4 3 0 A A N A LY T I C A L C H E M I S T R Y / A U G U S T 1,

Transcription

1 full page ad

2 4 3 A A N A LY T I C A L C H E M I S T R Y / A U G U S T 1, 2 1

3 Systematic LC/MS Metabolite Identification in Drug Discovery A four-step strategy to characterize metabolites by LC/MS techniques early in the pharmaceutical discovery process. he study of how a drug is absorbed, distributed, metabolized, and eliminated by the body is a vital but costly and time-consuming step in the drug discovery process. Metabolism can dictate the rate of absorption into the body, lead to the production of new and possibly toxic species, or activate the drug (1 6). For example, morphine s effect is primarily due to one of its glucuronide conjugate metabolites. Nigel J. Clarke, Diane Rindgen, Walter A. Korfmacher, and Kathleen A. Cox Schering-Plough Research Institute A U G U S T 1, 2 1 / A N A LY T I C A L C H E M I S T R Y A

4 Precursor ion scan (based on fragmentation data collected from standard) Sample arrives (bile/urine) Triple quadrupole Tandem MS/MS Product ion scan (for all possible metabolites identified from above plus expected metabolites) Quadrupole ion trap (sequential tandem MS) MS/MS, MS/MS/MS, or greater (for all possible metabolites detected in above steps; provides detailed structural information) Accurate mass MS or MS/MS (if necessary; provides accurate masses for empirical formula determination) Until recently, metabolite identification only took place once a compound had been chosen for drug development. However, the discovery of a toxic metabolite can set a research program back significantly. As a result, many pharmaceutical companies are now conducting metabolite identification studies in the early phases of drug candidate selection. But this is no easy task. The metabolism of a drug within a test animal can be extremely complex, involve multiple enzymatic pathways, and lead to a range of compounds with varying concentrations (7). Other drugs have one or two major metabolic pathways that dominate their metabolism, but several minor pathways can produce at least one metabolite. We have seen >3 detectable, and theoretically identifiable, metabolites form from a single compound. MS has emerged as an ideal technique for the identification of such structurally diverse metabolites. When coupled with on-line HPLC, the technique is extremely robust, rapid, sensitive, and Constant neutral loss (for common conjugates such as glucuronide or sulfate) FIGURE 1. Systematic approach to metabolite identification and characterization. easily automated (1, 2, 8 11). Not surprisingly, LC/MS and LC/MS/MS have become the methods of choice for pharmacokinetic studies, yielding concentration versus time data for drug compounds from in vivo samples such as plasma (1,2). Nevertheless, identifying meta bolites remains a time-consuming process because a range of instrumental techniques and software applications are needed to obtain the appropriate data. The analyst must be quite experienced to handle these various instruments and software packages. Furthermore, just as in a police investigation, the data is rarely obvious or completely conclusive, but rather requires previous experience to unravel it. For all these reasons, interpreting the data is typically the largest bottleneck in meta bo - lite identification. In this report, we present the choices and decisions involved in metabolite identification and how they can be merged into a systematic approach. The discussion covers the tools and techniques available to the mass spectrometrist, the complementary natures of different types of mass spectrometers, and innovative software that reduces the size of data sets. This new approach greatly increases sample throughput and turnaround time for useful information. A complete metabolite identification study is also described, which uses all of the techniques and instrumentation discussed. The flow chart in Figure 1 outlines the basic approach to meta bolite identification. The approach assumes that it is possible to predict numerous common alterations to the drug such as oxidation and oxidative conjugation. Typically, the compounds under investigation in a single drug discovery project are structurally very similar because a prospective lead compound s structure is fine-tuned for better selectivity and potency toward the receptor of interest. Therefore, the mass spectro metrist who analyzes compounds in an analogous series quickly learns the most common metabolic alterations to the parent structure and any novel modifications. This experience and information allows for a guided analysis with targeted A A N A LY T I C A L C H E M I S T R Y / A U G U S T 1, 2 1

5 searches for expected metabolites. (To ensure that novel metabolites caused by less-common metabolic pathways are detected, precursor ion scanning is available.) Techniques and instrumentation An exhaustive description of all available MS instruments and their modes of operation is beyond the scope of this article and not necessary to appreciate the strategies being applied here. However, a basic knowledge of a few types of major systems and their important modes of operation is required and included. Tandem MS. Tandem MS is the cornerstone of metabolite identification (12 14). Tandem MS actually covers a variety of scanning techniques including product ion, precursor ion, and neutralloss scanning. Tandem mass spectrometers usually contain an isolation stage and a fragmentation stage within the same device. Although many different ways exist to complete a tandem MS experiment, all of them follow the same basic series of events. First, the ion of interest is isolated on the basis of its m/z ratio and then passed into the collision cell a region of local high pressure. (In trapping instruments, the isolation and fragmentation normally take place within the same space between electrodes, and the stages are separated by time rather than space.) The collision cell is filled with an inert gas such as argon or helium, and a voltage is applied. Energized ions collide with the target gas, and each collision imparts a small amount of energy to the ion until sufficient energy is deposited to cleave an internal bond or bonds. The resulting ion fragments pass out of the cell and into the detector. The result provides a series of fragment ion masses, all of which correspond to part of the unknown molecule. From these masses, possible structures for the unknown molecule can be calculated. To aid interpretation, the parent compound standard is normally infused into the mass spectrometer in a separate experiment and forced to undergo collisional fragmentation. The resulting fragment ions can then be compared with the unknown and used as a guide during interpretation. Using this method, an experienced operator can rapidly identify the overall structure of the unknown and often pinpoint changes in the molecule. Precursor ion and constant neutral-loss scans. Tandem MS includes the precursor ion and constant neutral-loss scanning modes (14). Both scan types are employed in the first step of metabolite identification in Figure 1. These two scanning modes are an ideal place to start in early drug discovery because they Many pharmaceutical companies are now conducting metabolite identification studies in the early phases of drug candidate selection. provide a large amount of information and require very little knowledge regarding the metabolite s structure. Precursor ion scanning experiments identify likely metabolites by determining if they contain an unaltered portion of the parent molecule or an expected alteration (e.g., +16 Da for oxidation). The power of such an approach is that the operator only needs to know the fragmentation pattern of the parent ion, not the alteration to the parent compound. Constant neutral-loss experiments require no knowledge of the parent compound. This scanning technique searches for expected neutral losses from the analyte. In a metabolite identification study, the neutral losses normally are the characteristic losses detected when a conjugate moiety is broken apart from a conjugated metabolite (e.g., loss of 176 Da from a glucuronide conjugate). This tech - nique provides data on the molecular weights of the conjugated metabolites. These two scanning techniques are incredibly effective at detecting molecules that closely resemble the dosed drug and greatly reduce the amount of data the operator has to analyze. Reducing the data - set size narrows the list of possible meta - bolites for further study. While the triple quadrupole is the only instrument that possesses the hardware to perform precursor ion and constant neutral loss experiments in real time, software is being developed to allow rapid evaluation of data on-the-fly and to provide this type of information on other types of mass spectrometers in pseudo real-time (within the run time of the LC.) Targeted product ion analysis In drug discovery, each new compound normally arises from a small alteration to an initial lead template that was discovered in a receptor-binding assay. The final compound may bear very little similarity to that initial lead, but it has a lineage that stretches back to the initial compound. Consequently, after the first one or two compounds have been fully assayed, any structural regions or soft spots within the compounds in a drug series, which are highly susceptible to metabolism, are determined. Common metabolic alterations can be predicted, and a list of expected metabolites can be compiled, on the basis of a previously analyzed series. By combining this list with a list of suspected metabolites identified by precursor and neutral-loss scan data, a series of ions can be targeted for product ion analysis. Any type of mass spectrometer capable of product ion scanning can be used at this point, including a triple quadrupole. Thus, all of the experiments discussed so far can be completed A U G U S T 1, 2 1 / A N A LY T I C A L C H E M I S T R Y A

6 using a single instrument. However, to localize alterations within a molecule to a specific site, additional stages of tandem MS (MS n ) require a more specialized mass spectrometer. Determining sites of modification Multiple stages of MS can provide large amounts of structural information regarding each analyte, thereby allowing for a more detailed characterization of the metabolites (15, 16). Complet- Accurate mass measurements can often be the deciding factor that chooses one structure among several candidates. ing MS n experiments requires a mass spectrometer that can capture and store ions (13, 15, 16). While the ions are stored, they can be subjected to excitation and collisional fragmentation. The trapping instrument can then capture the resultant fragment ions, which can then be forced to undergo further fragmentation. The second-generation mass spectrum will now give structural information regarding the isolated fragment, allowing easier characterization of that ion. Because this procedure can be applied to each of the initial parent ion fragments, detailed structural information can be acquired rapidly. This technique often allows the isolation of a small region of the parent ion molecule that has been modified and, in some cases, even the individual atom that are different. At present, the Fourier transform (FT) and the quadrupole ion trap mass spectrometers are the only trapping mass spectrometers available. FTMS instruments are not yet capable of high throughput on a regular basis because they are expensive and require skilled operators. Thus, cheaper and simpler quad rupole ion trap mass spectrometers are typically used for these types of trapping experiments. Although MS n experiments can often pinpoint the site of modification very accurately, occasionally a metabolite fragments in a manner that does not provide the required information to identify the type of modification. This is typically the case when an expected fragment ion appears as a modified entity within the tandem MS/MS spectrum, but the modification cannot be explained by a common metabolic alteration (e.g., M+4 to a dimethylbenzene ring). In these cases, the answer can sometimes be determined by using accurate mass measurement, which, in turn, allows the calculation of an empirical formula for the fragment. Although accurate mass measurement is an established technique, the instrumentation required to complete the measurements have traditionally been large, expensive, and specialized. Therefore, accurate mass measurement experiments have not been widely exploited within metabolite identification. However, TOF and hybrid quadrupole-tof (Q-TOF) instruments are easy to use, making accurate mass a feasible part of routine metabolite identification. Our laboratory has successfully obtained accurate mass data by using a Q-TOF instrument. We routinely achieve mass accuracies of <1 ppm. With this high accuracy, a smaller number of possible empirical formulas can be calculated from the data, aiding greatly in characterizing the analyte. Accurate mass MS/MS data for determining empirical formulae of fragment ions can also be acquired, but this experiment is hampered by a lack of reference materials with structures and molecular weights similar to the compound under investigation. In such an experiment, the window for isolating the analyte of interest is opened slightly to allow the reference material into the collision cell. Both the analyte and reference compound undergo fragmentation, and the reference compound fragment ions are used as accurate mass reference points. Although this is certainly not the only meth od for obtaining high-accuracy MS/MS data, in our hands, it has given more consistent and accurate results with samples derived in vivo. Differentiating between two or more possible structures is often possible using this type of accurate MS and MS/MS data. As a result, accurate mass measurements can often be the deciding factor that chooses one structure among several candidates. Identification of a dosed compound With the basics of instrumentation and techniques described, a full metabolite identification study of a compound dosed in vivo can be examined. Sprague-Dawley male rats were dosed intravenously with Schering-Plough discovery compound SCH X at 2 mg/kg body weight and orally at 1 mg/kg. Urine and bile were collected over 24 h at regular time intervals. Before analysis, the individual time point samples for each animal were pooled for each dosing region to generate one - to 24-h bile and one - to 24-h urine sample. No further sample preparation was performed. Minimal sample cleanup is used because the nature, number, and concentrations of metabolites present are unknown, and it is therefore impossible to determine if any will be lost during a sample preparation procedure. Neat bile and urine samples were simply centrifuged to remove any particulates, and the supernatant was injected directly onto an analytical LC column minimizing metabolite loss and decreasing sample preparation time outweigh the reduced life of the analytical columns A A N A LY T I C A L C H E M I S T R Y / A U G U S T 1, 2 1

7 (a) 1 Total ion chromatogram Step 1: Collecting precursor ion scan and neutral-loss data. After sample preparation, a list of potential metabolites in the bile and urine sample needs to be compiled from precursor ion and constant neutral-loss analyses completed on a triple quad - rupole mass spectrometer. If the compound had been radiolabeled, an online radioactivity detector would have also been used to mark time points where the compound elutes within the chromatographic run. Figure 2a demonstrates the type of data recorded in a precursor scan. At least eight apparent metabolites are evident in a single experiment, solely on the basis of the fragment ions similarity to those produced by the parent compound standard. An additional five possible metabolites were found in the urine via precursor ion scanning data (data not shown). The total analysis took 44 min and required one bile and one urine injection to detect all 13 meta - bolites. Typically, more than a single series of fragment ions are scanned, corresponding to characteristic fragments from the top, middle, and lower portions of the parent compound. Common alterations to these fragments, such as hydroxylation, are monitored by looking for precursors of the analyte at the native fragment mass and fragments corresponding to the metabolic hydroxylation, which are 16 Da higher. Constant neutral-loss data were also acquired on the same two samples. Figure 2b shows the results of a constant neutral loss of 176 Da. This is a characteristic loss associated with a glucuronide conjugate. Four po - tential (M+16) glucuronide conjugates and two potential (M+32) glucuronide conjugates were detected in bile, but none were Relative abundance () (b) Relative abundance () Parent drug (M 14) glucuronide metabolite (M 4) metabolite (M 14) metabolite (M + 16) metabolites Time (min) Total ion chromatogram (M + 32) glucuronide metabolites (M + 16) glucuronide metabolites Time (min) FIGURE 2. Precursor ion and constant neutral-loss scan experiments. (a) Precursor ion scan and (b) constant neutral-loss (176 Da) scan experiments using rat bile. In both cases, the data are shown as an individual mass trace. Each trace is labeled with the putative metabolite s potential identity on the basis of their detected mass. found in the urine. Data from the precursor and constant neutral-loss experiments only indicate possible metabolites. Product ion analysis A U G U S T 1, 2 1 / A N A LY T I C A L C H E M I S T R Y A

8 (a) 1 Relative abundance Relative abundance (b) 1 Total ion chromatogram I 5.97 II of these masses can confirm the identity of these analytes as true metabolites derived from compound SCH X. Step 2: Product ion analysis of potential metabolites. The product ion data in the next step were acquired on the Q-TOF mass spectrometer, but a triple quadrupole or quadrupole ion trap mass spectrometer could also have been used Metabolite D MS/MS mass chromatogram Time (min) Metabolite A MS/MS mass chromatogram Metabolite B MS/MS mass chromatogram MS/MS of I MS/MS of II Metabolite C MS/MS mass chromatogram m/z FIGURE 3. Product ion scan data used to identify several putative metabolites first detected in Figure 2. (a) Mass chromatograms for each of the examined metabolites. The individual traces are labeled with the potential identity of the metabolite on the basis of information from Figure 2 experiments. (b) Tandem MS spectra from peaks I and II. The Q-TOF instrument was used to examine a list of expected metabolites that had been previously observed for analogs from this compound series and included +16 Da (hydroxylation), 14 Da (demethylation), and +32 Da (dihydroxylation). In addition, any putative metabolites identified by the precursor ion and neutral-loss scans also underwent product ion analysis. The rapid scanning abilities of the Q-TOF instrument allow several product ion experiments to be performed in rapid succession in a single LC run. In the SCH X example, up to six potential metabolites were analyzed per LC injection. Figure 3 shows an example of the chromatographic trace and product ion data. Overlapping peaks in the LC spectrum are no problem because the MS data provides a second dimension of independent data, which unambiguously identifies the individual peaks. Although some of the chromatographic peaks (I and II in Figure 3a) are not completely resolved, analytes can still be identified without baseline resolution because the product ion data for each peak is independent from the other responses (Figure 3b). Compounds analyzed in this manner were confirmed to be meta - bolites if their product ion spectra produced at least two fragment ions that are characteristic of the proposed structure. After determining that these analytes are indeed meta bolites, a more complete determination of their structure can be accomplished by using the power of MS n. Step 3: Structural elucidation of metabolites by MS n. Because of constraints due to space, expense, and complexity, the quadrupole ion trap is the instrument of choice for MS n exper A A N A LY T I C A L C H E M I S T R Y / A U G U S T 1, 2 1

9 iments in a typical metabolite identification laboratory. For an MS experiment, the masses of the intact metabolite and a related fragment ion are required. Because this information comes from the MS/MS experiments described in Step 2, MS 3 experiments can be set up directly on the instrument without any further experimentation. MS 3 often gives data sufficient to determine the type of meta - bolite modification and indicate which parts of the molecule were changed. In most cases however, especially for molecules that only break into a few large fragments, MS 4 or even MS 5 may be required to locate the site of modification. As shown in Figure 4, an O-glucuronide metabolite had to undergo MS 5 before locating the site of oxidation. In each trace, the fragment ion that will undergo additional fragmentation is highlighted. Thus, this technique quickly pinpoints a very small area that has been altered. Despite the obvious versatility and utility of MS n at providing detailed structural information, the exact structure of the metabolite is still impossible to deter- Instrument manufacturers are creating software packages that help the metabolite identification process. mine in certain cases. MS n will often provide only enough information to narrow down the choices to two or three possible structures. When MS analysis fails to provide absolute identification of analytes or their isomers, LC/NMR can be used. Accurate MS or mass product ion data can often narrow the choices down to only one possible structure in a short time. LC/NMR is really a technique of last resort, because it is a specialty study requiring considerable additional preparation and time. Step 4: Accurate mass measurement. The power of accurately determining the mass of a metabolite lies in being able to determine a list of possible empirical formulae. Obviously, the more accurate the mass measurement, the fewer degrees of freedom are available to the software calculating a formula, and the shorter the list of possibilities. Accurate mass product ion data further limits the number of possible formulae by providing data that are even more specific to the empirical formulae calculator. Product ion experiments narrow the site of modification to a small portion of the molecule and an accurate mass determination of this fragment limits the software to fewer structural possibilities. Furthermore, the operator normally has prior knowledge of the parent compound s structure, which limits the number and types of atoms the software should take into consideration. Until recently, accurate mass measurement required doublesector instruments or Fourier transform ion cyclotron resonance instruments, which are very large, complex, and expensive. However, the advent of less complex and expensive benchtop TOF instruments and hybrid instruments such as Q-TOF has allowed the application of accurate mass measurement in the metabolite identification laboratory. The techniques previously described and the power of accurate mass data are illustrated in Figure 5. Data were generated from the analysis of a novel metabolite of SCH X that was detected in both bile and urine. This molecule was first identified by precursor ion scanning as the addition of 18 mass units to the parent molecule (M+18). The parent compound contained a bromine atom, and the (M+18) ion exhibited a bromine isotopic pattern, which helped confirm that this analyte is indeed a metabolite. Moreover, bromine does not naturally occur in the body. Neutral-loss scan data suggested the presence of a glucuronide moiety within the molecule. Typically, if a free hydroxyl was present, such an alteration to a compound would appear as an addition of 176 mass units (M+176). If there was no free hydroxyl within the parent compound, the observed alteration would normally be the addition of 192 mass units (M+192) due to hydroxylation followed by glucuronidation. Therefore, a mass increase of 18 Da suggested a less-obvious form of glucuronidation. Product ion and MS n experiments were completed next, narrowing the area of alteration to the lower portion of the molecule, which is a 2,6-dimethyl pyri dine ring. This ring was adding 18 Da. Examining the structure of the ring suggested two obvious routes to such a mass transformation: the loss of two methyl groups and the addition of two hydroxyl groups and a glucuronide molecule, or the dearomatization of the pyridine ring and subsequent glucuronidation (Figure 5). Both transformations would give the same nominal mass, but they would be quite different from each other if an accurate mass measurement was completed. Internal and external calibration were used to obtain the accurate mass data required to determine the correct structure. The Q-TOF mass spectrometer was externally calibrated against a solution containing a mixture of various molecular-weight poly - ethylene glycol compounds. This allowed for a mass accuracy in the region of ±.5 Da. However, a higher level of mass accuracy was required to distinguish the two possible structures. To achieve this higher accuracy, an internal calibration was completed during the experiment by infusing a reference material into the LC column eluent just before the electrospray source. This reference material was of known exact mass, and the acquisition software could therefore adjust the slope of the external calibra- A U G U S T 1, 2 1 / A N A LY T I C A L C H E M I S T R Y A

10 Relative abundance 1 MS 4 MS MS 5 MS 3 Glucuronide tion curve slope in real time for each scan. This allowed much higher accuracy by compensating for scan-to-scan variations and fluctuations. Using this approach, an accuracy of ±.5 Da was readily achieved, which was sufficient to differentiate between the two possible structures by mass measurement of the unknown metabolite. The results showed that the metabolite differed by.82 Da from the theoretical mass for the dearomatized N-glucuronide structure. However, the observed metabolite mass differed by only.95 Da from the theoretical mass for the demethylated, dihydroxylated glucuronide structure. This allowed the dearomatized N-glucuronide structure to be discounted, thereby leaving the demethylated dihydroxylated glucuronide structure as the most likely metabolite. Any further confirmation of the structure would be left to NMR and LC/NMR; however, the information provided by MS studies is usually sufficient within drug discovery. Still searching The strategies detailed in this report are the result of more than two years of work. They are not meant to exhaustively identify every metabolite present within in vivo samples, but rather to rapidly and accurately identify sites of metabolic lability that might affect pharmacokinetic measurement results and detect m/z MS 615 MS MS MS FIGURE 4. MS 1 through MS 5 data from the analysis of an O-glucuronide metabolite. 219 MS 5 Each trace shows the next level of tandem MS in the sequence. Inset illustrates the sequential cleavage observed in the metabolite. the formation of possibly toxic metabolites. Therefore, only major metabolites are identified, but the definition of a major metabolite differs widely among researchers. Because ~5 of our samples are radioactive, our laboratory definition of major includes any resolved radioactive peak that constitutes 5 or greater of the overall radioactivity detected in that experiment. However, inevitably, other low-level metabolites are automatically detected using the prescribed strategies. The definition of a major metabolite is more difficult to quantify for nonradiolabeled samples. Again, the experiments detailed here are designed to complement each other like a series of filters. If a compound is of sufficient importance to become a possible lead candidate, it will be radiolabeled and reanalyzed later. At this point, any metabolites missed in the nonradiolabeled study should become apparent. However, in our experience over the past two years with numerous lead compounds going through nonradiolabeled and radiolabeled studies, no major metabolites were detected in the radioactive study that were not detected in the initial study. Despite improvements made by these approaches, physical interpretation of the data is still a major bottleneck in the meta - bolite identification process because it is so time consuming. These new approaches to data acquisition have led to many more samples being run in the same time period, which, in turn, increase the amount of data. Software that can reduce operator workload by using a series of criteria to analyze data and report apparent meta bolites will significantly improve throughput in metabolite identification. Instrument manufacturers are creating software packages that help the metabolite identification process. Although these packages are simple at present, the impetus is growing rapidly, and software such as Metabolynx (Micromass), Metabolite ID (PE Sciex), and the new Metabolite Data Browser (Thermo Finnigan) are helping reduce the workload of analysts. These packages automatically perform functions that researchers presently complete by hand such as the background subtraction of a control data file or the application of an isotope cluster A A N A LY T I C A L C H E M I S T R Y / A U G U S T 1, 2 1

11 analysis for chlorine- and bro - mine-containing compounds. At a minimum, the software searches for a list of expected metabolites from full-scan data and returns a list of possible hits. This allows numer - ous extraneous noise responses to be automatically discarded. As these packages evolve, they will handle correlation analysis of MS/MS data and data-dependent MS/MS acquisition of potential meta - bolites. All of these factors aid in reducing the data set that the analyst has to examine, thereby increasing throughput. The new direction for metabolite identification is the integration of automated data interpretation with Monkey bile pool IV 8 h (a) 1 LC/MS/MS in combination with the protocols described in this report. This approach will significantly accelerate metabolite identification. The authors would like to acknowledge José Castro-Perez, Sean Bennett, and Steve Preece from Micromass Inc.; Yves LeBlanc and Byron Kieser from Applied Biosystems; and Sarah Kambouris from Thermo Finnigan. All of the authors are currently conducting research at Schering- Plough Research Institute. Nigel J. Clarke and Diane Rindgen are both associate principle scientists working in the discovery metabolite identification group under the direction of Kathleen A. Cox. The group s interests lie mainly in automation of metabolite identification and structural elucidation through the use of new instrumentation and software. The group uses LC/tandem MS on a variety of instrument platforms to perform de novo characterization and identification of discovery drug candidate metabolites from various biological matrices, both in vivo and in vitro in source. Walter A. Korfmacher is the director of the discovery bioanalytical MS group of which the metabolite identification group is part. Address correspondence about this article to Clarke at Schering-Plough Research Institute, 215 Galloping Hill Rd., Kenilworth, NJ 733 or Nigel.Clarke@ spcorp.com. References (1) Cox, K. A.; Dunn-Meynell, K.; Korfmacher, W. A.; Broske, L.; Nomeir, A. A.; Lin, C.-C.; Cayen, M. N.; Barr, W. H. Drug Disc. Today 1999, 4, (2) Korfmacher, W. A.; Cox, K. A.; Bryant, M. S.; Veals, J.; Ng, K.; Watkins, (b) Relative abundance FIGURE 5. Accurate mass data from monkey bile (a) Raw data and b) accurately measured data after the external reference mass calibration was applied. The partial chemical structure shows the two most likely alterations that may have occurred. m/z R.; Lin, C.-C. Drug Disc. Today 1997, 2, (3) Lee, W. M. Aliment Phamacol. Ther. 1993, 7, (4) Pirmohamed, M.; Madden, S.; Park, B. K. Clin. Pharmacokinet. 1996, 31, (5) Pohl, L. R.; Pumford, N. R.; Martin, J. L. Eur. J. Haematiol. Suppl. 1996, 6, (6) Kreunter, W.; Hey, J. A.; Anthes, J.; Barnett, A.; Young, S.; Tozzi, S. Arzneimittelforschung 2, 5, (7) Glue, P.; Clement, R. P. Cell Mol. Neurobiol. 1999, 19, (8) Zhang, N.; Fountain, S. T.; Honggang, B.; Rossi, D. T. Anal. Chem. 2, 72, (9) Loo, J. A.; DeJohn, D. E.; Du, P.; Stevenson, T. I.; Ogorzalek Loo, R. R. Med. Res. Rev. 1999, 19, (1) Lim, H. K.; Stellingweif, S.; Sisenwine, S.; Chan, K. W. J. Chromatogr. 1999, 83, (11) Schultz, G.; Lowes, S.; Henion, J.; Hop, C.; Carlin, J. Comparison of a Triple Quadrupole Using SRM to a TOFMS for Quantitative LC-MS Support of Drug Discovery Programs; American Society of Mass Spectrometry: Orlando, FL, (12) Yost, R. A.; Enke, C. G. J. Am. Chem. Soc. 1978, 1, (13) Yost, R. A.; Enke, C. G.; McGilvery, D. C.; Smith, D.; Morrison, J. D. Int. J. Mass Spectrom. Ion. Proc. 1979, 3, 127. (14) Busch, K. L.; Glish, G. L.; McLuckey, S. A. Mass Spectrometry/Mass Spectrometry: Techniques and Applications of Tandem Mass Spectrometry; VCH Publishers: New York, (15) Stafford, Jr., G. C.; Kelley, P. E.; Syka, J. E. P.; Reynolds, W. E.; Todd, J. F. J. Int. J. Mass Spectrom. Ion. Proc. 1984, 6, 85. (16) Louris, J. N.; Cooks, R. G.; Syka, J. E. P.; Kelley, P. E.; Stafford, Jr., G. C.; Todd, J. F. J. Anal. Chem. 1987, 59, R N or HO R R OH N N A U G U S T 1, 2 1 / A N A LY T I C A L C H E M I S T R Y A