Article pubs.acs.org/ac
Calculation and Mitigation of Isotopic Interferences in Liquid Chromatography−Mass Spectrometry/Mass Spectrometry Assays and Its Application in Supporting Microdose Absolute Bioavailability Studies Huidong Gu,*,† Jian Wang,† Anne-Françoise Aubry,† Hao Jiang,† Jianing Zeng,† John Easter,‡ Jun-sheng Wang,§ Randy Dockens,§ Marc Bifano,§ Richard Burrell,‡ and Mark E. Arnold† †
Bioanalytical Sciences, Research and Development, Bristol-Myers Squibb, Route 206 and Province Line Road, Princeton, New Jersey 08543, United States ‡ Discovery Chemistry Synthesis, Research and Development, Bristol-Myers Squibb, 5 Research Parkway, Wallingford, Connecticut 06492, United States § Discovery Medicine and Clinical Pharmacology, Research and Development, Bristol-Myers Squibb, 311 Pennington-Rocky Hill Road, Pennington, New Jersey 08534, United States S Supporting Information *
ABSTRACT: A methodology for the accurate calculation and mitigation of isotopic interferences in liquid chromatography− mass spectrometry/mass spectrometry (LC−MS/MS) assays and its application in supporting microdose absolute bioavailability studies are reported for the first time. For simplicity, this calculation methodology and the strategy to minimize the isotopic interference are demonstrated using a simple molecule entity, then applied to actual development drugs. The exact isotopic interferences calculated with this methodology were often much less than the traditionally used, overestimated isotopic interferences simply based on the molecular isotope abundance. One application of the methodology is the selection of a stable isotopically labeled internal standard (SIL-IS) for an LC−MS/MS bioanalytical assay. The second application is the selection of an SIL analogue for use in intravenous (IV) microdosing for the determination of absolute bioavailability. In the case of microdosing, the traditional approach of calculating isotopic interferences can result in selecting a labeling scheme that overlabels the IV-dosed drug or leads to incorrect conclusions on the feasibility of using an SIL drug and analysis by LC−MS/MS. The methodology presented here can guide the synthesis by accurately calculating the isotopic interferences when labeling at different positions, using different selective reaction monitoring (SRM) transitions or adding more labeling positions. This methodology has been successfully applied to the selection of the labeled IV-dosed drugs for use in two microdose absolute bioavailability studies, before initiating the chemical synthesis. With this methodology, significant time and cost saving can be achieved in supporting microdose absolute bioavailability studies with stable labeled drugs.
S
and the number of labeling positions, the isotope distribution from an analyte with molecular weight of M can potentially interfere with the SIL-IS with a molecular weight of M + N
table isotopically labeled (SIL) internal standards (IS) are widely used in liquid chromatography−mass spectrometry/ mass spectrometry (LC−MS/MS) assays for the quantitative determination of drugs and their metabolites in biological matrixes1 since they have “identical” physicochemical properties regarding sample extraction, chromatographic separation, and mass spectrometry ionization. Depending on their structures © 2012 American Chemical Society
Received: February 14, 2012 Accepted: April 27, 2012 Published: April 27, 2012 4844
dx.doi.org/10.1021/ac300442v | Anal. Chem. 2012, 84, 4844−4850
Analytical Chemistry
■
Article
EXPERIMENTAL SECTION Materials and Reagents. BMS-708163 (avagacestat), 13 C6-avagacestat, BMS-790052 (daclatasvir), 15N413C2-daclatasvir, and 13C10-daclatasvir were synthesized at Bristol-Myers Squibb. Formic acid and acetic acid were purchased from EMD (Darmstadt, Germany). Acetonitrile, ACS HPLC grade, was purchased from EM Sciences (Gibbstown, NJ). Dimethyl sulfoxide (DMSO) and ammonium acetate were purchased from J.T.Baker Chemicals (Phillipsburg, NJ). LC−MS/MS Instrumentation. LC−MS/MS data were acquired by Analyst software on a Sciex API 4000 (Analyst 1.4.2) and Sciex API 5500 (Analyst 1.5.1) (Applied Biosystems, Foster City, CA) for the avagacestat assay and daclatasvir assay, respectively. The separation was achieved on a Waters (Milford, MA) Atlantis dC18 analytical column (2.1 mm × 50 mm, particle size 3 μm) with gradient elution using mobile phases of 0.01% formic acid in water (A) and 0.01% formic acid in acetonitrile (B) for the avagacestat assay and 0.01% acetic acid in 5 mM ammonium acetate in water (A) and acetonitrile (B) for the daclatasvir assay. Two Shimadzu (Tokyo, Japan) LC10ADvp pumps and a Shimadzu SIL-HTC autosampler were used as the delivery pumps and autosampler, respectively. The flow rate was 0.4 mL/min. Sample Preparation for LC−MS/MS Measurements of Isotopic Interferences. All stock solutions for avagacestat, 13 C6-avagacestat, daclatasvir, 15N413C2-daclatasvir, and 13C10daclatasvir were prepared in acetonitrile/DMSO (1:1, v/v). Avagacestat at 500, 10 000, and 50 000 ng/mL and 13C6-avagacestat at 500 ng/mL were prepared by appropriate dilution of 0.2 mg/mL of corresponding stock solutions with water/ acetonitrile (1:1, v/v). Daclatasvir at 2000, 10 000, and 200 000 ng/mL and 15N413C2-daclatasvir at 2000 ng/mL were prepared by appropriate dilution of the 1.0 mg/mL corresponding stock solutions with water/acetonitrile (1:1, v/v). An amount of 300 μL of each sample was transferred into a 96-well plate and 2−10 μL was injected into the LC−MS/MS system.
(where N is the number of isotope labeling positions) in LC− MS/MS assays. Obviously, there is no interference from a labeled compound to its nonlabeled or less labeled compound if isotopic purity is not a concern. Therefore, the evaluation of the isotopic interference from an analyte to its SIL-IS is the first step in the selection of an SIL-IS for an LC−MS/MS assay. The isotope distribution of a molecule can be easily calculated. There are several online calculators that exist for this purpose2,3 and are commonly used to estimate the maximum isotopic interference from an analyte to its SIL-IS in an LC−MS/MS assay. However, this approach overestimates the interference as it does not take into account the fragmentation. The calculations presented herein include the selective reaction monitoring (SRM) transition to calculate accurately the isotopic interference from an analyte to its SIL-IS in triple-quadrupole mass spectrometers. In all cases, the actual interference was less than the estimate based on simple isotope distribution. Our approach was also applied to the selection of SIL drugs suitable for microdosing studies. Absolute bioavailability (BA) of a drug is normally assessed by sequential oral and intravenous (IV) administrations of the drug to the same subjects in a crossover study. Recently, with the successful application of accelerator mass spectrometry (AMS) in microdose metabolism and pharmacokinetic studies,4−7 several microdose absolute BA studies have been reported that consisted of the administration of a very small amount of 14C-labeled IV-dosed drug (100 μg or less) concurrently with a therapeutic oral dose of the unlabeled drug.8,9 In these studies, LC−MS/MS was normally used for the analysis of the unlabeled oral drug, while AMS was used for the determination of the 14C-labeled IV drug because of its ultrahigh sensitivity.9 This strategy has gained broad acceptance in the pharmaceutical industry. However, there are two drawbacks for this strategy. The first is the cost and complexity of the AMS measurement. The second is the potential data inconsistency since the samples are analyzed by two different types of technologies and normally in two separate laboratories, as well as that AMS requires chromatographic separation of the drug from all of its metabolites due to the lack of specificity in the AMS detector. With the sensitivity improvements of newer mass spectrometry instrumentation into the picomolar or even femtomolar range, there is renewed interest in applying a microdosing approach with SIL IV dosing and LC−MS/MS determination of both the labeled and unlabeled drugs. Instead of single-atom 14 C labeling, a multiposition stable labeling scheme (e.g., 13C, 15 N) is given as a microdose IV administration. An SIL-IS is essential for the accurate LC−MS/MS analysis due to the potential ion suppression caused by ionization competition between the unlabeled oral drug and labeled IV drug.10 With not just one, but two SIL molecules, accurate calculations of the isotopic interferences among the unlabeled oral drug, the labeled IV drug, and the labeled IS become critical to guide the chemical synthesis process. A new methodology to calculate and mitigate the isotopic interferences in triple-quadrupole mass spectrometer based on the number of labeled atoms, the labeling positions, and the product ion selected has been developed and is demonstrated here using a simple molecular entity. The application of this methodology to the selection of the labeled IV drug and labeled IS in supporting two microdose absolute BA studies at Bristol-Myers Squibb is also presented.
■
RESULTS AND DISCUSSION Theory Based On a Simple Model. To demonstrate the methodology for the isotopic interference calculation in a simple and easy way, HA1A2A3 is used to represent a simple molecular entity. A1, A2, and A3 are three different atoms of mass m1, m2 and m3, respectively. A two-position labeled compound is HA1m2+1A2m3+1A3, with one label on A2 and the other label on A3. It is assumed that the relative isotope abundances for naturally occurring isotopes of A1, A2, and A3 are available and shown in Table 1.1.11 It is also assumed that the Table 1.1. Assumed Isotope Abundance (%) for Naturally Occurring Isotopes of A1, A2, and A3 mass
A1 (abundance %)
A2 (abundance %)
A3 (abundance %)
+0 +1 +2
100 1.00 NA
100 0.30 NA
100 0.04 0.20
precursor ion is [HA1A2A3 − H]− and the product ion is [A1A2]− under negative electrospray ionization (ESI) mode. Therefore, the SRM transitions (m/z) are m1 + m2 + m3 → m1 + m2 and m1 + m2 + m3 + 2 → m1 + m2 + 1 for HA1A2A3 and HA1m2+1A2m3+1A3, respectively. The isotope distribution for [HA1A2A3 − H]− is shown in Table 1.2. (Note that, for a real 4845
dx.doi.org/10.1021/ac300442v | Anal. Chem. 2012, 84, 4844−4850
Analytical Chemistry
Article
Table 1.2. Isotope Distribution for [HA1A2A3 − H]− A1 mass 1 2 3 4 5 6 7 8 9 10 11 12
m1 m1 m1 m1 m1 m1 m1 m1 m1 m1 m1 m1
abundance (%)
+1 +1 +1 +1 +1 +1
100 100 100 1.00 100 1.00 1.00 100 1.00 1.00 100 1.00
A2 mass m2 m2 m2 m2 m2 m2 m2 m2 m2 m2 m2 m2
+1 +1 +1 +1 +1 +1
abundance (%) 100 100 0.30 100 0.30 100 0.30 100 0.30 100 0.30 0.30
A3 mass m3 m3 m3 m3 m3 m3 m3 m3 m3 m3 m3 m3
+1
+1 +1 + + + + +
2 1 2 2 2
abundance (%)
isotopic molecular mass
combined abundance (%)
isotope distribution for [HA1A2A3 − H]− (%)
100 0.04 100 100 0.04 0.04 100 0.20 0.04 0.20 0.20 0.20
m1 + m2 + m3 m1 + m2 + m3 + 1
100.00000 0.04000 0.30000 1.00000 0.00012 0.00040 0.00300 0.20000 0.00000 0.00200 0.00060 0.00001
100.0000 1.3400
m1 + m2 + m3 + 2
m1 + m2 + m3 + 3
m1 + m2 + m3 + 4
0.2035
0.0026
0.0000
Table 1.3. Isotopic Interferences for Different Product Ions Having [HA1A2A3 − H]− as Precursor Ion product ion A2A3− A1A2− A2− A3− A1−
SRM transition monitored for [HA1A2A3 − H]−
contributing combinations in Table 1.2
actual isotopic interference/ overestimated isotopic interference (%)
0.2001
5, 8
98.3
0.0005
5, 6
0.2
0.0031
5, 7
1.5
0.0005
5, 6
0.2
0.2001
5, 8
98.3
SRM transition monitored isotopic interference from [HA1A2A3 − for [HA1m2+1A2m3+1A3 − H]− H]− to [HA1m2+1A2m3+1A3 − H]− (%)
m1 + m2 + m3 → m2 + m1 + m2 + m3 + 2 → m2 + m3 m3 + 2 m1 + m2 + m3 → m1 + m1 + m2 + m3 + 2 → m1 + m2 m2 + 1 m1 + m2 + m3 → m2 m1 + m2 + m3 + 2 → m2 + 1 m1 + m2 + m3 → m3 m1 + m2 + m3 + 2 → m3 + 1 m1 + m2 + m3 → m1 m1 + m2 + m3 + 2 → m1
molecule, it would be easily calculated with an online calculator.2,3) This is actually what would be observed in the Q1 full scan of [HA1A2A3 − H]−. The isotope abundance of [HA1A2A3 − H]− at mass of m1 + m2 + m3 + 2 is 0.2035%, and this number is traditionally used to estimate the isotopic interference from the analyte (HA1A2A3) to the two-position labeled compound (HA1m2+1A2m3+1A3) in LC−MS/MS analysis. However, this is an overestimation of the interference since m/z = m1 + m2 + m3 + 2 meets only one of the two criteria for HA1A2A3 to pass through the SRM transition (m/z m1 + m2 + m3 + 2 → m1 + m2 + 1) of the labeled HA1m2+1A2m3+1A3. In other words, for HA1A2A3, not all of the isotopic ions that passed through Q1 (m/z = m1 + m2 + m3 + 2) would pass through Q3 (m/z = m1 + m2 + 1). Table 1.2 summarizes all possible isotopic molecules, their masses, and how much they contribute to the isotopic distribution of [HA1A2A3 − H]−. Among them, four isotopic molecules have an isotopic molecular mass of m1 + m2 + m3 + 2 (species 5, 6, 7, and 8 in Table 1.2). Of those four, only species 5 and 6 can yield a product ion with a mass of m1 + m2 + 1 for [A1A2]−, and these are the only two species that can pass through both Q1 and Q3 in the SRM transition of the labeled HA1m2+1A2m3+1A3. Therefore, the isotopic interference is only 0.00052% (0.00012% + 0.00040%) instead of 0.2035%, or about 400-fold less. Table 1.3 shows the isotopic interferences obtained when different product ions are used. In this example, for the purpose of demonstration, it is assumed that A1−, A2−, A3−, A1A2−, and A2A3− are available for the product ions. According to the results shown in Tables 1.2 and 1.3, several observations and the strategy to mitigate the isotopic interferences in LC−MS/MS assays are summarized as follows: (1) Depending on the product ion selected, the isotopic interference varies significantly. However, the maximum possible isotopic interference will not exceed the corresponding isotope
distribution of the precursor ion, which is traditionally used to estimate the isotopic interferences in LC−MS/MS assays. (2) The failure to take into account the SRM transition when calculating isotopic interferences can result in a labeling scheme that uses more labeling positions than is really needed. At times, it may seem impossible to mitigate the interference when, if calculated accurately, the actual isotopic interference would have been acceptable. (3) A different product ion can sometimes be used to mitigate high isotopic interference. (4) Labeling at different positions is another option to mitigate the high isotopic interference if the choice for a different product ion is unavailable and if the chemistry allows. In this example, if A2A3− was the only major product ion, and the isotopic interference of 0.2001% was unacceptable, labeling on A1 and A2 as Hm1+1A1m2+1A2A3 would reduce the isotopic interference to 0.0034% (0.0030% + 0.0004%, species 6 and 7 in Table 1.2). (5) Labeling at additional positions can also be considered if other options are exhausted. In this example, if A2A3− was the only major product ion, labeling all three atoms as Hm1+1A1m2+1A2m3+1A3 will reduce the isotopic interference to 0.002% (0.0000% + 0.0020%, species 9 and 10 in Table 1.2). (6) Finally, and most importantly, with the use of this calculation along with the knowledge about the possible SRM transitions and feasibility of chemical labeling, one can inform the labeling scheme before the chemical synthesis which will result in significant time and cost savings. It should be pointed out here that the calculation was demonstrated with an assumption of using a unit resolution mass spectrometer. Application in Supporting Microdose Absolute BA Studies. Example 1: Avagacestat. Avagacestat (Figure 1a) is a γ secretase inhibitor which is being developed for the treatment of Alzheimer’s disease. By using 13C6-labeled avagacestat (Figure 1b) as the IS, two LC−MS/MS assays with standard 4846
dx.doi.org/10.1021/ac300442v | Anal. Chem. 2012, 84, 4844−4850
Analytical Chemistry
Article
interference from avagacestat (m/z 565 → 175) to 13C6avagacestat (m/z 571 → 181) in an LC−MS/MS assay, based on the isotope distribution of avagacestat formate adduct precursor ion at m/z of 571, is 0.0875%.2 On the basis of this number, any six-position labeled avagacestat, such as 13C6avagacestat (Figure 1b), is qualified to be used as the IS for avagacestat in an LC−MS/MS assay since the estimated isotopic interference to 13C6-avagacestat at the assay upper limit of quantitation is only 0.44% (= 0.0875% × 1000/200), assuming the assay range is 1−1000 ng/mL and the equivalent IS concentration is 200 ng/mL. However, the same six-position labeled drug would not be suitable for use in the microdose BA study as the isotopic interference would be 78.75% (= 0.0875% × 900). The isotope abundances for all seven possible combinations of the product ion fragmentation [C6H5O2SCl − H]− and neutral loss C15H14F4N4O4, which could make the isotopic molecular mass of 571 for [C20H17ClF4N4O4S − H + HCOOH]−, are listed in Table 2.1. Of all the combinations shown in Table 2.1, only one has a mass of 571 for the parent ion and 181 for the product ion, and this is the only combination (species 1) that can pass through the SRM transition of 13C6-avagacestat (m/z 571 → 181). The actual isotopic interference from avagacestat to 13C6-avagacestat is calculated as 0.0157%, more than 5 times lower than the estimated 0.0875%. However, 13C6-avagacestat is not suitable for use as the labeled IV microdose drug since the total isotopic interference from avagacestat to 13C6-avagacestat (0.0157% × 900 = 14.23%) would be too high. Therefore, it was necessary to synthesize an IV microdose drug with different labeling scheme as 13C6-avagacestat while avagacestat was used as the oral drug and 13C6-avagacestat was used as the assay IS. Several labeling schemes, listed as follows, were proposed as potential labeled IV drug candidates: (A) additional labels on 13 C6-avagacestat inside its product ion; (B) additional labels on 13 C6-avagacestat outside its product ion; (C) labels on avagacestat outside its product ion. The isotopic interferences from avagacestat and 13C6avagacestat to A, B, and C were calculated, and the results are shown in Table 2.2. No isotopic interferences are expected from A, B, and C to avagacestat and 13C6-avagacestat. It was found that at least three additional labeling positions are needed for both A and B to reduce the isotopic interference from 13C6-avagacestat (IS) to an acceptable level although two additional labeling positions would already eliminate practically
Figure 1. (a) Chemical structure and MS/MS fragmentation of avagacestat. (b) Chemical structure and MS/MS fragmentation of 13 C6-avagacestat.
curve ranges of 0.1−100 ng/mL and 1−1000 ng/mL for avagacestat were developed to support phase I and II clinical studies, respectively. The formate adduct ion was used as the precursor ion under negative ESI mode to achieve better sensitivity.12 In addition, only one major product ion with m/z of 175 was observed (Figure 1). Therefore, the SRM transitions (m/z) for avagacestat and 13C6-avagacestat are 565 → 175 and 571 → 181, respectively. In the proposed microdose absolute BA study, the estimated maximum plasma concentration of the labeled avagacestat, administered IV was about 900-fold lower than that of the unlabeled avagacestat administered orally. The isotope distribution for the avagacestat formate adduct precursor ion, [C20H17ClF4N4O4S − H + HCOOH]−, is presented in Supporting Information A. The estimated isotopic
Table 2.1. Calculated Abundances for All Seven Naturally Occurring Isotopic Ions of Avagacestat Formate Adduct Precursor Ion, [C20H17ClF4N4O4S − H + HCOOH]−, at Mass of 571a parent ion (m/z = 565) at mass of 571
lost in collision cell (neutral loss)
product ion (m/z = 175)
[C20H17ClF4N4O4S − H + HCOOH]−
isotope abundance of C15H14F4N4O4
isotope abundance of [C6H5O2SCl − H]−
mass
mass
abundance
mass
abundance
1 571 390 100 181 0.0157 2 571 391 18.0532 180 0.1087 3 571 392 2.3411 179 1.6905 4 571 393 0.2267 178 2.7325 5 571 394 0.0177 177 37.4666 6 571 395 0.0010 176 7.4229 7 571 396
