbal, C. C.; Xiong, Y.; Dufresne, C.; Luo, M. Assay Drug Dev. Technol. 2009, 7, 495–506. (18) Langsdorf, E. F.; Malikzay, A.; Lamarr, W. A.; Daubaras, D.; Kravec, C.; Zhang, R.; Hart, R.; Monsma, F.; Black, T.; Ozbal, C. C.; Miesel, L.; Lunn, C. A. J. Biomol. Screening 2010, 15, 52–61. (19) Brown, A.; Bickford, S.; Hatsis, P.; Amin, J.; Bell, L.; Harriman, S. Rapid Commun. Mass Spectrom. 2010, 24, 1207–1210. € (20) Lim, K. B.; Ozbal, C. C.; Kassel, D. B. J. Biomol. Screening 2010, 15, 447–452. (21) Luippold, A. H.; Arnhold, T.; J€org, W.; S€ussmuth, R. D. Int. J. Mass Spectrom. 2010, 296, 1–9. (22) Wagner, A. D.; Kolb, J. M.; Ozbal, C. C.; Herbst, J. J.; Olah, T. V.; Weller, H. N.; Zvyaga, T. A.; Shou, W. Z. Rapid Commun. Mass Spectrom. 2011, 25, 1231–1240. (23) Luippold, A. H.; Arnhold, T.; Jorg, W.; Kruger, B.; Sussmuth, R. D. J. Biomol. Screening 2011, 16, 370–377. (24) Jemal, M.; Ouyang, Z.; Xia, Y. Q. Biomed. Chromatogr. 2010, 24, 2–19. (25) Remane, D.; Wissenbach, D. K.; Meyer, M. R.; Maurer, H. H. Rapid Commun. Mass Spectrom. 2010, 24, 859–867. (26) Tong, X. S.; Wang, J.; Zheng, S.; Pivnichny, J. V.; Griffin, P. R.; Shen, X.; Donnelly, M.; Vakerich, K.; Nunes, C.; Fenyk-Melody, J. Anal. Chem. 2002, 74, 6305–6313. (27) Shou, W. Z.; Naidong, W. Rapid Commun. Mass Spectrom. 2003, 17, 589–597. (28) Xu, X.; Mei, H.; Wang, S.; Zhou, Q.; Wang, G.; Broske, L.; Pena, A.; Korfmacher, W. A. Rapid Commun. Mass Spectrom. 2005, 19, 2643– 2650. (29) Hughes, N. C.; Wong, E. Y.; Fan, J.; Bajaj, N. AAPS J. 2007, 9, E353–360. (30) Yan, Z.; Caldwell, G. W.; Jones, W. J.; Masucci, J. A. Rapid Commun. Mass Spectrom. 2003, 17, 1433–1442. (31) Hatsis, P.; Kapron, J. T. Rapid Commun. Mass Spectrom. 2008, 22, 735–738. (32) Enders, J. R.; McLean, J. A. Chirality 2009, 21 (Suppl 1), E253–264. (33) Matuszewski, B. K.; Constanzer, M. L.; Chavez-Eng, C. M. Anal. Chem. 1998, 70, 882–889.

ARTICLE

(34) Schuhmacher, J.; Zimmer, D.; Tesche, F.; Pickard, V. Rapid Commun. Mass Spectrom. 2003, 17, 1950–1957. (35) Schmidt, A.; Karas, M.; Dulcks, T. J. Am. Soc. Mass Spectrom. 2003, 14, 492–500.

8266

dx.doi.org/10.1021/ac202017c |Anal. Chem. 2011, 83, 8259–8266