- Email: [email protected]
MS using surrogate standards
Journal of Pharmaceutical and Biomedical Analysis 52 (2010) 534–543
Contents lists available at ScienceDirect
Journal of Pharmaceutical and Biomedical Analysis journal homepage: www.elsevier.com/locate/jpba
Quantitative determination of hippuric and benzoic acids in urine by LC–MS/MS using surrogate standards Natalia Penner 1 , Ragu Ramanathan 2 , Joanna Zgoda-Pols ∗ , Swapan Chowdhury Drug Metabolism and Pharmacokinetics, Merck Research Laboratories, 2015 Galloping Hill Road, Kenilworth, NJ 07033, United States
a r t i c l e
i n f o
Article history: Received 17 November 2009 Received in revised form 7 January 2010 Accepted 9 January 2010 Available online 18 January 2010 Keywords: LC–MS/MS Hippuric acid Benzoic acid Biomarkers
a b s t r a c t An LC–MS/MS method for simultaneous determination of hippuric acid (HA) and benzoic acid (BA) in monkey urine after direct injection was developed. Since HA and BA are endogenous compounds in urine, surrogate standards (13 C6 -hippuric and 13 C6 -benzoic acid) were employed to generate calibration curves. l-Phenylalanine-ring-D5 served as an internal standard. Multiple reaction monitoring in the negative ionization mode with an APCI source was used for detection of all components in the assay. The developed method is intended for determination of HA and BA in the range of 0.25–250 and 0.1–100 g/ml, respectively. Weighted (1/x) quadratic regression (r2 > 0.99) was used to generate calibration curves. Precision and accuracy of the method were assessed by analyzing 3 quality control samples (concentrations at low, medium, and high range of calibration curve) prepared in monkey urine. Stability for 48 h at room temperature and after 3 freeze–thaw cycles was also evaluated. The proposed method was successfully utilized for analysis of urine samples from female monkeys following the administration of everninomicin alone and in combination with gentamicin. The concentrations of endogenous HA and BA were calculated based on the peak area ratio of the analyte to the internal standard using a regression equation for corresponding surrogate standard. © 2010 Elsevier B.V. All rights reserved.
1. Introduction Exploring etiology of human diseases is one of the major goals of pharmaceutical research and medical science. A better understanding of how a disease originates and progresses could help in prevention or creating new approaches for detection and cure. Any living organism can be viewed as an extremely complex system, whose quantitative and qualitative composition reflects disease state, toxicological influences, environmental stress, age, and general health. Systems biology and physiological changes occurring in a living organism are usually studied from the point of view of multiple “omics”. Genomics targets genes and their mutations, transcriptomics deals with changes in mRNA level, proteomics studies protein expression, post-translational modifications, and up- and down-regulation in both. Metabonomics (metabolomics) targets the profile of endogenous compounds produced by a living organism and how this profile changes depending on a state of a biological system. Among the different “omics”, only metabonomics is able to generate information which reflects the function of the whole system in real time. This is achieved by
∗ Corresponding author. Tel.: +1 908 740 2170; fax: +1 908 740 3966. E-mail address: [email protected] (J. Zgoda-Pols). 1 Present address: Biogen Idec, Cambridge, MA, United States. 2 Present address: Bristol-Myers Squibb Company, Princeton, NJ, United States. 0731-7085/$ – see front matter © 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.jpba.2010.01.016
analyzing minimally invasive biofluids, such as urine and plasma, which are readily available in clinical and non-clinical setting. Analysis is usually carried out by high-resolution NMR spectroscopy or LC–MS with subsequent data reduction using statistical tools such as principal component analysis (PCA) [1–7]. Comparison of treatment groups may result in specific patterns or changes in metabolic profiles which can be then used for drug toxicity prediction and safety assessment [1–4,7], lead compound optimization [7], the study of gender-specific differences [5], inter-subject variability [6], and potentially, if a specific biomarker/group of biomarkers is identified, serve as an end-point in clinical trials. As with any complex biological matrix the biggest challenge is to recognize a comparatively small number of compounds that can inform us about possible toxicity, an adverse reaction, or pharmacodynamic effect. Identification of endogenous small molecule biomarkers that are related to natural, disease-related or drug-induced pharmacological events is the main objective of metabonomics. Standard workflow in most studies starts with creating a differential display of components changing in the system following external stimuli or as a response to disease progression, and then trying to identify them later. In the case of a disease marker, even when the concentration of an endogenous substance changes in response to a certain event or treatment, its use may not be straightforward. For example, hippuric acid (HA), the glycine conjugate of benzoic acid (BA), has been used for many years in occupational medicine for monitoring exposure
N. Penner et al. / Journal of Pharmaceutical and Biomedical Analysis 52 (2010) 534–543
to toluene [8]. Toluene is primarily metabolized to benzoic acid which then conjugates with glycine to form hippuric acid. HA is one of the main urinary endogenous metabolites and a principal route of benzoate elimination. HA participates in the natural regulation and prevention of stone formation [9]. Multiple studies conducted to evaluate the effect of various factors on excretion of hippuric acid showed that the HA concentration in urine is gender-, age-, race-, and generally subject-dependent [10]. Saito and Takeichi reported that urinary HA concentration in healthy human subjects (n = 6) was within a range of 5.5–15.8 g/ml [11]. In another larger study specifically designed to establish reference values of HA in a general population in one area of Brazil [12], HA levels of 0.18 ± 0.10 g/g creatinine corresponding to a mean of 280 g/ml (n = 115) was observed. These and other studies [13] confirmed, that the normal physiological level of HA in humans and animals is highly variable and depends on diet. Consumption of benzoate- or polyphenol-containing food and beverages (e.g. black tea) increased HA excretion [14]. It is also known that highly significant differences of hippurate levels may point to certain diseases e.g. diabetes in children and renal failure [15]. A similar situation occurs for BA. In some species, including humans, urinary excretion of minor amounts of benzoic acid is normal. However, impaired hippurate synthesis accompanied by elevated benzoate excretion in preterm babies can be an indication of fatal benzyl alcohol poisoning [16]; a similar picture in adults has even been associated with mental illness. When certain compounds are considered potential biomarkers, it is beneficial to monitor them over the course of a study or treatment duration to eliminate inherent inter-subject differences and to determine a trend. Based on literature information, the HA/BA pair can be a prospective biomarker combination. There have been a number of methods developed over the years utilizing colorimetric, LCUV [17], CE-MS [18], and GC–MS [19] techniques for quantitative determination of HA/BA in urine. Spectrophotometric methods are usually less selective and can suffer from interference from other urinary metabolites. The main disadvantage of GC–MS is the derivatization step which is necessary to convert HA/BA into more volatile esters or trimethylsilyl derivatives [19] prior to injection. The lack of an appropriate blank matrix for quantification of endogenous HA and BA has not been addressed before. A “surrogate standard” approach for quantitation of endogenous ␣ketoisocaproic acid in plasma was recently reported by Cohen and co-worker [20]. Deuterated ␣-ketoisocaproic acid was used to generate a standard curve and the concentration of naturally occurring ␣-ketoisocaproic acid was back-calculated using the regression equation derived for the deuterated surrogate standard. A similar approach was used for quantitation of endogenous sorbitol and fructose in human erythrocytes [21]. We also demonstrated earlier [22] that HA can be quantified based on calibration curve from 13 C -hippuric acid. The work described here was devoted to the 6 development of an LC–MS/MS method for simultaneous determination of HA and BA in monkey urine after direct injection. The method was qualified with precision and accuracy criteria set as described by Korfmacher [23] for discovery-supporting non-GLP assays requiring QC samples (Level III) at 25% and 75–125%, respectively, and used for analysis of urine samples from female monkeys following exposure to known nephrotoxicants everninomicin and gentamicin [24].
2. Experimental 2.1. Instruments and chemicals LC–MS experiments were performed using a TSQ Quantum mass spectrometer (Thermo Electron Corp., San Jose, CA) coupled with a
535
Waters Alliance 2695 HPLC module. TSQ Quantum mass spectrometer was operated with Q1 and/or Q3 resolution set at 0.1 Th FWHM (where Th = Thompson and FWHM = Full width at half maximum). Hippuric acid (HA), benzoic acid (BA), 5-bromo-3-hydroxyhippuric acid (5-Br-HA), acetic acid, and ammonium acetate were obtained from Sigma–Aldrich (Milwaukee, WI). 13 C6 -benzoic acid (13 C-BA) and l-phenylalanine-ring-D5 were purchased from Cambridge Isotope Laboratories, Inc. (Andover, MA). 13 C6 -hippuric acid (13 C-HA) was synthesized by Radiochemistry Group at Schering-Plough Research Institute. Ammonium carbonate was obtained from Acros Organic (Belgium). Acetonitrile and methanol were from Burdick and Jackson (Muskegon, MI). Deionized water was purified by MilliQ water purification system (Millipore, Bedford, MA). DMSO was from Fisher Scientific (Fair Lawn, NJ). A Chromolith Performance RP-18e 100 mm × 4.6 mm monolithic column (Merck KGaA, Darmstadt, Germany) and Luna C18(2) 150 mm × 3 mm column, 5 m particle size (Phenomenex, Inc., Torrance, CA) were assessed during chromatographic method development. 2.2. Sample source Urine samples from female cynomolgus monkeys collected following an intravenous (IV) administration of everninomicin and/or gentamicin for up to 1 week [24] were used for method development and subsequent analysis. Animals (n = 4 per dose group) were dosed at SPRI, Lafayette, NJ. Control animals were treated with 0.9% saline solution. For everninomicin treatment, monkeys were administered a single dose of 30 and 60 mg/kg. For gentamicin treatment, monkeys were administered a single dose of 10 mg/kg. For treatment with a combination of everninomicin and gentamicin, monkeys were administered a single dose of 10 mg gentamicin/kg and 30 mg everninomicin/kg. Urine samples were collected in 0–24 h blocks pre-dose and on Days 1, 3, and 7 and stored frozen at −30 ◦ C or below. 2.3. Calibration standards and quality control samples Stock solutions of HA (10 mg/ml), 13 C-HA (10 mg/ml), BA (2 mg/ml), 13 C-BA (2 mg/ml), and 5-Br-HA (2 mg/ml) were prepared in DMSO. A 2 mg/ml stock solution of l-phenylalanine-ring-D5 which served as an internal standard (ISTD) in the final method was prepared in 20 mM ammonium carbonate. Standard solutions containing a mixture of HA, BA and 5-Br-HA (40 g/ml each), HA, BA and l-phenylalanine (40 g/ml each), or HA, 13 C-HA, BA, 13 CBA and l-phenylalanine-ring-D5 (40 g/ml each) were prepared in water and/or pooled pre-dose monkey urine. All urine samples were centrifuged at ∼10,000 × g for 15 min before preparing solutions. Calibration curve range for 13 C-BA was 0.1–100 g/ml and for 13 C-HA it was 0.25–250 g/ml. Calibration standards VIII through I were prepared at 0.1, 0.25, 0.5, 2, 5, 20, 50, and 100 g/ml for 13 C-BA and at 0.25, 0.625, 1.25, 5, 12.5, 50, 125, and 250 g/ml for 13 C-HA. High, medium, and low QC samples were prepared for 13 C-BA at 80, 4, and 0.3 g/ml, respectively and for 13 C-HA at 200, 10, and 0.75 g/ml, respectively. Dilution QC samples were prepared at 80 (5×) and 40 (10×) g/ml for 13 C-BA and at 200 (5×) and 100 (10×) g/ml for 13 C-HA. On each day of analysis, an aliquot (2–3 ml) of pre-dose monkey urine was spiked with an appropriate volume of ISTD stock solution to the final concentration of 80 g/ml. This spiked urine sample was used for preparation of all calibration standards and QC samples. Calibration standard I and QC High were prepared by spiking calculated volume of 13 C-HA and 13 C-BA stock solutions into urine containing ISTD. Standards II–VIII were prepared from Standard I by serial dilution with ISTD-spiked urine. The same approach was
536
N. Penner et al. / Journal of Pharmaceutical and Biomedical Analysis 52 (2010) 534–543
used for preparation of QC Medium and QC Low samples from QC High. Dilution QCs were prepared in two steps. First, appropriate volume of 13 C-HA and 13 C-BA stock solutions was spiked into urine containing ISTD to achieve a concentration 5 or 10 times higher than the above specified nominal concentrations. These concentrated samples were then diluted with the ISTD-spiked urine to the nominal concentration of 40 and 80 g/ml for 13 C-BA, and 100 and 200 g/ml for 13 C-HA. All calibration standards and QCs were freshly prepared on the day of each run. 2.4. Preparation of monkey urine samples for analysis Approximately 500 l of each urine sample were transferred into a separate microcentrifuge tube and centrifuged at ∼10,000 × g for 15 min. An aliquot of supernatant (240 l) was then spiked with ISTD (10 l), vortexed, and transferred into injection vial for analysis.
3. Results and discussion 3.1. Development of the LC–MS/MS method Previously, different conditions for quantitative determination of HA in urine samples were evaluated in our laboratory [22]. Luna octadecylsilica or Chromolith column were used for separation. Detection was performed with either ESI or APCI ion sources, HA/13 C-HA pair was monitored in the positive and 5-Br-HA (ISTD) in the negative mode. These conditions did not provide the necessary separation and sensitivity of detection required for routine quantitation of the HA/BA mixture. Additional method development was necessary and involved evaluation of other chromatographic columns and mobile phases (buffer type and its pH, organic solvent). Since all analytes were acids, the negative ionization mode theoretically could provide superior sensitivity in comparison to the positive
Fig. 1. Matrix effect on the peak shape of HA in the negative (A) and positive (B) ionization mode using different chromatographic columns.
N. Penner et al. / Journal of Pharmaceutical and Biomedical Analysis 52 (2010) 534–543
Fig. 2. MS/MS spectra of BA, HA, and l-phenylalanine-ring-D5 in the negative ionization mode.
537
538
N. Penner et al. / Journal of Pharmaceutical and Biomedical Analysis 52 (2010) 534–543
Table 1 Structures of analytes, surrogate standards, and ISTD with corresponding SRM transitions selected for quantitation. Compound name
Structure
Transition (m/z)
Benzoic acid (BA)
121 → 77
13
127 → 83
C-benzoic acid (13 C-BA)
Hippuric acid (HA)
178 → 77
13
184 → 83
C-hippuric acid (13 C-HA)
l-Phenylalanine-ring-D5 (ISTD)
mode. Unfortunately, interference from endogenous compounds in mass spectrometer was much higher in the negative mode (Fig. 1). For example, under the same LC–MS/MS conditions the peak of HA after injection of standard solution in water was symmetrical (Fig. 1A) while in urine HA showed significant peak tailing. This effect was purely related to the matrix ion interferences in the mass spectrometer and was not a chromatography issue. In the positive mode (Fig. 1B) minor tailing was observed with Luna C18(2) but not with the monolithic Chromolith column. Additional method development showed that BA signal was poor in both positive and negative mode with ammonium acetatecontaining mobile phase. Ammonium carbonate (pH 7.5) provided satisfactory chromatographic separation of all three components of the mixture, however, under these conditions 5-Br-HA could not be used as ISTD due to poor MS response reproducibility. Based on its structural similarity to analytes, the deuterated ring-D5 analog of l-phenylalanine was selected as potential ISTD. Tandem mass spectra (MS/MS) of HA, BA, and l-phenylalanine-ring-D5 in the negative mode are presented in Fig. 2. Structures for all analytes, surrogate standards, and ISTD with corresponding SRM transitions selected for quantitation are presented in Table 1. The following chromatographic conditions were selected for method validation and sample analyses: the mobile phase consisted of (A) 20 mM ammonium carbonate (pH adjusted to 7.5 with glacial acetic acid) and (B) acetonitrile/methanol (50:50, v/v). The Chromolith Performance RP-18e monolithic column was kept at room temperature. Separation was achieved using linear gradient elution program (A/B ratios of 100/0, 100/0, 80/20, 50/50, 10/90, 10/90, 100/0, and 100/0 at 0, 2, 6, 7, 7.5, 8.5, 8.6, and 10 min, respectively) at a flow rate of 1 ml/min with 100% of effluent directed to the mass spectrometer. The final MS method was divided into two segments to achieve optimal sensitivity for detection of 13 C-HA/HA and 13 C-BA/BA pairs. In segment 1 the mass spectrometer was tuned for detection of BA while in segment 2 a tune file for HA was used. The APCI ion source was used in the negative ionization mode. Collision energy was set at 25 eV, capillary temperature was 300 ◦ C, and vaporizer temperature was 500 ◦ C. Sheath and auxiliary gases were set at 49 and 25 (arbitrary units), respectively. Discharge current, tube lens and scan
169 → 108
time for the detection of BA in segment 1 were set to 4 A, 100 V, and 0.5 s, respectively. The same parameters for the detection of HA in segment 2 were set to 10 A, 78 V, and 0.3 s, respectively. A representative extracted ion chromatogram of surrogate standards with ISTD spiked into blank monkey urine is shown in Fig. 3. l-Phenylalanine-ring-D5 provided acceptable signal in segment 2. 3.2. Dynamic range It is well documented in the literature that endogenous levels of BA and especially HA can vary significantly among species and depend on many external factors including diet [10–14]. Therefore, prior to method qualification and sample analysis, preliminary information on the expected concentration range of analytes in a given set of samples was obtained. In our case, 1 H NMR experiments conducted earlier on selected monkey urine samples (data not shown) indicated that compared to pre-dose levels, a decrease in urinary HA concentration following the dose of everninomicin alone and in combination with gentamicin was observed. Pooled pre-dose monkey urine was injected to obtain an estimate for background level of HA and BA in samples to be analyzed (Fig. 4). Based on these data the upper level of quantitation (Standard I) for HA and BA was set at 250 and 100 g/ml, respectively. To accommodate possible higher concentrations, two dilution QCs were introduced (see Section 2) which extended the concentration range of the two analytes to 1000 and 400 g/ml, respectively. The lower level of quantitation for HA and BA was set at 0.25 and 0.1 g/ml, respectively. Representative SRM chromatograms of 13 C-BA and 13 C-HA at low (standard VIII) and high (Standard I) level of quantitation are shown in Fig. 5. 3.3. Selectivity Since both HA and BA are endogenous compounds there are no blank urine samples devoid of them. Therefore, selectivity of the method developed in this study was tested based on surrogate standards 13 C-HA and 13 C-BA. There was no response from either of the two 13 C-compounds observed in pooled blank monkey urine at the appropriate retention times (Fig. 4). Furthermore, the lack of
N. Penner et al. / Journal of Pharmaceutical and Biomedical Analysis 52 (2010) 534–543
539
Fig. 3. Representative extracted ion chromatograms for 13 C-HA, 13 C-BA, and internal standard (ISTD) following injection of Standard I.
the signal from surrogate standards in the presence of significant amounts of HA and BA confirmed that SRM transitions were chosen correctly and interference was minimal. 3.4. Precision and accuracy Huge HA concentration variability in animals and humans is well documented [10–14]. Therefore, in this study it was important to determine a trend or a concentration change over the course of a study for each animal and relate it to toxicological findings [24].
For assay qualification, we adopted rules for discovery-supporting non-GLP assays (Level III with QC samples) as described in the literature [23]. A precision of 25% and accuracy of 75–125% for QC samples were set over the whole concentration range of HA and BA. The number of QC sample levels was limited to three: at low, medium, and high end of the standard curve. QC sample at low limit of quantitation (LLOQ) was not used. At least 2/3 of the QC samples had to be within 25% of their nominal value. At least 75% of assayed calibration standards had to be used for calibration curves and regression analysis. Standard samples included into the
Fig. 4. Representative extracted ion chromatograms for 13 C-HA, HA, 13 C-BA, and BA in pooled pre-dose monkey urine.
540
N. Penner et al. / Journal of Pharmaceutical and Biomedical Analysis 52 (2010) 534–543
Fig. 5. Representative extracted ion chromatograms for 13 C-HA and 13 C-BA at high (Standard I) and low (Standard VIII) limits of quantitation.
regression analysis had to be within 27.5% of their nominal value. Representative extracted ion chromatograms for 13 C-HA and 13 CBA at high (Standard I) and low (Standard VIII) limits of quantitation are shown in Fig. 5. Since the range of analyte concentrations was wide (3 orders of magnitude), the potential for carry-over was evaluated by injecting a blank monkey urine sample immediately after a high QC sample. No peak corresponding to 13 C-BA was detected in blank urine (data not shown). The response from 13 C-HA was below LLOQ which indicated that carry-over was very minor (<0.1%). The inter- and intra-assay performance data for 13 C-BA and 13 CHA are presented in Tables 2 and 3. Intra-day precision was within 13% for 13 C-BA and within 16% for 13 C-HA. The inter-day assay precision did not exceed 22% for both analytes. Intra-day accuracy was within 10%. Similar results were obtained for dilution QCs with accuracy <10% for both analytes.
ical run. Weighted (1/x) quadratic regression (r2 ≥ 0.99) provided the best fit for calibration curves over the concentration ranges of 0.1–100 and 0.25–250 g/ml for 13 C-BA and 13 C-HA, respectively (Fig. 6).
3.5. Stability Stability of HA in urine has been well studied [8]; no degradation of HA occurred under working conditions typically encountered during sample handling (including storage at 4 and −20 ◦ C and 3 freeze–thaw cycles). Therefore, only the stability of surrogate standards was assessed in this work using QC samples of low, medium, and high level. It was found that both surrogate standards spiked into monkey urine were stable at room temperature for up to 48 h. In addition, no degradation was observed following 3 freeze–thaw cycles. Long-term stability of surrogate standards was not investigated since solutions can be freshly prepared as needed. 3.6. Calibration curves Eight calibration standards prepared in pooled pre-dose monkey urine were injected at the beginning and at the end of each analyt-
Fig. 6. Typical calibration curves for 13 C-HA and 13 C-BA in monkey urine. The calibration samples were analyzed in duplicate.
N. Penner et al. / Journal of Pharmaceutical and Biomedical Analysis 52 (2010) 534–543
541
Table 2 Accuracy and precision for 13 C-BA in monkey urine. Theoretical amount (g/ml)
Day
Calculated amount (g/ml)
%CV
%Diff
80.0 (QC High)
Day 1 Day 2 Day 3 Day 4 Day 5 Intra-day
61.5 75.8 83.3 75.8 81.4 75.6
5.1 8.5 21 17 6.3 11
−23 −5.2 4.2 −5.0 1.6 −5.5
4.00 (QC Medium)
Day 1 Day 2 Day 3 Day 4 Day 5 Intra-day
4.40 4.81 4.11 3.26 4.23 4.16
7.1 1.3 17 8.0 5.1 13
10 20 2.3 −19 5.8 4.1
0.300 (QC Low)
Day 1 Day 2 Day 3 Day 4 Day 5 Intra-day
0.267 0.370 0.284 0.292 0.301 0.303
16 0.1 19 20 8.5 13
−11 24 −4.5 −3.0 0.30 1.1
3.7. Analysis of study samples The method developed in this study was used to analyze urine samples from female monkeys dosed with everninomicin alone or in combination with gentamicin. Concentration of endogenous HA and BA were calculated using the peak area ratio of the analyte to the internal standard and the regression equation of the calibration curve for the corresponding surrogate standard. Since 13 C-labeled compounds were used as surrogate standards, no chromatographic separation of the standard and the analyte occurred, and similar ion suppression due to matrix interference would be expected for both of them. It was previously shown that the chromatographic and mass spectrometric behavior of HA and 13 C-HA was identical and the calibration curve obtained from 13 C-HA could be used for quantitation of HA [22]. The same was true for the 13 C-BA/BA pair which allowed us to assume that the response factor of 13 C-BA/BA
and 13 C-HA/HA in urine should be equal or close to 1. On each analysis day the peak area ratio of the analyte to ISTD was determined (Xcalibur 1.4, Quantitative analysis), and concentrations of endogenous BA and/or HA were calculated using the following regression equation: y = ax2 + bx + c, where x is the concentration (g/ml), y is the peak area ratio of surrogate standard (13 C-BA or 13 C-HA) or the endogenous analyte (BA or HA) to ISTD, and a, b, and c are the regression coefficients. Back-calculated concentration (g/ml) versus time profiles of HA and BA in monkey urine are presented in Figs. 7 and 8, respectively. Concentration of urinary HA decreased significantly (10 times or more by Day 7) when everninomicin was administered to monkeys. Following the administration of gentamicin alone, the
Fig. 7. Calculated concentrations of HA in urine (g/ml) following IV administration of everninomicin and/or gentamicin to female monkeys.
542
N. Penner et al. / Journal of Pharmaceutical and Biomedical Analysis 52 (2010) 534–543
Table 3 Accuracy and precision for 13 C-HA in monkey urine. Theoretical amount (g/ml)
Day
Calculated amount (g/ml)
%CV
%Diff
200 (QC High)
Day 1 Day 2 Day 3 Day 4 Day 5 Intra-day
168 222 212 153 180 187
6.0 13 15 6.1 4.5 16
−16 11 6.0 −24 −10 −6.5
10.0 (QC Medium)
Day 1 Day 2 Day 3 Day 4 Day 5 Intra-day
7.0 12 11 4.7 7.6 16
−4.0 12 13 −27 2.0 −0.9
0.750 (QC Low)
Day 1 Day 2 Day 3 Day 4 Day 5 Intra-day
7.8 16 22 4.2 9.7 9.9
−8.0 −2.0 −7.0 −25 −6.9 −9.8
9.56 11.2 11.3 7.31 10.2 9.91 0.693 0.731 0.701 0.559 0.698 0.676
Fig. 8. Calculated concentrations of BA in urine (g/ml) following IV administration of everninomicin and/or gentamicin to female monkeys.
N. Penner et al. / Journal of Pharmaceutical and Biomedical Analysis 52 (2010) 534–543
decrease in HA level was minor. For most pre-dose samples, concentration of HA was very high (>250 g/ml). Since a very clear trend was observed, an additional analysis of the diluted predose samples was not necessary. These data are in agreement with results of 1 H NMR analysis of the same sample set (data not shown). One possible explanation for such dramatic changes in HA levels could be that the antibiotic everninomicin suppressed gut microflora, which is partially responsible for HA synthesis. However, the changes in benzoic acid concentration were less prominent. Another factor potentially affecting HA levels in urine is nephrotoxicity observed in the studied dose groups [24]. The experimental oligosaccharide antibiotic everninomicin (development discontinued) and aminoglycoside antibiotic gentamicin are known nephrotoxicants causing extensive renal damage [24]. Histopathological findings and gene expression analyses for this study were also described [24]. Monkeys receiving 60 mg daily dose of everninomicin developed renal lesions by Day 7, while an administration of a combination of both antibiotics dramatically increased the severity of toxicity with renal lesions occurring as early as Day 1. The same levels of everninomicin and gentamicin when dosed separately did not result in nephrotoxicity. In this study the severity of observed toxicity may be correlated with how fast the decrease in urinary HA occurs, however, sample collection during narrow intervals (e.g. 4-h blocks instead of 24-h blocks as in this study) would be required for a better understanding of renal toxicity-dependant HA behavior to conclude if this compound could be used as a biomarker of renal toxicity. 4. Conclusions A “surrogate standard” approach to quantitative biomarker analysis was tested by measuring hippuric and benzoic acids in monkey urine. An LC–MS method for quantitative determination of HA and BA was developed and successfully applied for analysis of urine obtained from female monkeys following dosing with everninomicin and/or gentamicin. Acknowledgement The authors thank Sumei Ren for synthesis of 13 C6 -hippuric acid used in this study. References [1] H.C. Keun, Metabonomic modeling of drug toxicity, Pharmacol. Ther. 109 (2006) 92–106. [2] J.C. Lindon, E. Holmes, J.K. Nicholson, Metabonomics techniques and applications to pharmaceutical research & development, Pharm. Res. 23 (2006) 1075–1088. [3] B.L. Ackermann, J.E. Hale, K.L. Duffin, The role of mass spectrometry in biomarker discovery and measurement, Curr. Drug Metab. 7 (2006) 525–539.
543
[4] D.G. Robertson, Current challenges and future developments in metabonomics technology, in: D.G. Robertson, J. Lindon (Eds.), Metabonomics in Toxicity Assessment, CRC Press, Boca Raton, FL, 2005, pp. 499–516. [5] S. Kochhar, D.M. Jacobs, Z. Ramadan, F. Berruex, A. Fuerholz, L.B. Fay, Probing gender-specific metabolism differences in humans by nuclear magnetic resonance-based metabonomics, Anal. Biochem. 352 (2006) 274–281. [6] M. Kussmann, F. Raymond, M. Affolter, OMICS-driven biomarker discovery in nutrition and health, J. Biotechnol. 124 (2006) 758–787. [7] F. Dieterle, G. Schlotterbeck, A. Ross, U. Niederhauser, H. Senn, Application of metabonomics in a compound ranking study in early drug development revealing drug-induced excretion of choline into urine, Chem. Res. Toxicol. 19 (2006) 1175–1181. [8] M.Y. Huang, C. Jin, Y.T. Liu, B.H. Li, Q.S. Qu, Y. Uchida, O. Inoue, H. Nakatsuka, T. Watanabe, M. Ikeda, Exposure of workers to a mixture of toluene and xylenes. I. Metabolism, Occup. Environ. Med. 51 (1994) 42–46. [9] E. Ramachandran, S. Natarajan, Growth habits of hippuric acid in gel, Cryst. Res. Technol. 37 (2002) 1274–1279. [10] A. Lof, H.E. Wigaeus, A. Colmsjo, B.O. Lundmark, A. Norstrom, N. Sato, Toxicokinetics of toluene and urinary excretion of hippuric acid after human exposure to 2H8-toluene, Br. J. Ind. Med. 50 (1993) 55–59. [11] T. Saito, S. Takeichi, Simultaneous detection of hippuric acid and methylhippuric acid in urine by Empore disk and gas chromatography–mass spectrometry, J. Pharm. Biomed. Anal. 30 (2002) 365–370. [12] M.E. Siquera, M.J.N. Paiva, Hippuric acid in urine: reference values, Rev. Saúde Pública 36 (2002) 723–727, Available from: http://www.scielosp.org/scielo.php?script=sci arttext&pid=S003489102002000700010&lng=en&nrm=iso. [13] A.N. Phipps, J. Stewart, B. Wright, I.D. Wilson, Effect of diet on the urinary excretion of hippuric acid and other dietary-derived aromatics in rat. A complex interaction between the diet, gut microflora and substrate specificity, Xenobiotica 28 (1998) 527–537. [14] M.N. Clifford, E.L. Copeland, J.P. Bloxsidge, L.A. Mitchell, Hippuric acid as a major excretion product associated with black tea consumption, Xenobiotica 30 (2000) 317–326. [15] C. Zuppi, I. Messana, P. Tapanainen, M. Knip, F. Vincenzoni, B. Giardina, M. Nuutinen, Proton nuclear magnetic resonance spectral profiles of urine from children and adolescences with type 1 diabetes, Clin. Chem. 48 (2002) 660–662. [16] M. LeBel, L. Ferron, M. Masson, J. Pichette, C. Carrier, Benzyl alcohol metabolism and elimination in neonates, Dev. Pharmacol. Ther. 11 (1988) 347–356. [17] M.J. Arin, M.T. Diez, J.A. Resines, M.T. Alemany, Quantitative estimation of urinary metabolites in ruminants by high-performance liquid chromatography, J. Liq. Chromatogr. 13 (1990) 2465–2473. [18] K.B. Elgstoen, J.Y. Zhao, J.F. Anacleto, E. Jellum, Potential of capillary electrophoresis, tandem mass spectrometry and coupled capillary electrophoresistandem mass spectrometry as diagnostic tools, J. Chromatogr. A 914 (2001) 265–275. [19] U. Langenbeck, J.E. Seegmiller, Gas chromatograph–mass spectrometry studies of hippuric acid derivatives. I. Evaluation of methods for quantitative determination hippuric acid, J. Chromatogr. 80 (1973) 81–88. [20] W. Li, L.H. Cohen, Quantitation of endogenous analytes in biofluid without a true blank matrix, Anal. Chem. 75 (2003) 5854–5859. [21] H.R. Liang, T. Takagaki, R.L. Foltz, P. Bennet, Quantitative determination of endogenous sorbitol and fructose in human erythrocytes by atmospheric pressure chemical ionization LC tandem mass spectrometry, J. Chromatogr. B: Anal. Technol. Biomed Life Sci. 824 (2005) 36–44. [22] N.A. Penner, R. Ramanathan, S. Chowdhury, K. Alton, Development of a liquid chromatography–mass spectrometry method for quantitative determination of hippuric acid in urine, in: Proceedings of the 52nd American Society for Mass Spectrometry Meeting, Nashville, TN, 2004. [23] W.A. Korfmacher, Bioanalytical assays in drug discovery environment, in: W.A. Korfmacher (Ed.), Using Mass Spectrometry for Drug Metabolism Studies, CRC Press LLC, Boca Raton, FL, 2005, pp. 1–34. [24] J.W. Davis, F.M. Goodsaid, C.M. Bral, L.A. Obert, G. Mandakas, C.E. Garner, N.D. Collins, R.J. Smith, I.Y. Rosenblum, Quantitative gene expression analysis in a nonhuman primate model of antibiotic-induced nephrotoxicity, Toxicol. Appl. Pharmacol. 200 (2004) 16–26.
Contents lists available at ScienceDirect
Journal of Pharmaceutical and Biomedical Analysis journal homepage: www.elsevier.com/locate/jpba
Quantitative determination of hippuric and benzoic acids in urine by LC–MS/MS using surrogate standards Natalia Penner 1 , Ragu Ramanathan 2 , Joanna Zgoda-Pols ∗ , Swapan Chowdhury Drug Metabolism and Pharmacokinetics, Merck Research Laboratories, 2015 Galloping Hill Road, Kenilworth, NJ 07033, United States
a r t i c l e
i n f o
Article history: Received 17 November 2009 Received in revised form 7 January 2010 Accepted 9 January 2010 Available online 18 January 2010 Keywords: LC–MS/MS Hippuric acid Benzoic acid Biomarkers
a b s t r a c t An LC–MS/MS method for simultaneous determination of hippuric acid (HA) and benzoic acid (BA) in monkey urine after direct injection was developed. Since HA and BA are endogenous compounds in urine, surrogate standards (13 C6 -hippuric and 13 C6 -benzoic acid) were employed to generate calibration curves. l-Phenylalanine-ring-D5 served as an internal standard. Multiple reaction monitoring in the negative ionization mode with an APCI source was used for detection of all components in the assay. The developed method is intended for determination of HA and BA in the range of 0.25–250 and 0.1–100 g/ml, respectively. Weighted (1/x) quadratic regression (r2 > 0.99) was used to generate calibration curves. Precision and accuracy of the method were assessed by analyzing 3 quality control samples (concentrations at low, medium, and high range of calibration curve) prepared in monkey urine. Stability for 48 h at room temperature and after 3 freeze–thaw cycles was also evaluated. The proposed method was successfully utilized for analysis of urine samples from female monkeys following the administration of everninomicin alone and in combination with gentamicin. The concentrations of endogenous HA and BA were calculated based on the peak area ratio of the analyte to the internal standard using a regression equation for corresponding surrogate standard. © 2010 Elsevier B.V. All rights reserved.
1. Introduction Exploring etiology of human diseases is one of the major goals of pharmaceutical research and medical science. A better understanding of how a disease originates and progresses could help in prevention or creating new approaches for detection and cure. Any living organism can be viewed as an extremely complex system, whose quantitative and qualitative composition reflects disease state, toxicological influences, environmental stress, age, and general health. Systems biology and physiological changes occurring in a living organism are usually studied from the point of view of multiple “omics”. Genomics targets genes and their mutations, transcriptomics deals with changes in mRNA level, proteomics studies protein expression, post-translational modifications, and up- and down-regulation in both. Metabonomics (metabolomics) targets the profile of endogenous compounds produced by a living organism and how this profile changes depending on a state of a biological system. Among the different “omics”, only metabonomics is able to generate information which reflects the function of the whole system in real time. This is achieved by
∗ Corresponding author. Tel.: +1 908 740 2170; fax: +1 908 740 3966. E-mail address: [email protected] (J. Zgoda-Pols). 1 Present address: Biogen Idec, Cambridge, MA, United States. 2 Present address: Bristol-Myers Squibb Company, Princeton, NJ, United States. 0731-7085/$ – see front matter © 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.jpba.2010.01.016
analyzing minimally invasive biofluids, such as urine and plasma, which are readily available in clinical and non-clinical setting. Analysis is usually carried out by high-resolution NMR spectroscopy or LC–MS with subsequent data reduction using statistical tools such as principal component analysis (PCA) [1–7]. Comparison of treatment groups may result in specific patterns or changes in metabolic profiles which can be then used for drug toxicity prediction and safety assessment [1–4,7], lead compound optimization [7], the study of gender-specific differences [5], inter-subject variability [6], and potentially, if a specific biomarker/group of biomarkers is identified, serve as an end-point in clinical trials. As with any complex biological matrix the biggest challenge is to recognize a comparatively small number of compounds that can inform us about possible toxicity, an adverse reaction, or pharmacodynamic effect. Identification of endogenous small molecule biomarkers that are related to natural, disease-related or drug-induced pharmacological events is the main objective of metabonomics. Standard workflow in most studies starts with creating a differential display of components changing in the system following external stimuli or as a response to disease progression, and then trying to identify them later. In the case of a disease marker, even when the concentration of an endogenous substance changes in response to a certain event or treatment, its use may not be straightforward. For example, hippuric acid (HA), the glycine conjugate of benzoic acid (BA), has been used for many years in occupational medicine for monitoring exposure
N. Penner et al. / Journal of Pharmaceutical and Biomedical Analysis 52 (2010) 534–543
to toluene [8]. Toluene is primarily metabolized to benzoic acid which then conjugates with glycine to form hippuric acid. HA is one of the main urinary endogenous metabolites and a principal route of benzoate elimination. HA participates in the natural regulation and prevention of stone formation [9]. Multiple studies conducted to evaluate the effect of various factors on excretion of hippuric acid showed that the HA concentration in urine is gender-, age-, race-, and generally subject-dependent [10]. Saito and Takeichi reported that urinary HA concentration in healthy human subjects (n = 6) was within a range of 5.5–15.8 g/ml [11]. In another larger study specifically designed to establish reference values of HA in a general population in one area of Brazil [12], HA levels of 0.18 ± 0.10 g/g creatinine corresponding to a mean of 280 g/ml (n = 115) was observed. These and other studies [13] confirmed, that the normal physiological level of HA in humans and animals is highly variable and depends on diet. Consumption of benzoate- or polyphenol-containing food and beverages (e.g. black tea) increased HA excretion [14]. It is also known that highly significant differences of hippurate levels may point to certain diseases e.g. diabetes in children and renal failure [15]. A similar situation occurs for BA. In some species, including humans, urinary excretion of minor amounts of benzoic acid is normal. However, impaired hippurate synthesis accompanied by elevated benzoate excretion in preterm babies can be an indication of fatal benzyl alcohol poisoning [16]; a similar picture in adults has even been associated with mental illness. When certain compounds are considered potential biomarkers, it is beneficial to monitor them over the course of a study or treatment duration to eliminate inherent inter-subject differences and to determine a trend. Based on literature information, the HA/BA pair can be a prospective biomarker combination. There have been a number of methods developed over the years utilizing colorimetric, LCUV [17], CE-MS [18], and GC–MS [19] techniques for quantitative determination of HA/BA in urine. Spectrophotometric methods are usually less selective and can suffer from interference from other urinary metabolites. The main disadvantage of GC–MS is the derivatization step which is necessary to convert HA/BA into more volatile esters or trimethylsilyl derivatives [19] prior to injection. The lack of an appropriate blank matrix for quantification of endogenous HA and BA has not been addressed before. A “surrogate standard” approach for quantitation of endogenous ␣ketoisocaproic acid in plasma was recently reported by Cohen and co-worker [20]. Deuterated ␣-ketoisocaproic acid was used to generate a standard curve and the concentration of naturally occurring ␣-ketoisocaproic acid was back-calculated using the regression equation derived for the deuterated surrogate standard. A similar approach was used for quantitation of endogenous sorbitol and fructose in human erythrocytes [21]. We also demonstrated earlier [22] that HA can be quantified based on calibration curve from 13 C -hippuric acid. The work described here was devoted to the 6 development of an LC–MS/MS method for simultaneous determination of HA and BA in monkey urine after direct injection. The method was qualified with precision and accuracy criteria set as described by Korfmacher [23] for discovery-supporting non-GLP assays requiring QC samples (Level III) at 25% and 75–125%, respectively, and used for analysis of urine samples from female monkeys following exposure to known nephrotoxicants everninomicin and gentamicin [24].
2. Experimental 2.1. Instruments and chemicals LC–MS experiments were performed using a TSQ Quantum mass spectrometer (Thermo Electron Corp., San Jose, CA) coupled with a
535
Waters Alliance 2695 HPLC module. TSQ Quantum mass spectrometer was operated with Q1 and/or Q3 resolution set at 0.1 Th FWHM (where Th = Thompson and FWHM = Full width at half maximum). Hippuric acid (HA), benzoic acid (BA), 5-bromo-3-hydroxyhippuric acid (5-Br-HA), acetic acid, and ammonium acetate were obtained from Sigma–Aldrich (Milwaukee, WI). 13 C6 -benzoic acid (13 C-BA) and l-phenylalanine-ring-D5 were purchased from Cambridge Isotope Laboratories, Inc. (Andover, MA). 13 C6 -hippuric acid (13 C-HA) was synthesized by Radiochemistry Group at Schering-Plough Research Institute. Ammonium carbonate was obtained from Acros Organic (Belgium). Acetonitrile and methanol were from Burdick and Jackson (Muskegon, MI). Deionized water was purified by MilliQ water purification system (Millipore, Bedford, MA). DMSO was from Fisher Scientific (Fair Lawn, NJ). A Chromolith Performance RP-18e 100 mm × 4.6 mm monolithic column (Merck KGaA, Darmstadt, Germany) and Luna C18(2) 150 mm × 3 mm column, 5 m particle size (Phenomenex, Inc., Torrance, CA) were assessed during chromatographic method development. 2.2. Sample source Urine samples from female cynomolgus monkeys collected following an intravenous (IV) administration of everninomicin and/or gentamicin for up to 1 week [24] were used for method development and subsequent analysis. Animals (n = 4 per dose group) were dosed at SPRI, Lafayette, NJ. Control animals were treated with 0.9% saline solution. For everninomicin treatment, monkeys were administered a single dose of 30 and 60 mg/kg. For gentamicin treatment, monkeys were administered a single dose of 10 mg/kg. For treatment with a combination of everninomicin and gentamicin, monkeys were administered a single dose of 10 mg gentamicin/kg and 30 mg everninomicin/kg. Urine samples were collected in 0–24 h blocks pre-dose and on Days 1, 3, and 7 and stored frozen at −30 ◦ C or below. 2.3. Calibration standards and quality control samples Stock solutions of HA (10 mg/ml), 13 C-HA (10 mg/ml), BA (2 mg/ml), 13 C-BA (2 mg/ml), and 5-Br-HA (2 mg/ml) were prepared in DMSO. A 2 mg/ml stock solution of l-phenylalanine-ring-D5 which served as an internal standard (ISTD) in the final method was prepared in 20 mM ammonium carbonate. Standard solutions containing a mixture of HA, BA and 5-Br-HA (40 g/ml each), HA, BA and l-phenylalanine (40 g/ml each), or HA, 13 C-HA, BA, 13 CBA and l-phenylalanine-ring-D5 (40 g/ml each) were prepared in water and/or pooled pre-dose monkey urine. All urine samples were centrifuged at ∼10,000 × g for 15 min before preparing solutions. Calibration curve range for 13 C-BA was 0.1–100 g/ml and for 13 C-HA it was 0.25–250 g/ml. Calibration standards VIII through I were prepared at 0.1, 0.25, 0.5, 2, 5, 20, 50, and 100 g/ml for 13 C-BA and at 0.25, 0.625, 1.25, 5, 12.5, 50, 125, and 250 g/ml for 13 C-HA. High, medium, and low QC samples were prepared for 13 C-BA at 80, 4, and 0.3 g/ml, respectively and for 13 C-HA at 200, 10, and 0.75 g/ml, respectively. Dilution QC samples were prepared at 80 (5×) and 40 (10×) g/ml for 13 C-BA and at 200 (5×) and 100 (10×) g/ml for 13 C-HA. On each day of analysis, an aliquot (2–3 ml) of pre-dose monkey urine was spiked with an appropriate volume of ISTD stock solution to the final concentration of 80 g/ml. This spiked urine sample was used for preparation of all calibration standards and QC samples. Calibration standard I and QC High were prepared by spiking calculated volume of 13 C-HA and 13 C-BA stock solutions into urine containing ISTD. Standards II–VIII were prepared from Standard I by serial dilution with ISTD-spiked urine. The same approach was
536
N. Penner et al. / Journal of Pharmaceutical and Biomedical Analysis 52 (2010) 534–543
used for preparation of QC Medium and QC Low samples from QC High. Dilution QCs were prepared in two steps. First, appropriate volume of 13 C-HA and 13 C-BA stock solutions was spiked into urine containing ISTD to achieve a concentration 5 or 10 times higher than the above specified nominal concentrations. These concentrated samples were then diluted with the ISTD-spiked urine to the nominal concentration of 40 and 80 g/ml for 13 C-BA, and 100 and 200 g/ml for 13 C-HA. All calibration standards and QCs were freshly prepared on the day of each run. 2.4. Preparation of monkey urine samples for analysis Approximately 500 l of each urine sample were transferred into a separate microcentrifuge tube and centrifuged at ∼10,000 × g for 15 min. An aliquot of supernatant (240 l) was then spiked with ISTD (10 l), vortexed, and transferred into injection vial for analysis.
3. Results and discussion 3.1. Development of the LC–MS/MS method Previously, different conditions for quantitative determination of HA in urine samples were evaluated in our laboratory [22]. Luna octadecylsilica or Chromolith column were used for separation. Detection was performed with either ESI or APCI ion sources, HA/13 C-HA pair was monitored in the positive and 5-Br-HA (ISTD) in the negative mode. These conditions did not provide the necessary separation and sensitivity of detection required for routine quantitation of the HA/BA mixture. Additional method development was necessary and involved evaluation of other chromatographic columns and mobile phases (buffer type and its pH, organic solvent). Since all analytes were acids, the negative ionization mode theoretically could provide superior sensitivity in comparison to the positive
Fig. 1. Matrix effect on the peak shape of HA in the negative (A) and positive (B) ionization mode using different chromatographic columns.
N. Penner et al. / Journal of Pharmaceutical and Biomedical Analysis 52 (2010) 534–543
Fig. 2. MS/MS spectra of BA, HA, and l-phenylalanine-ring-D5 in the negative ionization mode.
537
538
N. Penner et al. / Journal of Pharmaceutical and Biomedical Analysis 52 (2010) 534–543
Table 1 Structures of analytes, surrogate standards, and ISTD with corresponding SRM transitions selected for quantitation. Compound name
Structure
Transition (m/z)
Benzoic acid (BA)
121 → 77
13
127 → 83
C-benzoic acid (13 C-BA)
Hippuric acid (HA)
178 → 77
13
184 → 83
C-hippuric acid (13 C-HA)
l-Phenylalanine-ring-D5 (ISTD)
mode. Unfortunately, interference from endogenous compounds in mass spectrometer was much higher in the negative mode (Fig. 1). For example, under the same LC–MS/MS conditions the peak of HA after injection of standard solution in water was symmetrical (Fig. 1A) while in urine HA showed significant peak tailing. This effect was purely related to the matrix ion interferences in the mass spectrometer and was not a chromatography issue. In the positive mode (Fig. 1B) minor tailing was observed with Luna C18(2) but not with the monolithic Chromolith column. Additional method development showed that BA signal was poor in both positive and negative mode with ammonium acetatecontaining mobile phase. Ammonium carbonate (pH 7.5) provided satisfactory chromatographic separation of all three components of the mixture, however, under these conditions 5-Br-HA could not be used as ISTD due to poor MS response reproducibility. Based on its structural similarity to analytes, the deuterated ring-D5 analog of l-phenylalanine was selected as potential ISTD. Tandem mass spectra (MS/MS) of HA, BA, and l-phenylalanine-ring-D5 in the negative mode are presented in Fig. 2. Structures for all analytes, surrogate standards, and ISTD with corresponding SRM transitions selected for quantitation are presented in Table 1. The following chromatographic conditions were selected for method validation and sample analyses: the mobile phase consisted of (A) 20 mM ammonium carbonate (pH adjusted to 7.5 with glacial acetic acid) and (B) acetonitrile/methanol (50:50, v/v). The Chromolith Performance RP-18e monolithic column was kept at room temperature. Separation was achieved using linear gradient elution program (A/B ratios of 100/0, 100/0, 80/20, 50/50, 10/90, 10/90, 100/0, and 100/0 at 0, 2, 6, 7, 7.5, 8.5, 8.6, and 10 min, respectively) at a flow rate of 1 ml/min with 100% of effluent directed to the mass spectrometer. The final MS method was divided into two segments to achieve optimal sensitivity for detection of 13 C-HA/HA and 13 C-BA/BA pairs. In segment 1 the mass spectrometer was tuned for detection of BA while in segment 2 a tune file for HA was used. The APCI ion source was used in the negative ionization mode. Collision energy was set at 25 eV, capillary temperature was 300 ◦ C, and vaporizer temperature was 500 ◦ C. Sheath and auxiliary gases were set at 49 and 25 (arbitrary units), respectively. Discharge current, tube lens and scan
169 → 108
time for the detection of BA in segment 1 were set to 4 A, 100 V, and 0.5 s, respectively. The same parameters for the detection of HA in segment 2 were set to 10 A, 78 V, and 0.3 s, respectively. A representative extracted ion chromatogram of surrogate standards with ISTD spiked into blank monkey urine is shown in Fig. 3. l-Phenylalanine-ring-D5 provided acceptable signal in segment 2. 3.2. Dynamic range It is well documented in the literature that endogenous levels of BA and especially HA can vary significantly among species and depend on many external factors including diet [10–14]. Therefore, prior to method qualification and sample analysis, preliminary information on the expected concentration range of analytes in a given set of samples was obtained. In our case, 1 H NMR experiments conducted earlier on selected monkey urine samples (data not shown) indicated that compared to pre-dose levels, a decrease in urinary HA concentration following the dose of everninomicin alone and in combination with gentamicin was observed. Pooled pre-dose monkey urine was injected to obtain an estimate for background level of HA and BA in samples to be analyzed (Fig. 4). Based on these data the upper level of quantitation (Standard I) for HA and BA was set at 250 and 100 g/ml, respectively. To accommodate possible higher concentrations, two dilution QCs were introduced (see Section 2) which extended the concentration range of the two analytes to 1000 and 400 g/ml, respectively. The lower level of quantitation for HA and BA was set at 0.25 and 0.1 g/ml, respectively. Representative SRM chromatograms of 13 C-BA and 13 C-HA at low (standard VIII) and high (Standard I) level of quantitation are shown in Fig. 5. 3.3. Selectivity Since both HA and BA are endogenous compounds there are no blank urine samples devoid of them. Therefore, selectivity of the method developed in this study was tested based on surrogate standards 13 C-HA and 13 C-BA. There was no response from either of the two 13 C-compounds observed in pooled blank monkey urine at the appropriate retention times (Fig. 4). Furthermore, the lack of
N. Penner et al. / Journal of Pharmaceutical and Biomedical Analysis 52 (2010) 534–543
539
Fig. 3. Representative extracted ion chromatograms for 13 C-HA, 13 C-BA, and internal standard (ISTD) following injection of Standard I.
the signal from surrogate standards in the presence of significant amounts of HA and BA confirmed that SRM transitions were chosen correctly and interference was minimal. 3.4. Precision and accuracy Huge HA concentration variability in animals and humans is well documented [10–14]. Therefore, in this study it was important to determine a trend or a concentration change over the course of a study for each animal and relate it to toxicological findings [24].
For assay qualification, we adopted rules for discovery-supporting non-GLP assays (Level III with QC samples) as described in the literature [23]. A precision of 25% and accuracy of 75–125% for QC samples were set over the whole concentration range of HA and BA. The number of QC sample levels was limited to three: at low, medium, and high end of the standard curve. QC sample at low limit of quantitation (LLOQ) was not used. At least 2/3 of the QC samples had to be within 25% of their nominal value. At least 75% of assayed calibration standards had to be used for calibration curves and regression analysis. Standard samples included into the
Fig. 4. Representative extracted ion chromatograms for 13 C-HA, HA, 13 C-BA, and BA in pooled pre-dose monkey urine.
540
N. Penner et al. / Journal of Pharmaceutical and Biomedical Analysis 52 (2010) 534–543
Fig. 5. Representative extracted ion chromatograms for 13 C-HA and 13 C-BA at high (Standard I) and low (Standard VIII) limits of quantitation.
regression analysis had to be within 27.5% of their nominal value. Representative extracted ion chromatograms for 13 C-HA and 13 CBA at high (Standard I) and low (Standard VIII) limits of quantitation are shown in Fig. 5. Since the range of analyte concentrations was wide (3 orders of magnitude), the potential for carry-over was evaluated by injecting a blank monkey urine sample immediately after a high QC sample. No peak corresponding to 13 C-BA was detected in blank urine (data not shown). The response from 13 C-HA was below LLOQ which indicated that carry-over was very minor (<0.1%). The inter- and intra-assay performance data for 13 C-BA and 13 CHA are presented in Tables 2 and 3. Intra-day precision was within 13% for 13 C-BA and within 16% for 13 C-HA. The inter-day assay precision did not exceed 22% for both analytes. Intra-day accuracy was within 10%. Similar results were obtained for dilution QCs with accuracy <10% for both analytes.
ical run. Weighted (1/x) quadratic regression (r2 ≥ 0.99) provided the best fit for calibration curves over the concentration ranges of 0.1–100 and 0.25–250 g/ml for 13 C-BA and 13 C-HA, respectively (Fig. 6).
3.5. Stability Stability of HA in urine has been well studied [8]; no degradation of HA occurred under working conditions typically encountered during sample handling (including storage at 4 and −20 ◦ C and 3 freeze–thaw cycles). Therefore, only the stability of surrogate standards was assessed in this work using QC samples of low, medium, and high level. It was found that both surrogate standards spiked into monkey urine were stable at room temperature for up to 48 h. In addition, no degradation was observed following 3 freeze–thaw cycles. Long-term stability of surrogate standards was not investigated since solutions can be freshly prepared as needed. 3.6. Calibration curves Eight calibration standards prepared in pooled pre-dose monkey urine were injected at the beginning and at the end of each analyt-
Fig. 6. Typical calibration curves for 13 C-HA and 13 C-BA in monkey urine. The calibration samples were analyzed in duplicate.
N. Penner et al. / Journal of Pharmaceutical and Biomedical Analysis 52 (2010) 534–543
541
Table 2 Accuracy and precision for 13 C-BA in monkey urine. Theoretical amount (g/ml)
Day
Calculated amount (g/ml)
%CV
%Diff
80.0 (QC High)
Day 1 Day 2 Day 3 Day 4 Day 5 Intra-day
61.5 75.8 83.3 75.8 81.4 75.6
5.1 8.5 21 17 6.3 11
−23 −5.2 4.2 −5.0 1.6 −5.5
4.00 (QC Medium)
Day 1 Day 2 Day 3 Day 4 Day 5 Intra-day
4.40 4.81 4.11 3.26 4.23 4.16
7.1 1.3 17 8.0 5.1 13
10 20 2.3 −19 5.8 4.1
0.300 (QC Low)
Day 1 Day 2 Day 3 Day 4 Day 5 Intra-day
0.267 0.370 0.284 0.292 0.301 0.303
16 0.1 19 20 8.5 13
−11 24 −4.5 −3.0 0.30 1.1
3.7. Analysis of study samples The method developed in this study was used to analyze urine samples from female monkeys dosed with everninomicin alone or in combination with gentamicin. Concentration of endogenous HA and BA were calculated using the peak area ratio of the analyte to the internal standard and the regression equation of the calibration curve for the corresponding surrogate standard. Since 13 C-labeled compounds were used as surrogate standards, no chromatographic separation of the standard and the analyte occurred, and similar ion suppression due to matrix interference would be expected for both of them. It was previously shown that the chromatographic and mass spectrometric behavior of HA and 13 C-HA was identical and the calibration curve obtained from 13 C-HA could be used for quantitation of HA [22]. The same was true for the 13 C-BA/BA pair which allowed us to assume that the response factor of 13 C-BA/BA
and 13 C-HA/HA in urine should be equal or close to 1. On each analysis day the peak area ratio of the analyte to ISTD was determined (Xcalibur 1.4, Quantitative analysis), and concentrations of endogenous BA and/or HA were calculated using the following regression equation: y = ax2 + bx + c, where x is the concentration (g/ml), y is the peak area ratio of surrogate standard (13 C-BA or 13 C-HA) or the endogenous analyte (BA or HA) to ISTD, and a, b, and c are the regression coefficients. Back-calculated concentration (g/ml) versus time profiles of HA and BA in monkey urine are presented in Figs. 7 and 8, respectively. Concentration of urinary HA decreased significantly (10 times or more by Day 7) when everninomicin was administered to monkeys. Following the administration of gentamicin alone, the
Fig. 7. Calculated concentrations of HA in urine (g/ml) following IV administration of everninomicin and/or gentamicin to female monkeys.
542
N. Penner et al. / Journal of Pharmaceutical and Biomedical Analysis 52 (2010) 534–543
Table 3 Accuracy and precision for 13 C-HA in monkey urine. Theoretical amount (g/ml)
Day
Calculated amount (g/ml)
%CV
%Diff
200 (QC High)
Day 1 Day 2 Day 3 Day 4 Day 5 Intra-day
168 222 212 153 180 187
6.0 13 15 6.1 4.5 16
−16 11 6.0 −24 −10 −6.5
10.0 (QC Medium)
Day 1 Day 2 Day 3 Day 4 Day 5 Intra-day
7.0 12 11 4.7 7.6 16
−4.0 12 13 −27 2.0 −0.9
0.750 (QC Low)
Day 1 Day 2 Day 3 Day 4 Day 5 Intra-day
7.8 16 22 4.2 9.7 9.9
−8.0 −2.0 −7.0 −25 −6.9 −9.8
9.56 11.2 11.3 7.31 10.2 9.91 0.693 0.731 0.701 0.559 0.698 0.676
Fig. 8. Calculated concentrations of BA in urine (g/ml) following IV administration of everninomicin and/or gentamicin to female monkeys.
N. Penner et al. / Journal of Pharmaceutical and Biomedical Analysis 52 (2010) 534–543
decrease in HA level was minor. For most pre-dose samples, concentration of HA was very high (>250 g/ml). Since a very clear trend was observed, an additional analysis of the diluted predose samples was not necessary. These data are in agreement with results of 1 H NMR analysis of the same sample set (data not shown). One possible explanation for such dramatic changes in HA levels could be that the antibiotic everninomicin suppressed gut microflora, which is partially responsible for HA synthesis. However, the changes in benzoic acid concentration were less prominent. Another factor potentially affecting HA levels in urine is nephrotoxicity observed in the studied dose groups [24]. The experimental oligosaccharide antibiotic everninomicin (development discontinued) and aminoglycoside antibiotic gentamicin are known nephrotoxicants causing extensive renal damage [24]. Histopathological findings and gene expression analyses for this study were also described [24]. Monkeys receiving 60 mg daily dose of everninomicin developed renal lesions by Day 7, while an administration of a combination of both antibiotics dramatically increased the severity of toxicity with renal lesions occurring as early as Day 1. The same levels of everninomicin and gentamicin when dosed separately did not result in nephrotoxicity. In this study the severity of observed toxicity may be correlated with how fast the decrease in urinary HA occurs, however, sample collection during narrow intervals (e.g. 4-h blocks instead of 24-h blocks as in this study) would be required for a better understanding of renal toxicity-dependant HA behavior to conclude if this compound could be used as a biomarker of renal toxicity. 4. Conclusions A “surrogate standard” approach to quantitative biomarker analysis was tested by measuring hippuric and benzoic acids in monkey urine. An LC–MS method for quantitative determination of HA and BA was developed and successfully applied for analysis of urine obtained from female monkeys following dosing with everninomicin and/or gentamicin. Acknowledgement The authors thank Sumei Ren for synthesis of 13 C6 -hippuric acid used in this study. References [1] H.C. Keun, Metabonomic modeling of drug toxicity, Pharmacol. Ther. 109 (2006) 92–106. [2] J.C. Lindon, E. Holmes, J.K. Nicholson, Metabonomics techniques and applications to pharmaceutical research & development, Pharm. Res. 23 (2006) 1075–1088. [3] B.L. Ackermann, J.E. Hale, K.L. Duffin, The role of mass spectrometry in biomarker discovery and measurement, Curr. Drug Metab. 7 (2006) 525–539.
543
[4] D.G. Robertson, Current challenges and future developments in metabonomics technology, in: D.G. Robertson, J. Lindon (Eds.), Metabonomics in Toxicity Assessment, CRC Press, Boca Raton, FL, 2005, pp. 499–516. [5] S. Kochhar, D.M. Jacobs, Z. Ramadan, F. Berruex, A. Fuerholz, L.B. Fay, Probing gender-specific metabolism differences in humans by nuclear magnetic resonance-based metabonomics, Anal. Biochem. 352 (2006) 274–281. [6] M. Kussmann, F. Raymond, M. Affolter, OMICS-driven biomarker discovery in nutrition and health, J. Biotechnol. 124 (2006) 758–787. [7] F. Dieterle, G. Schlotterbeck, A. Ross, U. Niederhauser, H. Senn, Application of metabonomics in a compound ranking study in early drug development revealing drug-induced excretion of choline into urine, Chem. Res. Toxicol. 19 (2006) 1175–1181. [8] M.Y. Huang, C. Jin, Y.T. Liu, B.H. Li, Q.S. Qu, Y. Uchida, O. Inoue, H. Nakatsuka, T. Watanabe, M. Ikeda, Exposure of workers to a mixture of toluene and xylenes. I. Metabolism, Occup. Environ. Med. 51 (1994) 42–46. [9] E. Ramachandran, S. Natarajan, Growth habits of hippuric acid in gel, Cryst. Res. Technol. 37 (2002) 1274–1279. [10] A. Lof, H.E. Wigaeus, A. Colmsjo, B.O. Lundmark, A. Norstrom, N. Sato, Toxicokinetics of toluene and urinary excretion of hippuric acid after human exposure to 2H8-toluene, Br. J. Ind. Med. 50 (1993) 55–59. [11] T. Saito, S. Takeichi, Simultaneous detection of hippuric acid and methylhippuric acid in urine by Empore disk and gas chromatography–mass spectrometry, J. Pharm. Biomed. Anal. 30 (2002) 365–370. [12] M.E. Siquera, M.J.N. Paiva, Hippuric acid in urine: reference values, Rev. Saúde Pública 36 (2002) 723–727, Available from: http://www.scielosp.org/scielo.php?script=sci arttext&pid=S003489102002000700010&lng=en&nrm=iso. [13] A.N. Phipps, J. Stewart, B. Wright, I.D. Wilson, Effect of diet on the urinary excretion of hippuric acid and other dietary-derived aromatics in rat. A complex interaction between the diet, gut microflora and substrate specificity, Xenobiotica 28 (1998) 527–537. [14] M.N. Clifford, E.L. Copeland, J.P. Bloxsidge, L.A. Mitchell, Hippuric acid as a major excretion product associated with black tea consumption, Xenobiotica 30 (2000) 317–326. [15] C. Zuppi, I. Messana, P. Tapanainen, M. Knip, F. Vincenzoni, B. Giardina, M. Nuutinen, Proton nuclear magnetic resonance spectral profiles of urine from children and adolescences with type 1 diabetes, Clin. Chem. 48 (2002) 660–662. [16] M. LeBel, L. Ferron, M. Masson, J. Pichette, C. Carrier, Benzyl alcohol metabolism and elimination in neonates, Dev. Pharmacol. Ther. 11 (1988) 347–356. [17] M.J. Arin, M.T. Diez, J.A. Resines, M.T. Alemany, Quantitative estimation of urinary metabolites in ruminants by high-performance liquid chromatography, J. Liq. Chromatogr. 13 (1990) 2465–2473. [18] K.B. Elgstoen, J.Y. Zhao, J.F. Anacleto, E. Jellum, Potential of capillary electrophoresis, tandem mass spectrometry and coupled capillary electrophoresistandem mass spectrometry as diagnostic tools, J. Chromatogr. A 914 (2001) 265–275. [19] U. Langenbeck, J.E. Seegmiller, Gas chromatograph–mass spectrometry studies of hippuric acid derivatives. I. Evaluation of methods for quantitative determination hippuric acid, J. Chromatogr. 80 (1973) 81–88. [20] W. Li, L.H. Cohen, Quantitation of endogenous analytes in biofluid without a true blank matrix, Anal. Chem. 75 (2003) 5854–5859. [21] H.R. Liang, T. Takagaki, R.L. Foltz, P. Bennet, Quantitative determination of endogenous sorbitol and fructose in human erythrocytes by atmospheric pressure chemical ionization LC tandem mass spectrometry, J. Chromatogr. B: Anal. Technol. Biomed Life Sci. 824 (2005) 36–44. [22] N.A. Penner, R. Ramanathan, S. Chowdhury, K. Alton, Development of a liquid chromatography–mass spectrometry method for quantitative determination of hippuric acid in urine, in: Proceedings of the 52nd American Society for Mass Spectrometry Meeting, Nashville, TN, 2004. [23] W.A. Korfmacher, Bioanalytical assays in drug discovery environment, in: W.A. Korfmacher (Ed.), Using Mass Spectrometry for Drug Metabolism Studies, CRC Press LLC, Boca Raton, FL, 2005, pp. 1–34. [24] J.W. Davis, F.M. Goodsaid, C.M. Bral, L.A. Obert, G. Mandakas, C.E. Garner, N.D. Collins, R.J. Smith, I.Y. Rosenblum, Quantitative gene expression analysis in a nonhuman primate model of antibiotic-induced nephrotoxicity, Toxicol. Appl. Pharmacol. 200 (2004) 16–26.